Universidad San Jorge

Escuela de Arquitectura y Tecnología

ANÁLISIS DE ESCENARIOS DE CLIMA FUTURO E IMPACTO

DEL CAMBIO CLIMÁTICO SOBRE LOS VIÑEDOS DEL
TERRITORIO ESPAÑOL IBÉRICO-BALEAR

TESIS DOCTORAL

Emma Gaitán Fernández

Villanueva de Gállego. 2022

 Universidad San Jorge

Escuela de Arquitectura y Tecnología

ANÁLISIS DE ESCENARIOS DE CLIMA FUTURO E IMPACTO

DEL CAMBIO CLIMÁTICO SOBRE LOS VIÑEDOS DEL
TERRITORIO ESPAÑOL IBÉRICO-BALEAR

TESIS DOCTORAL

Emma Gaitán Fernández

Villanueva de Gállego. 2022

Directora de Tesis

Mª Rosa Pino Otín

AGRADECIMIENTOS

Lo que obtienes al alcanzar tus metas no es tan importante como en lo

que te conviertes

(Henry David Thoreau)

Llegar al final de este viaje no solo supone alcanzar una meta que creía casi imposible,
sino que durante el proceso he coleccionado muchos momentos y enseñanzas que ya
son una parte de mí. Por ello no quería dejar de agradecer a todos los que han
contribuido a ello:

A Rosa, por ser el motor que ha movido este trabajo, con su apoyo y sus palabras de
aliento en los momentos en los que quise tirar la toalla y su optimismo en las épocas de
mayor entusiasmo.

A mis compañeros en la Fundación para la Investigación del Clima, no solo por lo que
aprendo de ellos cada día, sino también por los buenos momentos que pasamos juntos.
En especial a Robert y Javier, por darme siempre el empujoncito que necesitaba en
aquellos momentos en que estaba atascada. Sus consejos y buenas palabras son un
tesoro.

A Jaime y Luis, por creer en mí desde el principio y darme la maravillosa oportunidad de

seguir creciendo día a día.

A mis padres, Federico y Conchi, por ser referentes en mi vida y no cortarme nunca las
alas para alcanzar mis sueños. A mi hermana, Ana, mi apoyo incondicional.

A mi familia (Cristian, Leo y Vera), mi mayor logro, cuyo apoyo y amor me ha

acompañado durante todo el proceso. Sin ellos esto no habría sido posible.

¡Gracias! a todos los que de alguna manera han contribuido a que haya llegado a la
meta.
RESUMEN
El cultivo de la vid es una práctica ancestral en regiones mediterráneas como España y
la región de Aragón en particular. Las características climáticas actuales de esta región
la convierten en una zona especialmente óptima para su cultivo siendo un referente
mundial. En el contexto actual de calentamiento global, las alteraciones climáticas
esperadas (aumento de las temperaturas, variaciones en los regímenes pluviométricos
e intensificación de los eventos extremos) pueden suponer un gran peligro para la
idoneidad climática que caracteriza a la región. Por lo tanto, es esencial llevar a cabo
estudios precisos que evalúen el impacto del cambio climático a escala local y
orientados al sector vitícola en particular.
Con el fin de evaluar el impacto del cambio climático en el sector vitícola aragonés (como
caso de estudio particular) así como en el español (Península y Baleares) se han
considerado tres aspectos: variaciones en el clima promedio (temperatura y
precipitación), variaciones en eventos extremos (olas de calor y frío así como episodios
de sequía) e indicadores bioclimáticos de interés vitícola.
Para ello, se han generado por primera vez escenarios de clima futuro a escala local de
temperatura, precipitación, olas de calor y frío, indicadores de sequía e indicadores
bioclimáticos basados en nueve Earth System Models (ESMs) y dos Representative
Concentration Pathway (RCP4.5 y RCP8.5) pertenecientes a la quinta fase del Coupled
Model Intercomparison Project (CMIP5). Para la generación de escenarios a escala local
se ha empleado una metodología de downscaling estadístico en dos pasos desarrollada
por la Fundación para la Investigación del Clima (FIC) y que aporta un valor añadido a
la toma de decisiones en materia de adaptación al cambio climático.
Como indicadores de sequía se han utilizado el Índice estandarizado de Precipitación
(SPI) y el Índice estandarizado de Precipitación y Evapotranspiración (SPEI). Como
indicadores bioclimáticos se han empleado el Índice de Huglin (HI), el Índice de frío (CI),
el Índice de sequedad (DI) y el Índice hidrotérmico. Como complemento, se calcularon
dos indicadores combinados: el Sistema de Clasificación Climática Multicriterio (Sistema
MCC) y el Índice Compuesto (CompI).
Los resultados de temperatura y precipitación van en línea con las tendencias
observadas en las últimas décadas: aumentos progresivos de temperatura (siendo más
intensos en los meses estivales) y variaciones en el régimen pluviométrico a nivel de
concentración (es decir, misma cantidad de precipitación registrada en menos tiempo).
En promedio para todo el territorio, a finales de siglo y bajo el escenario más
desfavorable, se espera que las temperaturas experimenten sus mayores ascensos en
los meses de verano, pudiendo aumentar hasta 7 ºC en el caso de las temperaturas
máximas y 5.8 ºC en las temperaturas mínimas. Estos aumentos harán que la meseta
sur, la costa mediterránea y las Islas Baleares superen fácilmente los 38 ºC. Los
ascensos menos acusados se esperan en los meses de invierno (hasta 4 ºC) tanto para
la temperatura máxima como la mínima.
Los episodios de olas de calor se espera que aumenten, en promedio para la región y
bajo el escenario RCP8.5 a finales de siglo, su intensidad media en más de 2 ºC
(superándose los 38.8 ºC) y su intensidad máxima en 3.6 ºC (lo que supondrían
temperaturas por encima de los 41 ºC). Además se espera que la duración de la misma
aumente hasta alcanzar de 12 a 20 días, en función del área a la que afecten. Por el
contrario, los episodios de olas de frío se mantendrán en la ocurrencia e intensidad
actual.
Respecto a los episodios de sequía, los resultados difieren según se analice el SPI
(basado exclusivamente en precipitación) o el SPEI (considerando el efecto de la
temperatura a través de la Evapotranspiración). En el primero de los casos, no se
aprecian cambios de interés y la región se mantendrá en valores normales de
alternancia de periodos secos y húmedos. En el segundo de los casos, bajo el RCP8.5,
se aprecia como toda la región tiende hacía episodios de sequía cada vez más intensos
y extensos en el tiempo, sufriendo episodios de sequía de severos a extremos.
A nivel de territorio, la meseta sur peninsular, especialmente Andalucía, Extremadura y
Murcia, junto con el Valle del Ebro serán las regiones más afectadas desde el punto de
vista climático. Por un lado, son las zonas donde se espera que se alcancen los
incrementos y las temperaturas más elevadas y por otro lado, son zonas bastante secas,
con poca pluviometría. La combinación de ambas características hyará que estas
regiones sufran fuertes episodios de sequía severa.
El hecho de que una de las zonas dónde se esperan que los cambios de temperatura
sean más acusados sea el Valle del Ebro, es crucial para Aragón. Esta zona es la más
poblada y de mayor concentración socioeconómica de la región lo que puede suponer
un gran riesgo para la salud, la mortalidad, la movilidad y el bienestar socioeconómico
entre otros factores.
En base a los resultados obtenidos para los indicadores bioclimáticos se espera que los
indicadores térmicos (HI y CI) tiendan a aumentar a lo largo del siglo XXI, mientras que
la escasez de agua (DI) será más pronunciada. Las tendencias encontradas no tienen
la misma repercusión en todo el territorio. En el sur de la península, con valores de HI
superiores a 3500°C y de CI por encima de 20°C y DI por debajo de -200 mm, la
continuidad del sector vitivinícola en su estado actual se ve seriamente amenazada, con
una disminución de los años climáticamente óptimos como muestran los valores de
CompI. Por el contrario, el norte peninsular y las zonas montañosas, a pesar de los
aumentos previstos, con HI inferiores a 2500°C, noches frescas (IC inferior a 15°C) y un
aporte hídrico suficiente (DI superior a 150 mm) mejoran considerablemente su aptitud
climática (CompI) aunque se mantiene el riesgo de enfermedad de mildiu debido al
aumento de la temperatura y la humedad.
Disponer de información climática de calidad (basada en proyecciones climáticas
robustas y fiables) ajustada a las características micro climáticas de la región y
considerando los efectos climáticos que pueden tener implicaciones en el sector vitícola,
supone una información de valor añadido imprescindible en la toma de decisiones para
la adaptación al cambio climático.
Índice

1. INTRODUCCIÓN Y ANTECEDENTES .............................................................................. 1

1.1 CAMBIO CLIMÁTICO Y MODELIZACIÓN CLIMÁTICA .......................................... 2
1.1.1. Los modelos climáticos: herramienta para la simulación del clima ....... 4
1.1.2. Necesidad de pasar de escala global a escala local.................................... 9
1.1.3. Datos necesarios para llevar a cabo el downscaling estadístico .......... 11
1.1.4. Procesos de verificación y validación ........................................................... 12
1.1.5. Proyecciones climáticas futuras ..................................................................... 14
1.1.8. Simulaciones futuras de variables derivadas ............................................. 16
1.1.9. Consideraciones finales a la modelización climática................................ 16
1.2 EL RETO DE LOS EVENTOS EXTREMOS ............................................................. 17
1.2.1. Episodios de olas de calor y frío..................................................................... 21
1.2.2. Episodios de sequía ........................................................................................... 23
1.3 EL IMPACTO DEL CAMBIO CLIMÁTICO SOBRE EL CULTIVO DE LA VID:
INDICADORES BIOCLIMÁTICOS .................................................................................... 28
1.3.1. Clima y vid ............................................................................................................. 29
1.3.2. La vid: ¿cómo le influyen las variables meteorológicas? ........................ 31
1.3.3. Cambio Climático y viñedos............................................................................. 37
1.3.4. Indicadores bioclimáticos como herramientas para el estudio del
viñedo ................................................................................................................................ 39
1.3.5. Denominaciones de Origen .............................................................................. 40
2. JUSTIFICACIÓN Y OBJETIVOS ...................................................................................... 42
3. METODOLOGÍAS ............................................................................................................... 46
3.1. Generación de escenarios de clima futuro de temperatura y precipitación
................................................................................................................................................. 47
3.1.1. Metodología de downscaling estadístico en dos pasos........................... 48
3.1.2. Procesos de verificación y validación ........................................................... 49
3.1.3. Corrección del error sistemático .................................................................... 49
3.2. Generación de escenarios de clima futuro de valores extremos .................. 49
3.2.1. Olas de calor/frío ................................................................................................. 49
3.2.2. Episodios de sequía ........................................................................................... 50
3.3. Generación de escenarios de clima futuro de indicadores bioclimáticos .. 50
3.4. Datos y área de estudio ............................................................................................ 54
3.4.1. Área de estudio .................................................................................................... 54
3.4.2. Bases de datos empleadas ............................................................................... 57
 3.4.3. Denominaciones de Origen .............................................................................. 61
4. RESULTADOS ..................................................................................................................... 62
Resultado 1 .......................................................................................................................... 63
Proyecciones climáticas futuras de temperatura y de episodios de olas de
calor y de frío para Aragón (España) en base a las salidas de modelos
pertencientes al CMIP5 ................................................................................................. 63
Material suplementario Resultado 1.......................................................................... 89
Anexo R1. Escenarios de temperatura y olas de calor y frío para Península
Ibérica y Baleares ......................................................................................................... 100
Resultado 2 ........................................................................................................................ 109
Impacto del cambio climático en la ocurrencia de eventos de sequía de
Aragón (NE de España) .............................................................................................. 109
Material suplementario Resultado 2........................................................................ 135
Anexo resultado 2. Escenarios de precipitación sequía para Península
Ibérica y Baleares ......................................................................................................... 142
Resultado 3 ........................................................................................................................ 149
Evaluación del impacto del cambio climático en el sector vitícola español
mediante indicadores bioclimáticos ....................................................................... 149
Material suplementario Resultado 3........................................................................ 179
5. DISCUSIÓN GENERAL.................................................................................................... 194
5.1. Consideraciones generales sobre la generación de escenarios de clima
futuro ................................................................................................................................... 195
5.2. Consideraciones sobre los escenarios de clima futuro a escala local en la
región de Aragón: temperatura y olas de calor/frío ................................................ 196
5.3. Consideraciones sobre los escenarios de clima futuro a escala local en la
región de Aragón: precipitación y sequía meteorológica ..................................... 199
5.4. Consideraciones sobre los escenarios de indicadores bioclimáticos de
interés vitícola a escala local en España ................................................................... 200
5.5. Consideraciones sobre posibles medidas de adaptación y mitigación .... 204
5.6. Aspectos metodológicos a considerar ............................................................... 207
6. CONCLUSIONES .............................................................................................................. 209
BIBLIOGRAFÍA ...................................................................................................................... 214
ANEXO DE PUBLICACIONES ............................................................................................ 239
1. INTRODUCCIÓN Y
ANTECEDENTES

1
1.1 CAMBIO CLIMÁTICO Y MODELIZACIÓN CLIMÁTICA

El cambio climático es, sin lugar a dudas, uno de los grandes retos a los que se enfrenta
la humanidad en el próximo futuro a nivel planetario. Desde que se crea en 1998 el IPCC
(Panel Intergubernamental sobre Cambio Climático por sus siglas en inglés) con el fin
de aunar esfuerzos en la lucha frente al cambio climático, se han podido hacer
evaluaciones integrales del estado de los conocimientos científicos, técnicos y
socioeconómicos sobre el cambio climático, sus causas, posibles repercusiones y
estrategias de respuesta. Ha sido el encargado de unificar la metodología de trabajo así
como la definición de los nuevos escenarios de clima futuro que han de ser utilizados
por los centros de investigación en las simulaciones climáticas.

Los distintos informes de evaluación presentados por el IPCC destacan detalladamente

los cambios en el clima ya registrados. Los datos observados no dejan lugar a dudas:
se esperan cambios en los patrones de precipitación, aumento de las temperaturas,
cambios en la ocurrencia de eventos extremos, alteraciones en las fechas de los ciclos
fenológicos de los cultivos, eventos de sequía y olas de calor y frío más intensos y
alteraciones en la aparición de enfermedades en cultivos, entre otros- Esto pone de
manifiesto la necesidad de adaptarse a la nueva realidad climática y estar preparados
para los cambios que se espera que ocurran en el futuro.

Las primeras conclusiones presentadas por el sexto informe del IPCC6, por sus siglas
en inglés) (IPCC, 2021a y b) son contundentes respecto a la realidad actual (e
indiscutible) de emergencia climática. Estas conclusiones, centradas en las bases
físicas del clima, ponen de manifiesto cómo esta situación además está directamente
ligada a la actividad humana, de manera que no hay ninguna región del planeta que no
se haya visto afectada por algún efecto relacionado con el cambio climático.

Hay dos puntos clave en los que el informe es tajante: "la temperatura de la superficie
global continuará incrementándose al menos hasta mitad de siglo teniendo en cuenta
todos los escenarios posibles de emisiones" y “Es inequívoco que la actividad humana
ha calentado la atmósfera, el océano y la superficie terrestre". Esto significa que, incluso
en el escenario menos desfavorable, la temperatura seguirá en ascenso, ya que serán
necesarios como mínimo 30 años para poder encauzar el daño ocasionado por las
actividades humanas desarrolladas en las últimas décadas. Estas conclusiones no
pretenden ser alarmistas, sino remarcar que, igual que la actividad humana es la
principal causante del estado actual, también es la única que puede frenarla y que está
a tiempo de ello.

Actuar contra el cambio climático es un reto realmente importante. Muchos de los

cambios observados en las últimas décadas no tienen precedentes de situaciones
similares (IPCC, 2021a y b) por lo que las medidas a adoptar no están definidas. Para
hacer frente a esta problemática se tienen tres vías de trabajo. En primer lugar, diseñar
medidas de mitigación a gran escala, que sean inminentes y contundentes a la hora de
reducir su impacto. A continuación, adoptar medidas de adaptación adecuadas a cada
sector y a su problemática particular y por último, entender la investigación como una
herramienta de apoyo para adaptarse al cambio climático.

En las últimas décadas la comunidad científica ha centrado gran parte de sus esfuerzos
en simular el clima futuro a través del desarrollo de Modelos Climáticos (MC), los cuales

2
pretenden representar el comportamiento e interacciones entre los distintos
componentes del sistema climático y su evolución en las próximas décadas. Las
proyecciones climáticas simuladas por los MCs son los pilares en los que se basarán
los estudios de evaluación de impactos y se sustentarán las medidas de adaptación.
Por tanto, disponer de un conjunto fiable y robusto de proyecciones climáticas es el
punto de partida en cualquier estudio relacionado con el impacto que el cambio climático
pueda tener en cualquier sector.

Son varios aspectos los que han de tenerse en cuenta en el estudio del impacto del
cambio climático: 1) variaciones en las condiciones climáticas medias de la región, 2)
alteraciones en la frecuencia de ocurrencia e intensidad de los eventos extremos y 3)
consecuencias de dichas modificaciones climáticas. Los impactos como consecuencia
del cambio climático no afectan de forma homogénea a todas las regiones por lo que es
esencial que estos estudios se realicen a escala local considerando las distintas
condiciones micro meteorológicas que coexisten dentro de una misma región.

Las proyecciones climáticas de variables como la temperatura y la precipitación

proporcionan una visión del impacto general del cambio climático como consecuencia
del calentamiento global. A partir de esta información, se evalúan variables extremas
(como las olas de calor y los episodios de sequía) que proporcionan información sobre
eventos que no ocurren con tanta frecuencia y que completan la visión en conjunto del
cambio climático en una región, ya que se dispondrá de información promedio junto con
información de extremos. Pero esta información no es suficiente para evaluar el impacto
del cambio climático en la vid y se requiere de información climática adaptada a sus
requerimientos heliotérmicos e hídricos, los cuáles se evalúan mediante indicadores
bioclimáticos (Huglin 1978; Maglhaes 2008; Malherio et al., 2010).

En base a lo anteriormente expuesto, determinar la relación entre clima y viñedo y

evaluar su evolución futura, cobra un especial interés en regiones como España donde
el sector vitícola no es solo importante en términos de biodiversidad sino también
socioeconómicos.

España es una de las regiones más afectada por los efectos del cambio climático. Se
espera que en las próximas décadas la temperatura media ascienda en todo el territorio
español entre 2 y 4 °C y que se produzcan ligeros cambios en la cantidad de
precipitación acumulada anualmente. Los eventos de precipitación serán más
concentrados en el tiempo (es decir, misma cantidad pero precipitada en menos días),
y los fenómenos extremos serán cada vez más intensos (Porter and Semenov, 2005;
Gaitán et al., 2019, 2020). Estas alteraciones climáticas supondrán un aumento de la
demanda de regadío (Doll 2002), un aumento en la ocurrencia de enfermedades y
plagas (Alig et al., 2002) así como cambios en la zonificación vitícola (Malheiro et al.,
2010) entre otros impactos. Esto significa que eventos tales como olas de calor, sequías
o precipitaciones extremas, entre otros, se producirán de forma más recurrente
aumentándose considerablemente el riesgo socio-económico asociado a su presencia.

Existen varios estudios centrados en técnicas de regionalización o downscaling de

temperatura para España (Ribalaygua et al., 2013a, 2013b; Frías et al., 2005; Hervada-
Sala et al., 2000; Turco et al., 2014) pero estos no abordan la problemática de los
eventos extremos de temperatura (ya sean días con temperaturas muy altas/muy bajas
u olas de calor/frio). Los pocos estudios que existen relacionados con este aspecto se

3
centran en estudiar eventos históricos (Fernández-Montes and Rodrigo 2012; Fonseca
et al., 2016). En el caso concreto de Aragón, hay pocos estudios que hayan desarrollado
escenarios a escala local de temperatura y estos, en general, no contemplan situaciones
extremas ni usan modelos pertenecientes a la quinta fase del Proyecto de
Intercomparación de Modelos acoplados (CMIP5, en sus siglas en inglés, Coupled
Model Intercomparison Project) (Buerger et al., 2007; Goncalves et al., 2014; Ribalaygua
et al., 2013a).

Respecto a la generación de escenarios futuros de precipitaciones a escala local,

existen estudios que generan proyecciones de precipitación en base a distintas técnicas
de downscaling y conjunto de modelos climáticos para España (Osca et al., 2013; Turco
et al., 2011; Barrera-Escoda et al., 2014) aunque la mayoría con modelos anteriores al
CMIP5. El resto de estudios basados en modelos del CMIP5 se centran en regiones
concretas como una cuenca hidrográfica (Miró et al., 2021), la región mediterránea
(Monjo et al., 2016) o toda la península y lo hacen en base a modelos regionales
generados en el proyecto Coordinated Downscaling Experiment - European Domain
EURO-CORDEX (Lorenzo and Alvarez, 2021).

Por tanto, hasta donde sabemos no existía hasta la fecha ningún estudio para Aragón
en base a CMIP5 y mediante downscaling estadístico que simule proyecciones
climáticas ni de temperatura ni de precipitación. A nivel peninsular existen diversos
estudios publicados sobre proyecciones climáticas en base al CMIP5. La mayoría de
dichos estudios utilizan las salidas directas de los modelos o las salidas de EURO-
CORDEX, por lo que no utilizan una metodología estadística.
1.1.1. Los modelos climáticos: herramienta para la simulación del clima

A día de hoy la herramienta más potente en la simulación del clima futuro son los
modelos climáticos (MCs). Los MCs resuelven las ecuaciones fundamentales de las
leyes y principios de la física que rigen los procesos en cada componente del sistema
climático global (atmósfera, hidrosfera, criosfera, litosfera y biosfera), como los
intercambios de energía y de masa, mediante métodos numéricos (figura 1). Los MCs
están compuestos por diferentes módulos, cada uno de los cuales reproduce los
procesos de cada componente del sistema climático, siendo los principales módulos
atmosféricos (AGCM, de sus siglas en inglés) y los oceánicos (OGCM, de sus siglas en
inglés).

4
 Figura 1. Esquema de ecuaciones y procesos físicos simulados por un MC. Elaboración propia.

Estas ecuaciones son ecuaciones en derivadas parciales, es decir, no lineales y no es

posible hallar una solución analítica, por lo que en muchas ocasiones es preciso recurrir
al análisis numérico para determinar una solución aproximada. Para poder resolverlas,
los MCs consideran un espacio tridimensional discretizado en una cuadrícula (150km x
150km en la atmósfera y 100km x100km en el océano), así como la evolución temporal
(diaria). En cada punto de la cuadrícula, se representan los valores de las variables del
modelo y se resuelven las ecuaciones fundamentales obteniendo valores de las
variables meteorológicas para un cierto instante en el futuro. Aquellos procesos que la
rejilla del modelo no puede representar explícitamente (porque son de menor escala) se
parametrizan, es decir, se estudian como si fueran variables derivadas a partir de otras
y no como variables independientes (por ejemplo, la precipitación convectiva, la
microfísica de las nubes, los flujos de la capa límite planetaria etc.) (figura 2).

Figura 2. Figura esquemática de un MC y del concepto de rejilla espacial. Fuentes: National Oceanic and Atmospheric
Administration (NOAA) y Climate Research Unit (CRU), 2000.

5
Por lo tanto, las características que van a definir el comportamiento de un MC serán la
resolución espacial, el nivel de parametrización definido y la complejidad computacional
en la ejecución del modelo, diferenciando aquellos modelos simples de los más
complejos.

Los MCs han evolucionado mucho desde los años sesenta. De los modelos más simples
(condicionados por las limitaciones computacionales y la simple representación de los
procesos físicos a escala bidimensional) pasando por los modelos de circulación general
(representación tridimensional del sistema atmósfera-océano así como la mejora en la
capacidad computacional) hasta llegar a los más actuales.

La última generación de modelos climáticos son los Earth System Models (ESM), que
describen los procesos que tienen lugar dentro y entre la atmósfera, el océano, la
criosfera y la biosfera marina y terrestre. A parte de simular las ecuaciones físicas que
rigen el comportamiento de la atmósfera y del océano, estos modelos además recogen
los mecanismos químicos y biológicos que gobiernan a los elementos del sistema
terrestre e incluyen erupciones volcánicas y variaciones de la radiación solar entrante.

Otro de los avances incluidos en los ESMs es la representación del ciclo del carbono,
permitiendo el cálculo interactivo de las emisiones atmosféricas de CO 2 o similares.
También pueden incluir otros componentes, como la química de la atmósfera, las capas
de hielo, la vegetación dinámica, el ciclo del nitrógeno, modelos urbanos o de cultivos,
etc. El principal avance frente a los MCs utilizados anteriormente es que permiten la
interacción del sistema con el ciclo del carbono y tienen en cuenta la bioquímica y la
biogeología marina (figura 3).

Figura 3. Características principales de los MCs y de los ESMs. Fuente: Heavens et al. 2013

Cada nueva generación de modelos climáticos está vinculada a una nueva fase del
IPCC que establece las líneas de desarrollo generales que todos los centros
responsables de los modelos han de tener en cuenta. Por ejemplo: rango de tamaños
en las rejillas, cobertura espacial y temporal, variables mínimas a simular y número
mínimo de escenarios de emisiones con los que trabajar, entre otros. Es por eso por lo
que según se han ido sucediendo las distintas fases del IPCC, se ha ido modificando la
manera en que se definen los escenarios de emisiones sobre los que se construyen las
simulaciones o escenarios climáticos futuros. Los ESMs son la base del Climate Model

6
Intercomparison Project en sus fases quinta CMIP5 (Taylor et al., 2012) y sexta CMIP6
(Eyring et al., 2016) y constituyen los conjuntos más actuales de experimentos
coordinados de modelos climáticos disponibles. El CMIP es el organismo encargado de
hacer públicos los resultados de los múltiples modelos en formato estandarizado y sus
distintas fases van surgiendo de forma paralela al IPCC. Además, entre sus otros
objetivos, se encuentra comprender los cambios climáticos pasados, presentes y futuros
derivados de la variabilidad natural no forzada o en respuesta a cambios en el
forzamiento radiativo en un contexto multimodelo. Esta comprensión incluye la
evaluación del rendimiento de los modelos durante el periodo histórico y la cuantificación
de las causas de la dispersión en las proyecciones futuras.

Según el IPCC (IPCC 2013: Glosario) un escenario climático es:

“una representación plausible y a menudo simplificada del clima futuro, basada en un

conjunto internamente coherente de relaciones climatológicas, que se construye para
ser utilizada de forma explícita en la investigación de las consecuencias potenciales
del cambio climático antropogénico, y que sirve de insumo para las simulaciones de
los impactos”

De manera que un escenario de cambio climático será la diferencia entre un escenario

climático concreto y el clima actual.

Los escenarios climáticos se definen por la necesidad de evaluar cómo y cuánto se verá
afectado el clima futuro a consecuencia de la actividad humana (actividad
antropogénica).

Hay dos formas de evaluar la influencia de la actividad humana. Por un lado,

considerarla responsable de una emisión futura de gases de efecto invernadero y otras
sustancias, entonces hablaremos de escenarios de emisiones o por otro lado,
considerando el efecto final de las emisiones, es decir, el forzamiento radiativo. En este
caso, hablaremos de escenarios de forzamiento radiativo.

Figura 4. Diferencias entre escenario de emisiones y escenario de forzamiento radiativo. Definiciones extraídas del
Glosario del IPCC (IPCC, 2013: Glosario)

7
El concepto de escenario de emisiones fue usado hasta el Cuarto informe del IPCC
(IPCC 2007), de manera que en el Quinto informe (IPCC 2013) se ha trabajado con los
escenarios de forzamiento radiativo.

Los escenarios futuros asociados al quinto informe del IPCC son los Representative
Concentration Pathways (RCPs), definiéndose cuatro escenarios (en función del
forzamiento radiativo alcanzado en 2100): RCP2.6, RCP4.5, RCP6.0 y RCP8.5. Estos
escenarios se agrupan en lo que el IPCC denomina “Tier 1”, aquellos RCPs que han de
ser sí o sí simulados por todos los ESMs, y “Tier 2”, aquellos que son opcionales. Al
primer grupo pertenecerían los RCPs 4.5 y 8.5 y al segundo el resto; aunque la mayoría
de los modelos proporcionan información de al menos RCP 2.6, 4.5 y 8.5.

El nombre de los RCPs viene dado por dos de sus características principales:

1) Representative. Los RCPs se basan en escenarios de emisiones ya existentes, es

decir, cada RCP representa un conjunto de escenarios de emisiones ya existentes
(diferentes maneras de emitir gases de efecto invernadero y otras sustancias) con un
mismo forzamiento radiativo en el 2100. Por lo que el RCP debe ser compatible tanto
con escenarios extremos como con escenarios medios.

2) Concentration Pathway. Este término hace hincapié en que los RCPs son la
herramienta para la generación de escenarios de emisiones, y no escenarios en sí
mismos, ya que, si bien recogen las distintas componentes de los forzamientos
radiativos, no son un conjunto completo de proyecciones climáticas, socioeconómicas y
de emisión.

Figura 5. Esquema de los forzamientos radiativos y los escenarios de emisiones de CO2 bajo distintos RCPs. (Figura
basada en IPCC 2014)

8
Por lo tanto, para simular el clima futuro, se establecen hipótesis de cómo puede ser en
el futuro el forzamiento radiativo según evolucione la sociedad. Bajo estas hipótesis se
corren los MCs y se obtienen los escenarios climáticos.

A pesar de que los modelos climáticos suelen reproducir de forma adecuada las
características generales de la atmósfera presentan ciertas limitaciones, de manera que
tienen dificultades a la hora de simular fenómenos de pequeña escala o muy locales
como consecuencia 1) de su insuficiente resolución espacial (2-3º), 2) la topografía está
descrita con poco detalle dejándose fuera algunos forzamientos relacionados con la
misma y 3) los procesos recogidos mediante parametrizaciones directas se ajustan de
manera estadística para todo el planeta pudiendo ser ineficientes en regiones concretas.

1.1.2. Necesidad de pasar de escala global a escala local

Dado que en la mayor parte de los estudios de evaluación de impactos es necesaria la

presencia de escenarios climáticos con resolución local de variables cercanas a la
superficie terrestre (temperatura a 2 m., precipitación, etc.), surge la necesidad de
adaptar la información proporcionada por los ESMs (de baja resolución espacial) a la
información requerida por los modelos de impacto (de mayor resolución espacial-local
en superficie).

Los estudios más recientes ponen de manifiesto la existencia de un cambio climático

global, pero estas alteraciones no afectan de la misma manera a todas las regiones del
mundo y dentro de una misma región, se observan cambios significativos ente puntos
muy próximos.

Figura 6. Cambio en la temperatura media de la superficie (a) y en la precipitación media (b) basados en las
proyecciones de la media multimodelo para 2081-2100 en relación con 1986-2005 bajo los escenarios RCP2.6
(izquierda) y RCP8.5 (derecha). El número de modelos utilizados para calcular la media multimodelo se indica en la
esquina superior derecha de cada panel. El punteado muestra las regiones en las que el cambio proyectado es grande
en comparación con la variabilidad interna natural, y en las que al menos el 90% de los modelos coinciden en el signo
del cambio. El rayado muestra las regiones en las que el cambio previsto es inferior a una desviación estándar de la
variabilidad interna natural. Fuente: IPCC5 Report

9
En muchos casos, conocer la evolución del clima en una región o como promedio en
una celda de un MC (200km x 200km) no es suficiente, necesitándose información a
escala mucho más local. Este proceso de reducción de escala es conocido como
regionalización o downscaling. En estos procesos se combina la información
proporcionada por un MC con información más detallada como la topografía, la
meteorología local, los usos del suelo etc.

Figura 7. Figura esquemática de regionalización o downscaling (Fuente: David Viner, Climatic Research Unit, University
of East Anglia, UK, 2000)

Existen múltiples técnicas de downscaling, desde las más básicas basadas en aplicar
algún tipo de tratamiento estadístico a las distintas salidas de los modelos climáticos
(denominadas MOS, Model Output Statistics) hasta las más avanzadas, fundamentadas
en procesos físicos. Dentro de estas últimas se encuentran las técnicas de downscaling
dinámico y downscaling estadístico:

 Downscaling estadístico: se obtienen relaciones empíricas entre variables a

gran escala (configuraciones atmosféricas de baja resolución, denominadas
predictores) procedentes de los Modelos Climáticos y variables de alta
resolución en superficie cuya simulación se quiere obtener (denominadas
predictandos).
 Downscaling dinámico: se incrementa la resolución del modelo en la zona de
interés mediante anidando un modelo climático regional (RCM, por sus siglas en
inglés) en las condiciones de contorno del modelo climático global o realizando
una técnica de “zoom” de la rejilla del propio modelo climático global.

Cada técnica tiene ciertas ventajas e inconvenientes que hacen que una sea más
apropiada que otra en función de las características del estudio en el que se vayan a
utilizar. En términos generales el downscaling dinámico tiene una base física más fuerte
y no requiere de una base de datos observados histórica frente al downscaling
estadístico que exige asumir la hipótesis de que las relaciones detectadas en el pasado
se mantendrán en el futuro, pudiendo provocar efectos de sobreajuste. Además requiere
de una base de datos observados histórica, lo que supone un factor muy limitante en

10
aquellas regiones con pocas observaciones. Por otro lado, el downscaling estadístico
permite recoger la micrometeorología local (siendo la mejor técnica en zonas de
compleja orografía) y generar un mayor número de proyecciones climáticas (facilitando
la cuantificación de la incertidumbre). Sin olvidar que su coste computacional es muy
bajo, mientras que el elevado coste computacional de los modelos dinámicos restringe
considerablemente el número de proyecciones climáticas a generar y su resolución
espacial (en torno a los 25km) mantiene alguno de los problemas asociados a la
resolución de los modelos climáticos

Figura 8. Características principales de las distintas técnicas de downscaling. Elaboración propia.

En base a sus ventajas y desventajas, las técnicas estadísticas serán las más
adecuadas en aquellas regiones ricas en datos meteorológicos y con una orografía
compleja, gran diversidad climática y ocurrencia de fenómenos de pequeña escala
(como la precipitación convectiva), mientras que las dinámicas serán de gran utilidad en
aquellas regiones pobres en datos observados y sin una meteorología muy local
(fenómenos micro meteorológicos fuera del alcance de la rejilla del modelo).

1.1.3. Datos necesarios para llevar a cabo el downscaling estadístico

 Datos observados procedentes de estaciones meteorológicas

Las técnicas de downscaling estadístico requieren de datos meteorológicos observados

ubicados en la zona sobre la que se realiza el estudio. Estos datos deben cumplir unos
estándares mínimos de calidad por lo que:

1) han de pasar por distintos controles que aseguren su fiabilidad. Todas las series de
datos se ven sometidas a distintos tests que detectan inhomogeneidades, datos
anómalos y huecos, desestimándose aquellas series que no los superen.

2) deben tener una extensión mínima de 5 años. Aunque lo deseable es que su

extensión sea lo mayor posible alcanzando los 30-50 años en muchas zonas esto no es
posible.

11
3) lo ideal es que cubran la mayor extensión del territorio y que lo hagan de forma lo
más homogénea posible.

 Reanálisis

Un reanálisis es un conjunto de estimaciones históricas sobre una malla regular de

variables meteorológicas y oceánicas generado por modelos avanzados de predicción
climática y oceánica mediante técnicas de simulación de datos. Los modelos se aplican
sobre un conjunto de datos observados procedentes, no solo de observatorios
meteorológicos, sino de múltiples fuentes de datos observados como las boyas marinas,
los globos sonda, los radiosondeos, los satélites y los radares.

El reanálisis ha de ser sometido a un proceso de calidad que permita eliminar aquellos

valores que no sean coherentes como, por ejemplo, precipitaciones negativas, áreas sin
datos, humedades relativas superiores al 100% o temperaturas mínimas superiores a
las máximas.

Los MCs necesitan ser validados frente a una base de datos observados completa, es
decir, sin datos faltantes. Por lo que el uso de un reanálisis es imprescindible para poder
aplicar el proceso de validación de los modelos.

 Modelos Climáticos

El último set de datos que se requiere son los modelos climáticos. Las técnicas
estadísticas, debido a su rapidez computacional, permiten trabajar con un número
elevado de modelos climáticos (n) y RCPs (m), por lo que se obtendrá un conjunto de
(n x m) proyecciones climáticas. Los MCs se corren de forma continua desde el pasado
hacía el futuro, una vez que se simula el periodo de control, la ejecución se separa en
tantas ejecuciones como RCPs se consideren. De cada modelo climático se dispone,
por tanto, de una simulación de control denominada “Historical” para el periodo 1951-
2005 y de entre 2 y 4 RCPs para el periodo 2006-2100.

Al igual que se hace con los datos observados y con el reanálisis, las simulaciones de
control también han de someterse a controles de calidad para evitar la presencia de
datos incoherentes.

La salida de control o Historical es imprescindible, no solo, para la generación de las

simulaciones futuras, sino también para evaluar el comportamiento de cada modelo
climático simulando el clima actual durante el proceso de validación.

1.1.4. Procesos de verificación y validación

Antes de proceder a generar las proyecciones climáticas a escala local, se llevan a cabo
dos procesos que permiten evaluar el comportamiento de la metodología y el
funcionamiento de cada modelo climático, información que ha de tenerse en cuenta e
incluirse en las simulaciones futuras. Ambos procesos se realizan a escala local, es
decir, en cada punto donde hay datos de un observatorio meteorológico.

12
  Verificación de la metodología

El proceso de verificación permite evaluar cómo de buena es la metodología de

downscaling empleada a la hora de simular el clima actual, generalmente los últimos 30-
50 años, según las bases de datos empleadas. Para ello, se comparan las series
simuladas de los datos del reanálisis mediante downscaling, con los datos observados
para un periodo común.

Durante el proceso de verificación han de tenerse en cuenta algunos aspectos:

1) Se han de simular correctamente tanto los valores medios como los extremos.

2) Se ha de simular correctamente tanto la escala climática (promedios de 30 años)

como la diaria (cambios de un día a otro).

3) Aplicar estudios de robustez como análisis de años cálidos/fríos o secos/húmedos.

Lo que se pretende con este análisis es comprobar que si los años elegidos para simular
son cálidos (secos) la metodología es capaz de simular años fríos (húmedos) y
viceversa.

4) Para poder realizar este proceso, la resolución espacial y temporal del reanálisis ha
de adecuarse a la del modelo climático con el que se va a trabajar posteriormente. La
resolución espacial de los reanálisis suele ser más fina que la de los MCs por lo que su
rejilla se “relajará” a la resolución del MC extrapolando. En el caso de la resolución
temporal se utilizará la diaria.

5) Dado que cada dato de la serie simulada por downscaling del reanálisis se
corresponde con un día con fecha real, es posible comparar dichas series con los datos
observados aunque no existen observaciones para todos los días del periodo, ya que
las comparaciones se hacen día a día. Este aspecto es fundamental, ya que sugiere
que se están reproduciendo de forma satisfactoria las relaciones físicas entre
predictores y predictandos.

 Validación de los modelos climáticos

En general, los MCs tienden a simular un clima más frío, cálido, húmedo o seco de lo
que éste es en realidad y además, no simulan de la misma manera todas las regiones
del planeta. Esta diferencia entre modelos a la hora de simular el clima actual es
consecuencia de las características de cada modelo, de las condiciones de contorno
establecidas y de las parametrizaciones asumidas. Por lo tanto, determinar como de
bueno es un MC simulando el clima de la región de estudio es una información clave a
tener en cuenta en la generación de las proyecciones climáticas futuras. El proceso
mediante el cual se evalúa a los MCs se conoce como validación y consiste en comparar
las series simuladas mediante downscaling a partir de la salida de control (Historical)
del MC con las series simuladas mediante downscaling a partir del reanálisis para un
periodo común.

En contra de lo que ocurría en el caso de la verificación, en la validación no se pueden

comparar las series simuladas día a día, ya que la salida de control no pretende simular
un día concreto sino las características climáticas generales, por lo que la comparativa
ha de hacerse a escala climática (en periodos de mínimo 30 años).

13
1.1.5. Proyecciones climáticas futuras

Una vez finalizados ambos procesos, si los resultados son buenos y se han descartado
aquellos MCs que no funcionan adecuadamente en la zona de estudio, se proceden a
generar las proyecciones climáticas futuras.

Por cada MC y RCP se tiene una simulación futura diaria hasta el 2100 sobre la que se
va a aplicar el downscaling, obteniéndose así un abanico de proyecciones futuras diarias
a escala local o regionalizadas. Aunque las series regionalizadas ofrecen datos diarios,
su interpretación ha de hacerse a escala climática (periodos de al menos 30 años) y
nunca sobre fechas concretas.

Las series simuladas pueden ser utilizadas desde dos enfoques:

1) atendiendo a los cambios esperados respecto al periodo de control. Para ello se

analizará la diferencia de la variable a analizar entre un periodo futuro (por ejemplo,
2041-2070) y el periodo de control (por ejemplo, 1971-2000). Esta información
proporciona una idea de cómo va a ser el cambio esperado de dicha variable.

2) atendiendo a los valores absolutos alcanzados. En ciertos estudios de impactos es

necesario evaluar los valores que va a alcanzar una variable (por ejemplo, saber si la
temperatura superará cierto umbral) más que la magnitud del cambio esperado.

Por ejemplo, si para un cultivo la temperatura máxima es crítica cuando se superan los
40 ºC en una época del año concreto, necesitaremos saber la evolución de la
temperatura máxima en valor absoluto mejor que la variación de dicha variable. Saber
que la variación esperada de temperatura máxima será de, por ejemplo, 2.5º֯C no nos
informa de si se superará el umbral crítico. En función de la aplicación que se le vaya a
dar a las simulaciones futuras, se escogerá un formato u otro.

En el caso de que sea necesario trabajar con valores absolutos, es necesario corregir
el error sistemático sobre las series simuladas. Para poder corregir dicho error, primero
se han de cuantificar las incertidumbres asociadas al proceso de downscaling.

1.1.6. Incertidumbres asociadas al proceso de downscaling. Corrección del error

sistemático

A lo largo del proceso de downscaling van surgiendo ciertas incertidumbres, es decir,

estados de conocimiento incompleto como, por ejemplo, imprecisión de los datos,
ambigüedad en la definición de un concepto o una proyección incierta de la conducta
humana.

Estas incertidumbres pueden dividirse en cuatro grandes grupos asociados a los

forzamientos radiativos futuros, a la técnica de downscaling empleada, a los modelos
climáticos y a la variabilidad natural del sistema climático y su ocurrencia de forma
encadenada por lo que reciben el nombre de cascada de incertidumbres.

14
 Figura 9. Esquema de las incertidumbres asociadas al proceso de downscaling. Elaboración propia.

Conocer estas incertidumbres es muy importante a la hora de manejar los escenarios

climáticos futuros. A mayor incertidumbre, mayor ha de ser la cautela y flexibilidad con
la que se manejen los escenarios. Cuantificar las incertidumbres es un proceso muy
complejo pero imprescindible para un análisis correcto de los escenarios.

Se puede intentar minimizar el efecto de la incertidumbre mediante la corrección del

error sistemático. El error sistemático se puede definir como el error que se comete al
simular cierta variable con respecto a su valor observado. En este caso, el error
sistemático es la combinación del error de verificación (asociado a la metodología de
downscaling) y el error de validación (asociado a cada uno de los distintos MCs). El error
asociado a la variabilidad climática y al forzamiento radiativo elegido se denominan
errores de impredecibilidad (no son cuantificables) y no se incluyen en el error
sistemático ya que no son medibles.

El error sistemático no se calcula directamente en el proceso de generación de

proyecciones climáticas futuras a escala diaria ya que la manera en la que se calcula
varía, no solo en función de la variable, sino también depende de la escala temporal a
considerar (diaria, mensual, estacional o anual).

1.1.7. Limitaciones de los escenarios climáticos futuros

Los escenarios de clima futuros son estimaciones que pretenden dan una idea de cómo
será el clima pero en ningún caso son predicciones categóricas y presentan ciertas
limitaciones que han de tenerse siempre en cuenta:

1) no todas las variables se simulan con la misma robustez. Por ejemplo, las
simulaciones de temperatura presentan mayor fiabilidad que las de precipitación ya que
esta última variable presenta una variabilidad interna mucho mayor que dificulta su
simulación.

2) al ser un estudio local, es decir, considerando la información proporcionada por los

datos observados es muy importante que la información de partida sea de calidad y que

15
el observatorio esté ubicado lo más cerca posible y mejor correlacionado,
climáticamente hablando, con el punto donde se quiere hacer el estudio.

3) ninguna proyección climática (asociada a una combinación MC-RCPs) es más

probable que otra por lo que cuanto mayor sea el número de combinaciones MC-RCP
más amplio será el abanico de proyecciones climáticas con el que trabajar y poder
cuantificar las incertidumbres en las proyecciones.

1.1.8. Simulaciones futuras de variables derivadas

Una vez que se han generado series de temperatura y precipitación futuras a escala
local, estas suponen el input para el cálculo de las variables derivadas de interés, como
por ejemplo, episodios de olas de calor, índices de sequía o indicadores bioclimáticos.

El cálculo de las variables derivadas también lleva un proceso de verificación en el que

se comparan los resultados obtenidos al calcular las variables en base a datos
observados con las variables obtenidas sobre las series simuladas del reanálisis.

1.1.9. Consideraciones finales a la modelización climática

Mediante las técnicas de downscaling es posible adaptar la salida de los modelos

climáticos (con resolución media de 200 km) a escala local y en el caso de las técnicas
estadísticas incorporar información meteorológica mediante datos observados. Al final
del proceso, se obtienen series simuladas diarias a escala local para todo el siglo XXI
de temperatura (máxima y mínima) y precipitación. El número de proyecciones
climáticas depende de la cantidad de modelos climáticos y RCPs considerados pero lo
ideal es trabajar con el mayor número de ellos posibles.

Finalmente, para que la robustez de las proyecciones climáticas sea lo mayor posible
es importante que se trabaje a escala local y diaria, que se utilicen los MCs más actuales
a fecha de la realización del estudio y que se hagan estudios de verificación y validación
completos que permitan evaluar las incertidumbres y calcular el error sistemático.

16
1.2 EL RETO DE LOS EVENTOS EXTREMOS

Existe un amplio consenso en la literatura científica en cuanto a que los efectos más
graves del calentamiento global estarán relacionados no solo con un cambio en el clima
medio, sino especialmente con un aumento de la frecuencia e intensidad de los
fenómenos extremos (como los huracanes, las sequías y las olas de calor, entre otros)
como se refleja en las conclusiones del IPCC y en estudios independientes (Cook et al.,
2013). Por su ubicación al sur de Europa, hay dos fenómenos extremos con fuerte
impacto en el territorio español: las olas de calor y los episodios de sequía.

Las olas de calor (definidas como un periodo de x días consecutivos con temperaturas
extremas) son una forma de tiempo extremo que aumentará en frecuencia, intensidad y
duración bajo la influencia de un clima cambiante durante los próximos años (IPCC
2013; Field et al., 2012). Los fenómenos de temperaturas extremas pueden afectar a
muchos aspectos de la vida humana como la mortalidad, la salud, el confort o la
agricultura y la hidrología (García-Herrera et al., 2005; Roldán et al., 2016) y de forma
considerable al viñedo, ya que una larga exposición de la vid a altas temperaturas puede
ocasionarle estrés térmico o suponer el achicharramiento de las hojas y/o el fruto.

El efecto que las altas temperaturas tienen sobre diversos sectores es un tema de gran
interés científico, aunque en la literatura es difícil encontrar estudios que presenten
proyecciones climáticas a escala local. La mayoría de estos estudios incluyen
información sobre temperaturas extremas como herramienta para explorar el impacto
de dichos extremos en un sector determinado (Tobias et al., 2014; Roldán et al. 2016;
Royé et al., 2021) pero no hacen una evaluación propia de las mismas.

A nivel europeo existen tanto estudios sobre temperaturas extremas actuales (Soares
et al., 2012; Seubert et al., 2014) como futuras. Estos últimos, o no están basados en
CMIP5 (Estrella and Menzel 2013; Hertig et al., 2010) o están en base a EURO-
CORDEX (Carvalho et al., 2021). Respecto a las olas de calor y eventos extremos, hay
más cantidad de estudios publicados pero la gran mayoría abarcan áreas completas
como toda Europa (Schoetter et al., 2015; Vautard et al., 2013), la región mediterránea
(Zittis et al., 2016) o todo un país (Kim et al., 2018; Planton et al., 2008) quedándose
Aragón fuera de casi todos ellos. En España los últimos estudios publicados sobre olas
de calor (Lorenzo et al., 2021; Pereira et al., 2021) y altas temperaturas (Pereira et al.,
2017; Lorenzo and Alvarez 2022) se basan principalmente en modelos provenientes de
EURO-CORDEX.

La sequía, por otra parte, es probablemente uno de los fenómenos climáticos extremos
con mayor impacto en la población mundial y que puede afectar a millones de personas
cada año en todo el planeta (Bryant 1993; Wilhite 2000). Los episodios de sequía tienen
graves efectos en la disponibilidad de agua y, por tanto, en actividades económicas
como el turismo y la agricultura (Lesk et al., 2016) o profundos impactos en la salud
humana (Stanke et al., 2013) y en los ecosistemas (Alary et al., 2014) que pueden
persistir en el tiempo (Dai 2011).

Según las conclusiones aportada por el IPCC (IPCC 2014), el análisis del régimen de
precipitaciones (Calbo, 2010; Lavaysse et al., 2012), las sequías (Burke and Brown,
2008; Lopez-Bustins et al, 2013) y las temperaturas extremas que aumentan

17
drásticamente la evapotranspiración (ET) (Rebetez et al., 2006) y disminuyen la
humedad del suelo (Sheffield and Wood 2008) sugieren que los episodios de sequía
podrían ser más graves en todo el mundo a lo largo del siglo XXI (Dai 2013).

Los escenarios de sequía publicados para España también son escasos: o bien
proceden de estudios realizados con anterioridad al IPCC5 y en áreas muy reducidas
(López-Bustins et al., 2013) o bien utilizan modelos del IPCC5 pero emplean información
de downscaling dinámico proveniente de EURO-CORDEX (Collados-Lara et al., 2018;
Marcos-García et al., 2017) u otros modelos dinámicos (Ojeda et al., 2021). En ambos
casos Aragón es una zona que no están bien representando debido a la escala espacial
de dichos estudios que no permite considerar con detalle las características climáticas
de menor escala que condicionan a Aragón.

En este contexto, la mayoría de las zonas europeas y la región mediterránea parecen

ser puntos calientes regionales destacados del cambio climático en los que se espera
un aumento de la ocurrencia de eventos extremos (Beniston et al., 2007; Skaugen et al.,
2004). En España, al igual que en el resto de Europa (Feyen and Dankers, 2009), se
han producido diferentes series de sequías importantes en las últimas décadas.
Además, la literatura parece indicar una tendencia al aumento de la escasez de agua
meteorológica en la Península Ibérica, ya sea por un aumento de la frecuencia de los
episodios de sequía o por un cambio en el régimen de precipitaciones (Fragoso et al.,
2018; Gallego et al., 2011; García-Barrón et al., 2011; Machado et al., 2011; Ojeda et
al., 2017; Vicente-Serrano et al., 2004). Esto hace necesario el desarrollo de escenarios
futuros de sequías a escala local que sean adecuados para evaluar los impactos del
cambio climático.

Por ello, la comunidad científica y las instituciones están dedicando muchos esfuerzos
a entender, identificar, documentar y monitorizar estos fenómenos de forma más
exhaustiva.

Uno de los grandes retos a los que se enfrenta la comunidad científica es la simulación
de eventos meteorológicos o climáticos extremos, y no solo porque son variables que
se mantienen fuera de los rangos promedios o porque se tiene menos información
histórica que permita evaluar su ocurrencia, sino porque en muchos casos existen varias
definiciones para un mismo evento, lo que dificulta su análisis y seguimiento. Un
fenómeno extremo es fácil de reconocer, pero difícil de definir.

Según el glosario del IPCC (IPCC 2013: Glosario), se define un evento meteorológico
o climático extremo como:

“fenómeno meteorológico raro en determinado lugar y época del año. Aunque las
definiciones de raro son diversas, la rareza normal de un fenómeno meteorológico
extremo sería igual o superior a los percentiles 10º o 90º de la estimación de la función
de densidad de probabilidad observada.”

distinguiendo entre fenómenos simples (aquellos eventos meteorológicos locales e

individuales que exceden un umbral crítico en una escala continuada) y fenómenos
complejos (tiempo severo asociado a un fenómeno climático particular como
consecuencia de una combinación crítica de variables).

18
Los extremos eventos no pueden describirse en función de una única variable ya que
vienen definidos por gran cantidad de atributos. Desgraciadamente, la naturaleza
multidimensional de estos fenómenos suele pasarse por alto, ya que se establecen
rankings basados en un único atributo, perdiéndose información muy valiosa para su
estudio.

Díaz and Murnane (2008), propusieron una taxonomía simple que permite analizar el
evento en función de su rareza, severidad y duración del evento (figura 10).

Figura 10. Clasificación de los eventos meteorológicos extremos en función del grado de severidad. Elaboración propia
basado en Díaz y Mumane, 2008.

Una de las preguntas más recurrentes dentro de la comunidad científica es ¿qué

procesos pueden propiciar la aparición de un fenómeno extremo? Stephenson 2008,
propone dos orígenes:

1) El principio evolutivo, es decir, el evento extremo no ocurre de forma estacionaria sino

que evolucionan de otros eventos menos extremos hasta su completo desarrollo.

2) El principio estacionario. Eventos extremos casi estacionarios, como las

temperaturas, presentan una ratio de cambio en su amplitud igual cero. Esta
característica implica que existe un balance entre la tendencia de los forzamientos y las
disipaciones.

Además, hay ciertos procesos que pueden dar lugar y/o favorecen la formación de
eventos extremos. Por ejemplo, el rápido crecimiento de un sistema debido a
inestabilidades, el desplazamiento de sistemas sinópticos a nuevas localizaciones o su
aparición en diferentes épocas del año. También la coincidencia temporal y espacial de
varios fenómenos que de forma individual no son considerados extremos pero que

19
combinados aumentan su intensidad o cambios en la persistencia u ocurrencia de los
fenómenos que pasan a ser crónicos, entre otros. La clave se encuentra en entender
cómo estos procesos han actuado en el pasado y como se comportarán en el futuro.
Según el IPCC6 (IPCC 2021a y b), la ocurrencia de eventos extremos no tiene
precedente en los registros observados y se espera que esta aumente en las próximas
décadas como consecuencia del calentamiento global, de manera que cada leve
aumento tiene un impacto importante. En los últimos años se ha reforzado la evidencia
de que ciertos fenómenos extremos se han recrudecido como consecuencia de la
influencia humana:

 En todas las regiones se ha observado un aumento en los episodios de calor,

especialmente las zonas continentales provocando un fuerte impacto en la salud
y el bienestar general. Estos episodios ocurren de forma más regular, ampliando
su periodo de aparición.
 Se ha observado un aumento en el número de días y noches cálidas y un
descenso en el número de días y noches frías.
 El cambio climático también juega un papel importante en la ocurrencia de
eventos de lluvia extremos, siendo cada vez más intensos y localizados espacial
y temporalmente. Se espera que estos eventos se intensifiquen en torno a un
7% por cada 1oC de calentamiento global.
 En el lado opuesto, ha aumentado el área afectada por sequías, cada vez más
severas y prolongadas.

Los datos observados son claros al respecto: el número de eventos extremos ha crecido
en casi todas las regiones del mundo, siendo muy pocas aquellas en las que ha
disminuido. El siguiente paso, para poder predecir los fenómenos extremos de la mejor
forma posible, es saber cuántos de esos fenómenos son debido a la variabilidad natural
del sistema climático y qué porcentaje consecuencia de la actividad humana.

Para responder esta pregunta, se aplican estudios de “atribución” que permiten evaluar
si la intensidad y/o frecuencia de un evento extremo ha cambiado o cambiará como
consecuencia del cambio climático. En este tipo de estudios hay que lidiar con ciertas
limitaciones que dificultan los trabajos de atribución. En primer lugar, la calidad de los
datos observados, que normalmente son muy cortos para recoger fenómenos
meteorológicos con periodos de ocurrencia de más de 30/50 años. En segundo lugar, la
dificultad de los modelos para simular un evento extremo concreto y finalmente, cómo
de bien los modelos conocen los procesos físicos que rigen dicho evento y de qué
manera se verán alterados por el cambio climático. En aquellos eventos donde se vean
bien representadas todas las condiciones, el grado de confiabilidad será mayor y
viceversa.

La importancia de poder evaluar el impacto que el cambio climático tiene en la

ocurrencia y características de estos fenómenos es una pieza clave para poder
simularlos con el mayor grado de confiabilidad posible. Es tanto el interés en avanzar
en este aspecto que desde el 2011 se publica de forma anual el informe “Explicando los
eventos extremos desde una perspectiva climática” publicado como suplemento
especial dentro del Boletín de la Sociedad Meteorológica Americana (BAMS de sus
siglas en inglés), donde se recogen decenas de estudios de atribución. Por desgracia a

20
día de hoy sabemos que el cambio climático es una causa más, añadida a las ya
existentes, de aparición de eventos extremos pero no podemos saber si es “la causa”.

Conocer el comportamiento futuro de los fenómenos extremos de mayor impacto en la

región considerada en este estudio como las olas de calor y de frío y las sequías con
gran efecto en el sector vinícola es de gran importancia para minimizar los impactos
negativos y evaluar el riesgo asociado a cada fenómeno. El riesgo asociado a un
episodio extremo puede entenderse como la combinación entre las características
naturales del propio fenómeno (duración, intensidad, extensión), la vulnerabilidad de la
sociedad frente a dicho fenómeno y la capacidad de prevención frente a dicho fenómeno
(figura 11).

Figura 11. Riesgo asociado a un evento meteorológico extremo en base a los elementos que lo condicionan.
Elaboración propia.

1.2.1. Episodios de olas de calor y frío

Existen muchas y muy variadas formas de definir una ola de calor. Por ejemplo:

 Según la Agencia Estatal de Meteorología (AEMET) una ola de calor es “un

episodio de al menos tres días consecutivos, en que como mínimo el 10% de las
estaciones consideradas registran máximas por encima del percentil del 95% de
su serie de temperaturas máximas diarias de los meses de julio y agosto del
periodo 1971-2000”.
 La Organización Meteorológica Mundial (WMO, de sus siglas en inglés) lo define
como “un evento extremo condicionado por un marcado calentamiento del aire
o por una invasión de aire cálido sobre una extensa región y con una duración
desde pocos días a varias semanas”.
 El IPCC lo define como “un periodo de calor anormal e incómodo”.
 EURO-CORDEX considera que un episodio de ola de calor es “un periodo de al
menos tres días consecutivos en los que la temperatura máxima diaria supere el
percentil 99 de las temperaturas máximas diarias de los meses de mayo a
septiembre para el periodo de control de 1971 a 2000”.

21
Además de las anteriormente citadas, una ola de calor se puede definir en función de
los datos que se manejan o del sector en el que se vayan a analizar. Por ejemplo,
algunas definiciones se han utilizado para identificar olas de calor en una serie temporal
de datos de temperatura (Smith et al., 2013). En otras ocasiones la elección de la
definición de ola de calor puede influir tanto en las tendencias proyectadas (Smith et al.,
2013) como en las estimaciones de los riesgos durante los eventos (Anderson and Bell
2009; Chen et al., 2016; Kent et al., 2014).

Según las recomendaciones de la WMO (2010), una definición práctica y cualitativa de

una ola de calor debe considerar la existencia de un tiempo caluroso marcado y no
habitual en una región durante al menos dos días consecutivos en el periodo caluroso
del año, basándose en las condiciones climáticas locales, con condiciones térmicas
registradas por encima de determinados umbrales.

Por otra parte, y a grandes rasgos, una ola de frío puede definirse como un
acontecimiento meteorológico caracterizado por un fuerte descenso de la temperatura
del aire cerca de la superficie que conduce a valores extremadamente bajos. Otras
definiciones manejadas por organismos nacionales e internacionales son:

 AEMET define una ola de frío como “un episodio de al menos tres días
consecutivos, en que como mínimo el 10% de las estaciones consideradas
registran mínimas por debajo del percentil del 5% de su serie de temperaturas
mínimas diarias de los meses de enero y febrero del periodo 1971-2000”
 La WMO define un episodio de ola de frío como “un enfriamiento marcado del
aire o una invasión de aire muy frío en una zona amplia”.
 Según el IPCC, una ola de frío es un evento que suele causar problemas y
graves impactos en la población, especialmente en las altitudes del norte.
 EURO-CORDEX define un episodio de ola de frío como “un periodo de al menos
tres días consecutivos en los que la temperatura mínima diaria no supere el
percentil 1 de las temperaturas mínimas diarias de los meses de octubre a abril
para el periodo de control de 1971 a 2000”.

Las directrices marcadas por la WMO recomiendan considerar un episodio de ola de frío
como "un tiempo frío marcado e inusual que se caracteriza por un descenso brusco y
significativo de las temperaturas del aire cerca de la superficie (máximas, mínimas y
medias diarias) en una zona amplia y que persiste por debajo de ciertos umbrales
durante al menos dos días consecutivos durante la estación fría".

Según recomienda la WMO (2015) en su guía “The Guidelines on the definition and
monitoring of extreme weather and Climate Events”, los episodios de ola de calor y frío
han de evaluarse atendiendo a las siguientes características:

 Duración media: número medio de días que duró la ola de calor.

 Intensidad media: valor medio de la temperatura durante la ola de calor.
 Intensidad máxima: valor máximo que alcanzó la temperatura durante la ola de
calor.
 Extensión espacial: área ocupada por la ola de calor.

Los episodios de ola de calor y ola de frío se calculan en base a las características
climáticas de cada punto, por lo que el umbral (el valor de temperatura) que determina

22
si un día es demasiado cálido (o frío) no es fijo y no debe ser tomado como referencia
fuera del punto para el que ha sido determinado.

1.2.2. Episodios de sequía

Las sequías son otro de los eventos climáticos más extremos y persistentes y uno de
los episodios menos entendidos ya que su propia naturaleza intrínseca lo convierte en
un fenómeno de gran complejidad (Hagman 1984). Además, es uno de los fenómenos
naturales que más afectan a la sociedad ya que su impacto no siempre es cuantificable
en el mismo momento en el que se detecta el fenómeno como sí ocurre con otros
fenómenos meteorológicos tales como los huracanes, los terremotos o las inundaciones
(Bryant 1991; Wilhite 1992). Por este motivo, es uno de los temas más recurrentes en
la comunidad científica (Scheffield et al., 2012; Dai 2013; Spinoni et al., 2015) y desde
las instituciones se están haciendo grandes esfuerzos por construir una base de datos
de sequía que permitan ayudar a cuantificar el fenómeno como el European Drought
Observatory (www.ge.uio.no/de) donde se recopila información del Combined Drought
Index, CDI (Sepulcre-Canto et al., 2012) y del European Drought Impact Inventory, EDII
(Stahl et al., 2012) o el National Drought Mitigation Center que realiza un inventario
semanal de las condiciones de sequía en US a través de varios índices de sequía. En
España se ha generado una base de datos del SPEI (Vicente-Serrano et al., 2017).

Las características que hacen que esté fenómeno sea más complejo de cuantificar son:

1) Los fenómenos de sequía suelen acumularse en el tiempo, desde pocos meses a

años, y sus efectos siguen repercutiendo en algunos sectores incluso años después de
que haya terminado el episodio.

2) No hay una definición estándar y global que determine un fenómeno de sequía, sino
que hay cientos de definiciones generalmente relacionadas con las características
climáticas y socioeconómicos de la región para la que se define (Wilhite and Glantz
1985).

3) A diferencia de otros fenómenos como los tornados o los terremotos, los episodios
de sequía no causan pérdidas de vida ni daños estructurales sino que sus efectos son
más silenciosos, por lo que muchos servicios de protección civil colocan a este
fenómeno en el ranking de los más peligrosos (Bryant 1991). Las consecuencias de la
sequía también dependen de su uso socioeconómico por lo que tienen una componente
social y otra cultural (Hayes 2011). Como ejemplo, se produce escasez de agua debido
al aumento de la población, a los usos agrícolas, usos energéticos o al consumo en
sectores industriales (Bates et al., 2008).

4) Debido a que los efectos asociados a un episodio de sequía pueden perdurar, e

incluso, intensificarse con el paso de los meses, en la mayoría de las ocasiones no se
cuantifican adecuadamente debido a la falta de registros históricos (Hayes 2011).
Además, es complicado por los múltiples mecanismos que lo causan y por su
operacionabilidad a diferentes escalas espaciales y temporales convirtiéndola en uno
de los fenómenos más difíciles de monitorizar y manejar (Wilhite 1992).

23
Atendiendo exclusivamente al punto de vista meteorológico, un episodio de sequía es
(Wilhite 2000):

“la interrupción durante un tiempo prolongado de los patrones normales de la

circulación global”

siendo contadas las ocasiones en las que este fenómeno se analiza exclusivamente
desde esta perspectiva, sino que en general se amplía el concepto considerando el
impacto que este fenómeno tiene en otros sectores. Es por esto, que se habla de cuatro
tipos de sequía cuyos efectos van encadenándose: sequía meteorológica, agrícola,
hidrológica y socioeconómica (figura 12).

Figura 12. Tipos de sequía según el sector en el que impacten y las componentes que se vean afectadas. Elaboración
propia.

Además del sector al que más afectan, estos tipos de sequía difieren entre sí por la
duración, intensidad y extensión espacial del fenómeno (Hayes et al., 2010). Además
pueden ser analizadas desde dos puntos de vista: conceptual u operacional (Wilhite and
Glantz 1985). Si se evalúa la sequía conceptualmente, solo se tiene en cuenta el
fenómeno físico que supone, es decir, alteraciones en el régimen pluviométrico, pero si
se evalúa operacionalmente, se intenta definir características de dicho fenómeno como
el comienzo y el fin del episodio, la recurrencia o la intensidad. Todas ellas
informaciones muy útiles para intentar predecir estos fenómenos.

Como la sequía meteorológica por definición es una alteración de los valores normales
de precipitación, es un fenómeno que se da en cualquier tipo de clima,
independientemente de que presente un régimen pluviométrico elevado o escaso. No
ha de confundirse con la aridez, fenómeno de escasez de precipitaciones permanente
y que es propio de una región y no una alteración temporal. Aunque la sequía
meteorológica hace referencia solo a las precipitaciones, la presencia de ciertos factores
como las rachas de viento fuerte (McVivar et al., 2012a), el contenido de agua (Willet et
al., 2014), la nubosidad y la duración solar (Wild et al., 2013) o las altas temperaturas
(Cook et al., 2014; Livneh and Hoerling 2016, Huo et al., 2017; Hartman et al., 2013)
pueden contribuir a su intensificación. Por otro lado, el impacto de la sequía no va a ser
el mismo y dependerá del estado en el que se encuentren los recursos hídricos y las
necesidades socioeconómicas en el momento en el que se produce el fenómeno.

24
Sequías de la misma magnitud pueden tener efectos muy diferentes en función de la
época del año y el lugar en el que ocurran.

Son muchos los sectores interesados en evaluar la presencia de episodios de sequía y

sus impactos, como por ejemplo, efecto en desastres naturales (Carroll et al., 2009) o
ecológicos (Lewis et al., 2011; Choat et al., 2012), influencia en incendios forestales
(Gudmundson et al., 2014; Westerling et al., 2006), en el abastecimiento de agua y la
gestión de recursos hídricos (Pedro-Monzonis et al., 2015; Hoerling et al., 2014; Seager
et al., 2015), en la agricultura (Tsakiris et al., 2010; Lesk et al., 2016) y la ganadería
(Alary et al., 2014; Clark et al., 2016), sin olvidar el impacto social y económico (García-
Herrera et al., 2010; Cheesman 2016; Stanke et al., 2013).

No existe una magnitud física ni un instrumento que permita medir la sequía, por lo que
la manera en la que se analizan los episodios de sequía es a través de indicadores o
índices de sequía, definidos como (Hayes et al., 2000, Hayes 2006):

“un indicador directo basado en información climática que resume en un único valor las
características principales del fenómeno facilitando su uso frente a valores puros de
las variables meteorológicas. Un índice es útil si proporciona una evaluación clara,
simple y cuantitativa de las características principales de la sequía”

Los índices de sequía suelen ser ajustados a la zona de estudio y a las necesidades de
evaluación del fenómeno por lo que existe una gran variedad de ellos. En las últimas
décadas se ha producido un gran avance en el desarrollo de estos índices. Se definen
índices basados exclusivamente en la precipitación como el SPI (McKee et al., 1993), el
RAI (Van-Rooy 1965), el DSI (Bryant et al 1992), el NRI (Gommes and Petrassi 1994) o
el DFI (Gonzalez and Valdes 2006). Otros combinan la precipitación con el efecto de la
temperatura como el SPEI (Vicente-Serrano et al., 2010a) y el RDI (Tsakaris and
Vangelis 2005; Thomas et al 2015) o el aridity index (Subrahmanyam and Subramaniam,
1964). Finalmente, otros índices aumentan su complejidad incluyendo variables como
como la cobertura nubosa, el stream flow (SWSI, Nagarajan 2003), la humedad del suelo
(SPDI, Palmer 1965) o la vegetación (NDVI, Nagarajan 2003).

Son varios los estudios que han intentado recopilar todos ellos (Zagar 2011; Niemeyer
2008; Wilhite and Glatz 1985) con el fin de poner en común todo lo existen referente a
indicadores de sequía, pero la selección de un indicador u otro no es tan sencilla ya que
hay que tener en cuenta la disponibilidad de datos para su cálculo, el uso que se le va
a dar a la información y si se va a aplicar a una región u otra (necesidad de índices
normalizados). Con el fin de acotar el uso de indicadores y dar una ruta de trabajo a los
interesados, son múltiples las guías que organismos, como el IPCC o la WMO, generan
con directrices para el cálculo de sequías. En una de ellas, “The Lincoln declaration on
Drought Indices” (Hayes et al., 2010) se determinó que el Índice de Precipitación
Estandarizado (SPI) es, probablemente, el único índice valido para cualquier región del
mundo y escala temporal.

En el contexto de calentamiento global actual, es evidente que el aumento de las

temperaturas ha llevado a una aceleración del proceso hidrológico. Por un lado,
aumenta la energía disponible mediante la evapotranspiración, por otro, aumenta la
capacidad de almacenamiento calorífico por parte de la atmósfera a razón teórica de un

25
7% por cada incremento de 1 ºC (Trenberth 2011) y por último, la energía disponible en
la superficie aumenta más despacio. La conjunción de todos los aspectos provoca
alteraciones en la humedad disponible y por ende en la precipitación global, lo que se
traduce en periodos secos, especialmente en verano (Vicente-Serrano et al., 2010b).

El SPI, es un índice que se calcula exclusivamente en base a la precipitación.

Atendiendo a lo anterior, Vicente-Serrano et al. (2010a), definieron el Índice
estandarizado de precipitación y evapotranspiración (SPEI) que supone una
modificación del SPI mediante la introducción de la evapotranspiración.

Ambos índices (SPI y SPEI) son índices estandarizados (permiten comparar regiones
entre sí fácilmente), permiten evaluar, mediante su comparativa, la influencia de la
temperatura en los episodios de sequía y han sido ampliamente aplicados en la zona de
estudio. Por todo ello, son los índices más adecuados para alcanzar los objetivos
planteados en este estudio. Además, pueden representarse a diferentes escalas en
función del alcance del estudio (figura 13).

Figura 13. Escala de intensidades usada para evaluar el SPI/SPEI así como implicaciones de la sequía según la escala
temporal considerada. Elaboración propia.

El Índice de precipitación estandarizado (SPI) fue desarrollado por Mckee et al.

(1993) y se basa en dos supuestos: que la variabilidad de la precipitación es mayor que
la de la temperatura y la evapotranspiración potencial (ETP), y que el resto de las
variables son estacionarias en el tiempo. El valor del SPI se define como:

“un valor numérico que representa el número de desviaciones estándar de la

precipitación, en el periodo de acumulación en cuestión, respecto a la media, una vez
transformada la distribución original de la precipitación en una distribución normal (es
decir, media de cero y desviación estándar de 1)”

Los valores del SPI pueden interpretarse como el número de desviaciones típicas en
que la anomalía observada se desvía de la media a largo plazo.

26
El Índice de evapotranspiración y de precipitación estandarizado (SPEI),
desarrollado por Vicente-Serrano et al., 2010a y revisado por Begueria et al., 2014, es
una variante del SPI. Tiene mayor potencial como índice de sequía ya que considera el
balance climático (a través de la diferencia entre la precipitación mensual y la ETP). Se
define el valor del SPEI como:

“un valor numérico que representa el número de desviaciones estándar del balance
climático (precipitación menos ETP), en el periodo de acumulación en cuestión,
respecto a la media, una vez transformada la distribución original de la precipitación en
una distribución normal (es decir, media de cero y desviación estándar de 1)”

Los valores del SPEI pueden interpretarse de la misma manera que los del SPI (número
de desviaciones estándar en que la anomalía observada se desvía de la media a largo
plazo).

Para el cálculo del SPI se utiliza la distribución Gamma con el fin de ajustarse a la serie
de precipitación original (WMO, 2012) y en el caso del SPEI se utiliza la distribución log-
logística (Vicente-Serrano et al., 2015; Vicente-Serrano and Beguería, 2016). Los
parámetros de estas distribuciones se obtienen por el método de momentos ponderados
probabilísticos insesgados (Vicente-Serrano and Beguería, 2016). La escala de valores
de SPI y SPEI utilizada en el estudio puede verse en la figura 13.

27
1.3 EL IMPACTO DEL CAMBIO CLIMÁTICO SOBRE EL CULTIVO DE LA
VID: INDICADORES BIOCLIMÁTICOS

El sector agrícola en general y el vitícola en particular, es uno de los más vulnerables al

cambio climático. El impacto del cambio climático no afecta de la misma manera a todas
las regiones ni cultivos por igual, por lo que es esencial que se evalué a escala local.
Para alcanzar este objetivo es necesario disponer de información climática proyectada
a escala local (por ejemplo, temperatura y precipitación), evaluar el impacto que dichas
proyecciones suponen para el sector vitícola (mediante indicadores bioclimáticos) e
identificar los posibles riesgos a los que ha de hacer frente el sector en las próximas
décadas.

A día de hoy, España es una de las regiones más apropiadas para cultivar vinos de alta
calidad por sus características climáticas y edafológicas (Van Leeuwen et al., 2004;
Sotés et al., 2012). Aunque en general el viñedo europeo se cultiva entre 100 y 300 m
de altura, las características climáticas del territorio ibérico-balear hacen posible que se
cultiven algunas variedades hasta los 1000m, ampliándose considerable el rango de
variedades de uvas y por ende de tipos y calidades de vinos que se pueden obtener.
Desgraciadamente, estas condiciones pueden verse modificadas en el futuro (Deque et
al., 2012; Lorenzo et al., 2013).

A grandes rasgos, los vinos españoles pueden agruparse en continentales (tipo Rioja o
Valdepeñas), atlánticos húmedos (tipo Alabariño o Chacolí) y mediterráneos (Penedés).
Estos factores propios y únicos de cada una de las distintas regiones donde se cultivan
vinos españoles, dan lugar a las Denominaciones de Origen (DOs). Las DOs españolas
constituyen el sistema utilizado en España para el reconocimiento de una calidad
diferenciada, consecuencia de unas características propias y diferenciables debidas al
medio geográfico en el que se producen las materias primas.

La importancia que el sector vitícola tiene para un país como España se reflejan en las
cifras (FEV, 2021): es el primer viñedo del mundo con 949.565 ha de vid, lo que supone
un 13% del total mundial; es el tercer productor mundial con una producción media anual
de vino y mosto de entre 40 y 42 millones de hectolitros; es el primer exportador mundial
en volumen (21 millones de hectolitros en 2019) y el tercero en valor (2.700 millones de
euros exportados en 2019. Tres de cada cinco botellas comercializadas en el mundo
proceden de Europa y España es responsable del 25% de la producción de vino
europeo. Además, es líder en viñedo ecológico, con 121.000 ha responsables de una
producción de más de 400.000 toneladas de uvas de vinificación (FEV, 2021).

En España hay cerca de 4300 bodegas (de las cuales más de 3000 son exportadoras)
que venden vinos españoles en más de 198 países. Estas cifras demuestran que el
sector vitícola aporta un gran valor a la economía, generando y manteniendo más de
427.000 empleos (FEV, 2021). Esto es especialmente relevante en zonas rurales pues
permite luchar contra el despoblamiento rural y fomentar el turismo gastronómico,
principalmente a través de los museos y bodegas que conforman la Red de Rutas del
Vino de España.

Organismos nacionales como la Federación Española del Vino, el Observatorio Español

del Vino o la Organización Interprofesional del Mercado del Vino, junto con
internacionales como la Organización Internacional del Vino están poniendo mucho

28
esfuerzo y recursos en elaborar un plan de acción frente al cambio climático, intentando
identificar las necesidades del sector con el fin de establecer ciertas directrices,
recomendaciones y buenas prácticas que velen por el futuro del sector vitícola español.
Estos esfuerzos se ven reflejados en los más 170 millones al año que se invierte en
proyectos de I+D+i (FEV, 2021).

1.3.1. Clima y vid

La vid es una planta leñosa perteneciente a las Vitáceas que comprende 19 géneros,
de los cuáles únicamente el género Vitis es cultivado. Este género se divide a su vez en
Euvitis y Muscadina. La mayoría de las vides cultivadas pertenecen al género Euvitis y
se cultivan, principalmente en Asia Oriental, América y Europa. En las dos primeras
regiones se dan más de 20 especies, mientras que en Europa se cultiva,
exclusivamente, la especie Vitis vinífera.

Debido al considerable aumento en el número de hectáreas de vides cultivadas en

regiones como Australia, Estados Unidos, Chile, Sudáfrica y Nueva Zelanda, la industria
vitícola europea ha visto la necesidad de ir un paso más allá para hacer frente a sus
crecientes “competidores”, probando nuevas variedades y mejorando las técnicas de
cultivo.

Tradicionalmente las zonas más óptimas para el cultivo de la vid se han ubicado entre
los 30-500N y los 20-50oS coincidiendo con aquellas regiones que presentan
temperaturas promedio anuales entre los 10 y los 20oC.

Figura 14. Localización de las principales regiones vitícolas del mundo (en color morado). En líneas generales, se
sitúan entre los 30-50 ºN y los 20-50 ºS. Estos límites se corresponden con las isotermas 10 y 20 ºC en ambos
hemisferios. Fuente: Sallis et al, 2009:3.

Estas regiones presentan un clima base (unas características climáticas promedio) que
las convierten en idóneas para su cultivo con una producción óptima de la vid dando
lugar a vinos equilibrados y de calidad. Si el clima base es demasiado frío, se obtienen
vinos desequilibrados como consecuencia de la escasa maduración de la uva. Por el
contrario, si el clima es demasiado cálido, también se obtienen vinos desequilibrados,
pero en este caso por sobre maduración de la uva (figura 15). El clima base determina

29
qué variedades de vid encajan en cada región y el estilo de vino que podrá producir, y
la variabilidad climática es la encargada de diferenciar unas añadas de otras dentro de
un mismo viñedo.

La variabilidad climática es una característica intrínseca del clima y desde siempre los
viticultores han ido adaptándose de forma gradual a las condiciones medioambientales
locales y regionales y a sus variaciones (Jones et al., 2012). Por tanto, el clima promedio
y la variabilidad climática son los factores medioambientales que más influyen en la
calidad y producción del vino (Spellman 1999; David 2000; Santos et al., 2011).

Figura 15. Calidad de la cosecha en función de las condiciones climáticas (basada en PricewaterhouseCoopers, 2009)

La vid es muy sensible al clima (Kenny 1993; Winkler 1974; Gladstones 1992; Jones
2006; Meier 2007; White 2006) y a las condiciones atmosféricas en un gran rango de
escalas temporales (Santos et al., 2019), desde impactos a corto plazo (heladas, rachas
de viento y granizo), a medio plazo (sequías e inundaciones) y a largo plazo (tendencias
climáticas). Por otro lado, las variables climáticas influyen en aspectos como el
crecimiento y la productividad de la vid, rigen la fotosíntesis, son el aporte de agua o
determinan la coloración, entre otros (Carbonneau 2003). Además, posibles
alteraciones en los requerimientos climáticos (heliotérmicos e hídricos) tienen un fuerte
impacto en las características organolépticas finales de la uva (acidez, azúcar, color,
etc) y por ende, en las características finales del vino. Por consiguiente, cambios en
patrones meteorológicos y/o climáticos son la consecuencia de los efectos del cambio
climático que pueden derivar en impactos para el sector vitícola (Schultz 2000; Jones et
al., 2005a).

Está demostrado que el cambio climático afecta al viñedo (Kenny and Harrinson 1992;
Jones 2006) y los resultados mostrados por las proyecciones climáticas parecen apuntar
a una tendencia hacía impactos cada vez más fuertes (Meehl 2007) y un desplazamiento
de las zonas óptimas para el cultivo de la vid hacía los Polos de unos 20º hacía el 2050
(Kenny and Harrinson 1993; Tate 2001). Se espera que los efectos del cambio climático
afecten al viñedo, no solo de forma global sino dentro de cada añada, modificando los
estados fenológicos y la definición de “gestión tradicional” del viñedo.

Son varios los estudios que han analizado la relación clima-viñedo en territorio español
durante las últimas décadas (Santos et al., 2012; Malheiro et al 2010; Ramos et al.,

30
2008; Lorenzo et al., 2013; Fraga et al., 2014, Ramos et al., 2008; Moral et al., 2015) o
han realizado una clasificación climática para el pasado (Blanco-Ward 2007; Moral et
al., 2015). En los últimos años, además, se ha añadido el esfuerzo de utilizar
proyecciones climáticas para estudiar las relaciones clima futuro-viñedo (Lorenzo et al.,
2013, 2015) y proyectar zonificaciones climáticas mediante distintos indicadores
(Blanco-Ward 2007; Lorenzo et al., 2012; Fraga et al., 2012, 2013; Resco et al., 2016;
Santos et al., 2019).

1.3.2. La vid: ¿cómo le influyen las variables meteorológicas?

La vid es una planta perenne caracterizada por presentar ciclos de vida anuales
interdependientes, es decir, las condiciones de un ciclo condicionan las características
que tendrá el siguiente. El ciclo biológico de la vid engloba el ciclo vegetativo y el ciclo
reproductor, de manera que ambos son simultáneos en el tiempo (figura 16). Dentro de
las distintas etapas, el periodo de maduración es el más importante, ya que es el
momento en el que se fijan las características organolépticas finales de la uva (acidez,
color y aroma) y que definen su calidad, alcanzándose los niveles óptimos de azúcar y
alcohol.

Figura 16. Etapas fenológicas de la vid. Elaboración propia.

Las distintas variables meteorológicas van a jugar un papel importante en el correcto

desarrollo de la vid, de manera que es necesario que se den las condiciones térmicas e
hídricas idóneas en cada una de ellas (Magalhaes 2008; Makra et al., 2009). Mientras
que la temperatura afecta principalmente a la calidad de la uva (Jones and Davis 2000;
Santos et al., 2011) y al equilibrio del vino mediante una correcta maduración, la
radiación y la precipitación tienen efectos en la cantidad de la cosecha ya que son
factores a tener en cuenta en la gestión del viñedo (Lopes et al., 2008).

Aunque el calor y la insolación son dos variables que aportan múltiples beneficios en el
correcto desarrollo de la vid, un exceso de ambos puede convertirlos en perjudiciales.
En el aporte justo de calor, agua e insolación está la clave para el éxito productivo y

31
cualitativo de la vid, de manera que cada variable meteorológica está ligada a la vid de
diferente manera.

En términos generales, las condiciones más favorables para la vid son aquellas en las
que la temperatura óptima en el periodo de crecimiento se sitúa entre los 12-22 ºC
(Jones et al., 2006) y la temperatura media en el periodo vegetativo entre los 20 y 30
ºC (Gouveia et al., 2011) asegurándose una adecuada acumulación de calor en el
periodo de maduración (White et al., 2006; Holland and Smith, 2014). Respecto a la
precipitación, se requieren entre 350 y 600 mm anuales para obtener vinos de calidad
(Hidalgo et al., 2012). Además, este cultivo es muy exigente en cuanto a las horas
solares, necesitando entre 10 (en octubre) y 15 horas (junio y julio). En cuanto a la
presencia de enfermedades, la humedad relativa máxima juega un papel esencial en los
meses de mayo y junio (favoreciendo enfermedades criptogámicas como el mildiu, el
oídium y la botritis), lo ideal son valores entre el 60 y el 70%.

A continuación, se analiza el papel que juega cada variable meteorológica en el

desarrollo de la vid.

 Temperatura

Como se ha comentado, la temperatura es la variable que más impacto tiene en el ciclo

de la vid. En cada una de las etapas fenológicas es necesario que se den unos valores
térmicos concretos para que la baya se desarrolle correctamente ya que tanto la
composición metabólica como la fisiológica de la uva están fuertemente relacionadas
con ella (Jones et al., 2012).

En términos generales, la actividad de la planta y por tanto su desarrollo comienza

cuando las temperaturas alcanzan los 10oC (normalmente en primavera). Una vez
alcanzado o superado dicho umbral (denominado cero vegetativo) la planta se mantiene
en desarrollo hasta que las temperaturas vuelven a situarse por debajo de dicho umbral
(normalmente en otoño). Este periodo de tiempo se denomina periodo activo de la vid.
La parte radicular de la vid tiene un periodo activo algo más largo que la parte aérea,
esto se debe a que el suelo alcanza el cero vegetativo antes que el aire y el descenso
de temperaturas es más lento llegando de nuevo a ese umbral con posterioridad. Estos
valores varían de una variedad a otra e incluso hay algunas variedades que pueden dar
señales de desarrollo una vez alcanzados los 0 ºC pero en términos generales se
acepta el umbral de 10 ºC para el cero vegetativo y el periodo de abril a octubre como
el periodo activo.

La temperatura juega un papel especialmente importante en las siguientes etapas

fenológicas:

1) Salida de dormancia de yemas latentes. Tras varios días consecutivos con

temperatura mínima inferior a un umbral establecido en función de la variedad, las
yemas despiertan de su reposo vegetativo. Ocurre en invierno.

2) Desborre. Las yemas brotan a partir de un umbral, normalmente 10 ºC.

3) Floración. Esta etapa es altamente dependiente de las temperaturas.

4) Maduración. Este es el periodo más importante y determina la calidad de la cosecha.

Los procesos que ocurren en esta etapa se producen a nivel de baya y es por eso que

32
cada variedad necesita que durante la maduración que la temperatura se mantenga
dentro de unos valores concretos (figura 17) (Jones et al., 2006).

Figura 17. Umbrales de temperatura durante la época de crecimiento para diferentes variedades de uva. Fuente: Jones
2006.

Temperaturas muy elevadas en este periodo (en función de los requerimientos de cada
variedad) ponen en riesgo la calidad de la uva. Los impactos en la baya por altas
temperaturas varían si éstas se producen en combinación con otros factores (como la
humedad relativa) y en función del momento del día en que se consideren (figura 18).

Figura 18. Efectos asociadas a las altas temperaturas registradas durante el periodo de maduración. Elaboración
propia.

33
En general, temperaturas altas son sinónimo de calidad del vino. Pero esta afirmación
no es 100% cierta ya que las temperaturas extremas (tanto altas como bajas) también
pueden afectar al viñedo de forma negativa. Además, cada variedad tolera de forma
diferente las altas temperaturas (Moutinho-Pereira et al., 2007). El mayor problema
relacionado con altas temperaturas es la posible desecación de las partes aéreas de la
planta e incluso su muerte. Mientras que temperaturas muy bajas pueden llegar a helar
los órganos herbáceos o limitar el desarrollo de la planta (figura 19).

Figura 19. Posibles complicaciones que pueden presentarse en la vid bajo condiciones extremas de temperatura.
Elaboración propia.

Las diferencias genéticas, morfológicas y fisiológicas entre variedades propician que su

adaptación al cambio climático sea diferente en cada caso. Variedades con una baja
demanda termal (pocas necesidades de acumulación de calor con un rango pequeño
de oscilación) tendrán más complicado adaptarse, las variedades que necesiten poca
demanda termal pero tengan un rango alto de oscilación tendrán un margen medio de
adaptación, mientras que aquellas variedades con una alta demanda termal tendrán
más fácil adaptarse al cambio climático. En general, las variedades de uva blancas
tienen una demanda termal más baja que las tintas.

 Pluviometría

El 80% de la vid es agua, por lo que es esencial para su constitución ya que la usa para
el transporte de nutrientes, para transpirar y para refrigerarse. Por lo tanto, mantener el
aporte hídrico es esencial para que el viñedo viva. Ha de tenerse en cuenta que la vid
no requiere el mismo aporte hídrico en todas las etapas fenológicas, por lo que es igual
o más importante conocer la evolución pluviométrica dentro del año que la precipitación
anual total. Tanto un exceso como un defecto de agua en momentos clave pueden ser
perjudicial para la vid (figura 20).

34
 Figura 20. Consecuencias de déficit o exceso de agua en función de la etapa fenológica en la que se produzca.
Elaboración propia.

 Humedad relativa

La humedad relativa no es una variable que influya en la vid tanto como la precipitación
o como la temperatura. Pero valores de humedad muy altos o bajos pueden resultar
perjudiciales para la planta. Si la humedad relativa es alta (>80%) puede favorecer la
proliferación de ciertas enfermedades mientras que situaciones de estrés hídrico
acompañadas de humedades bajas (<40%) resultan críticas para la vid. Humedades
relativas de entre el 60-70% son las óptimas, ya que favorecen la actividad fotosintética.

 Insolación

La vid necesita de un cierto número de horas de sol para su crecimiento, siendo muy
exigente durante las etapas de floración y maduración. En términos generales, la vid
necesita entre 10 y 14 horas de sol para su óptimo desarrollo.

 Fenómenos meteorológicos extremos adversos

Aunque la vid es una planta que se adapta muy fácilmente al estrés medioambiental, los
fenómenos extremos la pueden llegar a afectar irreversiblemente (Hidalgo 2012;
Easterling 2000; Jones et al., 2005; Menzel et al., 2011).

Dentro de los fenómenos extremos hay algunos que resultan especialmente

perjudiciales para la vid como los episodios de olas de calor, los episodios de sequía y
los episodios de granizo y heladas. Las consecuencias de su aparición van a depender
del momento del ciclo biológico de la vid en que aparezcan (figura 21).

35
Figura 21. Fenómenos extremos con mayor impacto en el viñedo y las posibles consecuencias asociadas
a su aparición. Elaboración propia.

Además de las variables meteorológicas anteriormente expuestas y que varían día a

día, hay otros factores que van a influir en el buen desarrollo del viñedo: los factores
geográficos (ubicación, continentalidad, orientación etc.); los factores climáticos (las
características promedio de una región contribuyen a la idoneidad climática para el
cultivo de la vid) y los factores socio-económicos (técnicas de manejo de la finca,
elección adecuada del cultivo etc.) (figura 22).

Figura 22. Combinación de factores que condicionan el cultivo y la producción vitícola. Elaboración propia.

36
1.3.3. Cambio Climático y viñedos

Durante las últimas décadas los cambios en las condiciones climáticas como
consecuencia del cambio climático (aumento de las temperaturas, alteraciones en el
régimen pluviométrico, alteraciones en la ETP o aumento en las concentraciones de
CO2), han afectado de forma directa a los viñedos a lo largo de todo el mundo. Son
múltiples los estudios que recopilan los cambios que los productores vitícolas han
experimentado en sus fincas (Battaglini et al., 2009; Alonso and O’Neill, 2011) y que
tienen relación directa con los requerimientos heliotérmicos e hídricos que la vid necesita
para un crecimiento óptimo, observándose cambios en cualquier tipo de clima.

Entre los principales cambios que se han reportado, a nivel de finca, destacan cambios
en la viabilidad de una zona para cultivar ciertos tipos de variedades, cambios en la
distribución geográfica (Schultz, 2000), cambios en el régimen hídrico y por tanto en las
necesidades de riego (Ruml et al., 2012), variaciones en las secuencias de plagas y
enfermedades (Gouevia et al., 2011), alteraciones de los parámetros edafológicos,
aumento de las temperaturas durante la época de crecimiento (Fraga et al., 2012;
Santos et al., 2012; Duchene and Schneider, 2005; Neumann and Matzarakis, 2011) y
baja predictibilidad y regularidad de la cosecha y la calidad del vino (Schlutz 2000; Jones
et al., 2005a).

Entre los principales cambios observados a nivel de baya y vino, los más reportados
fueron: cambios en la composición química de la uva y del vino, cambios en las
características organolépticas (Orduna 2010; Bureau et al., 2000), cambios en la
secuencia de las etapas fenológicas (Jones et al., 2005), especialmente en la época de
brotación, comienzo de la cosecha., y en comienzo de maduración (Stock et al., 2005;
Ganichot 2002; Duchene and Schneider 2005; Sigler 2008; Petgen 2007; Duchene et
al., 2010; García de Cortazar Atauri 2006; Webb et al., 2007; Pieri 2010; Koufus et al.,
2014, 2017; Ramos et al., 2008). También un aumento de los eventos extremos con el
peligro que su ocurrencia supone para la seguridad de la cosecha (Kerry and Harrinson,
1992; Jones et al., 2005).

A nivel español, junto con los impactos ya mencionados, se ha observado un descenso

en la superficie vitícola en el noreste peninsular como consecuencia del estrés hídrico
(Camps and Ramos, 2012), un aumento de la demanda de regadío (Alonso and O’Neill,
2011) y una reducción de la esperanza de vida de la vid en un 30%. Según un estudio
de la Universidad de la Rioja (Andrés et al., 2020), el 90% de los profesionales asociados
a alguna Denominación de Origen han notado los efectos del cambio climático y el 56%
consideran que esos impactos les están afectando de forma considerable. Entre los
riesgos climáticos que más les afectan sitúan a las heladas, los pedriscos, las sequías
y las olas de calor.

Todos los cambios anteriormente citados suponen impactos económicos asociados a la

actividad del viñedo, que dependiendo de la región donde se hayan observado han
supuesto un impacto positivo o negativo. Por ejemplo, el aumento de las temperaturas
mínimas durante la maduración ha supuesto un descenso en la calidad de ciertos vinos
procedentes de la Península Ibérica (Fraga et al., 2012; Malheiro et al., 2010; Kenny
and Harrinson 1992; Koundouras et al., 1999; Santos et al., 2010b) y el aumento de las
temperaturas máximas ha supuesto una degradación de las características
organolépticas en muchas de las principales regiones vitícolas (Orduna 2010; Buttrose

37
et al., 1971; Downey et al., 2006; Bureau et al., 2000). Mientras que esas mismas
variaciones térmicas han favorecido la producción vitícola y la mejora en la calidad del
vino en regiones frías (Gouveia et al., 2011), ver figuras 23 y 24. En resumen, los
cambios asociados al impacto del cambio climático, tanto a nivel de planta, finca o
socioeconómico, no son proporcionales en todo el territorio (Hannah et al., 2013).

De todas las alteraciones ocasionadas por el cambio climático, hay tres que destacan
especialmente: aumento del CO2, alteraciones en el régimen pluviométrico y aumento
de las temperaturas. Debido a su ubicación en el sur de Europa, España se espera que
sea una de las regiones más afectadas por los efectos del cambio climático,
especialmente por este aumento de las temperaturas y el estrés hídrico lo que hace
urgente determinar la relación entre clima y viñedo y evaluar su evolución futura.

Figura 23. Efectos negativos del cambio climático a nivel viñedo relacionados con el aumento de CO 2, con alteraciones
en el régimen pluviométrico y con el aumento de las temperaturas. Elaboración propia.

Figura 24. Efectos positivos del cambio climático a nivel viñedo relacionados con el aumento de CO 2, con alteraciones
en el régimen pluviométrico y con el aumento de las temperaturas. Elaboración propia.

38
1.3.4. Indicadores bioclimáticos como herramientas para el estudio del viñedo

La manera en la que es posible evaluar la relación entre clima/meteorología y los

distintos factores que afectan a la vid y al conjunto de la producción vitícola, es a partir
de los indicadores bioclimáticos (Fregoni 2003). Un indicador o índice bioclimático es un
parámetro que mide una variable en función de sus valores climáticos promedios. A
través de una selección de indicadores bioclimáticos se pueden estudiar distintos
factores de interés vitícola como la idoneidad climática de una región para cultivar la vid,
el tipo de variedad más idóneo o la posibilidad de aparición de ciertas pestes y/o
enfermedades.

En líneas generales, los estudios clásicos utilizan índices individuales calculados en

base a la temperatura y la precipitación y derivados de ellos. A continuación se
enumeran los relevantes:

Figura 25. Recopilación de estudios sobre impactos del cambio climático. Elaboración propia

Entre los indicadores individuales más utilizados destacan la temperatura y la

precipitación durante la época de crecimiento, la temperatura y la amplitud térmica diaria
durante la maduración, los días con temperaturas muy extremas, ocurrencia de heladas,
los grados día o distintos índices de clasificación climática como Winkler (Winkler et al
1994) o Huglin (Huglin 1978).

Estudios más recientes ponen de manifiesto la necesidad de trabajar con índices

combinados ya que representan de forma más completa la clasificación y
discriminación vitícola (Ibacache, 2010) y permiten caracterizar la calidad del vino
(Huglin 1978; Maglhaes 2008). Esta forma de trabajar se engloba dentro del concepto
de zonificación vitícola y es el primer paso para evaluar el potencial vitícola de una región
(Malherio et al., 2010). Entre los más utilizados se encuentran el CatI (Fraga et al., 2014),
índice de Zuluaga (Westphalen and Maluf 2000) o el Grape Water Index (Teixeira et al.,
2012), entre otros.

Tonietto y Carbonneau definieron en 2003 el Geoviticulture Multicriteria Climatic

Classification (MCC) System, siendo el primer estudio que plantea una metodología para
describir de forma macro climática la variabilidad vitícola de una región. El MCC System
se basa en la combinación de tres índices complementarios entre sí y que aportaban

39
tanto información heliotérmica como hídrica: Huglin Index, Cool night Index y Dryness
Index. Esta clasificación supuso el punto de partida para comparar y establecer grupos
de regiones similares de forma más eficiente que la que se venía usando hasta la fecha,
que normalmente solo incluía un único índice y que se ha comprobado que recoge de
forma más que suficiente la variabilidad vitícola (Blanco-Ward et al., 2007). El uso del
MCC System como clasificación vitícola está muy extendido y muestra de ellos son las
múltiples regiones en las que se ha aplicado y en las que se utiliza como herramienta
de adaptación al cambio climático (Fraga et al., 2014; Irimia et al., 2013; Tonietto et al.,
2010; Vukovic et al., 2010).

Por lo tanto, los índices bioclimáticos son herramientas que permiten evaluar cómo
ciertas variables pueden afectar a la vid en distintos momentos de su ciclo vegetativo y
cómo pueden condicionar las características de las uvas y de los tipos de vino. Disponer
de esta información es de gran utilidad para realizar una óptima planificación del viñedo,
no solo en las actividades temporales (es decir, año a año) sino en actividades que han
de planificarse a medio-largo plazo (como por ejemplo, elegir variables más resilientes
a las nuevas condiciones climáticas o trasladar el viñedo a localidades más altas).

1.3.5. Denominaciones de Origen

Una DO permite identificar un producto originario de un lugar determinado

cuyascaracterísticas principales y diferenciadoras son exclusivamente consecuencia del
medio geográfico en que se producen, de los factores humanos y naturales que
intervienen en su desarrollo y de las fases de producción que se realizan en dicha zona
geográfica.

Las DOs es el sistema que se utiliza en España para reconocer la calidad diferenciada
de un vino. A través del Consejo Regulador se entregan las calificaciones de DOs a
aquellos vinos que, por su materia prima y su forma de ser elaborados y envasados,
solo pueden ser generados en una determinada región.

Para que un vino sea reconocido con el distintivo de DO debe cumplir, además de las
características propias de la región donde se elaboran, que sea reconocido en el
mercado y que haya sido reconocido como vino de la zona durante mínimo 5 años.
Existe una denominación de origen más exclusiva, la Denominación de Origen
Calificada (DOCa.), que se concede a aquellos vinos que además de cumplir los
requisitos de DO cumplen otras condiciones como los controles de calidad a los que se
someten y las condiciones de su comercialización (figura 26).

40
 Figura 26. Requisitos necesarios para que un vino se distinguido como DO o DOca. Elaboración propia.

Desde la primera DO legalmente certificada, la Denominación de Origen Jerez-Xérés-

Sherry en 1935, hasta el 2021, el número de zonas con certificación de DO ha ascendido
hasta 101. En lo referente a las DOCa solo hay dos que cumplen esos requisitos: Rioja
(localizada en La Rioja, País Vasco y Navarra) y Priorat (en Cataluña). Se pueden
encontrar alguna DO en todo el territorio español como se aprecia en la figura 27.

Figura 27. Listado de DO clasificado por comunidades autónomas. Elaboración propia.

41
2. JUSTIFICACIÓN Y OBJETIVOS

42
De todo lo dicho anteriormente se desprende que el cambio climático supone nuevos retos
para la Península Ibérica que repercutirán en el sector vitícola español lo que hace
imprescindible disponer de información climática local predicha de calidad.

Ya existían varios estudios de downscaling específicamente sobre la temperatura en

España (Brands et al., 2011b; Frías et al., 2005; Frías et al., 2010; Hervada-Sala et al.,
2000; Miró et al., 2016; Turco et al., 2014), pero muy pocos estudios han explorado las
temperaturas extremas en este territorio, tanto a nivel histórico (Fernández-Montes y
Rodrigo, 2012; Fonseca et al., 2016) como a futuro (Pereira et al.2017; Lorenzo et al.,
2021, Pereira et al., 2021, Lorenzo and Alvarez, 2022).

Pero como ya se ha visto, solo unos pocos estudios han desarrollado escenarios de
temperatura en la región de Aragón (noreste de España) (Buerger et al., 2007; Goncalves
et al., 2014; Ribalaygua et al., 2013a), pero ninguno de ellos desarrolló escenarios de
temperaturas extremas u olas de calor. Sólo (Barrera-Escoda et al., 2014) desarrolló
escenarios de temperaturas extremas en la cuenca del Ebro. Ese estudio mostró que el
aumento previsto del número de noches tropicales y de las temperaturas extremas podría
tener un efecto negativo sobre la salud humana y las condiciones de confort. Otros
estudios también han analizado los posibles impactos del cambio climático en esta región
sobre la salud humana (Roldán et al., 2016). Ambos estudios utilizaron los modelos
climáticos asociados al 4º informe de evaluación del IPCC y no al 5º como es nuestro caso.

Los escenarios de sequía en España también son escasos y no informan del impacto que
estos extremos supondrán para la región.

En el caso de las temperaturas extremas, se han realizado varios estudios que exploran
las temperaturas extremas actuales (Kuglitsch et al., 2010) y futuras en Europa (Carvalho
et al., 2021). Sin embargo, o se basan en modelos anteriores al CMIP5 o usan downscaling
dinámico.

Aunque muchos modelos climáticos tienen dificultades para reproducir adecuadamente

los extremos climáticos, como las olas de calor (Stegehuis et al., 2015), unos pocos
estudios han informado sobre escenarios de olas de calor en Europa (Fischer and Schar
2010; Schoetter et al., 2015), o situados en la región mediterránea (Zittis et al., 2016), en
Francia (Planton et al., 2008), Italia (Tomozeiu et al., 2007) o Finlandia (Kim et al., 2018).
Schoetter et al. (2015) y Kim et al. (2018) son uno de los pocos estudios que ya utilizan
los modelos del IPPC5. Como parte del Experimento Europeo Coordinado de Desescalado
Regional (EURO-CORDEX) (Giorgi et al., 2009), la iniciativa EURO-CORDEX proporciona
proyecciones climáticas regionales para Europa. Vautard et al., (2013) utilizó el proyecto
EURO-CORDEX para simular olas de calor a escala regional europea, proporcionando
una reducción de escala de las simulaciones CMIP5. Estos estudios a escala europea sólo
incluían algunas localidades de España, pero no Aragón.

La industria vitícola es un ejemplo de globalización (Anderson 2003, Hussein et al., 2008)

con el desarrollo de nuevas regiones de consumo y producción, expansión del comercio
internacional y de innovación tecnológica. Las medidas llevadas a cabo por la industria
para reducir los impactos del cambio climático (Galbreath 2014, Alonso and O’Neill 2011,
Lereboullet et al., 2013a) refuerzan la necesidad de disponer de información climática
(Lemos et al., 2012) y estrategias de comunicación con la industria (Lemos et al., 2012;
Holland et al. 2010).

43
A pesar de que son muchos los indicios e informes sobre los efectos que el cambio
climático ya ha ocasionado en el sector vitícola así como el esfuerzo llevado a cabo por la
comunidad científica para identificar las relaciones entre variables meteorológicas y vid y
el impacto que dichos cambios tienen en el futuro, siguen siendo escasos los estudios
enfocados a analizar cómo los productores y los agricultores vitícolas pueden adaptarse a
dichos cambios (Holland et al. 2010).

Desde el punto de vista de nuestro territorio y aunque existen estudios que evalúan el
impacto del cambio climático en los viñedos del territorio español ibérico-peninsular, la
mayoría lo hacen en base a proyecciones climáticas dinámicas (que no han tenido en
cuenta la climatología local), salidas directas de los modelos climáticos o un número
reducido de localizaciones españolas ya que forman parte de estudios que abarcan áreas
geográficas más extensas. No existe hasta la fecha ningún estudio que evalúe el impacto
del cambio climático en los viñedos del territorio ibero-español mediante índices
bioclimáticos calculados en base a proyecciones climáticas regionalizadas a escala local
con una técnica de downscaling estadístico (considerando la climatología local) generadas
a partir de modelos climáticos pertenecientes a la quinta fase del CMIP.

Por tanto, los escenarios de olas de calor/frío y sequía en Aragón a escala local así
como los escenarios de indicadores bioclimáticos de interés vitícola en territorio
Ibérico-peninsular español no han sido obtenidos hasta la fecha, lo que justifica los
objetivos de esta tesis pues es imprescindible disponer de escenarios locales para
determinar el impacto del cambio climático en la realidad ambiental o socioeconómica de
cada región para tomar decisiones de adaptación al cambio climático. La manera en la que
se han generado proyecciones de temperatura y precipitación a escala local en base a
modelos CMIP5 cubre una laguna de información climática en la región de Aragón.
Finalmente, y dado que la industria vitícola aragonesa se puede beneficiar de las prácticas
y soluciones llevadas a cabo por regiones climáticas similares, se justifica la necesidad de
ampliar el estudio de indicadores vitícolas a todo el territorio ibérico-peninsular.

OBJETIVO:

Por lo tanto, el objetivo principal de este estudio es generar escenarios de clima futuro a
escala local en España para el siglo XXI de temperatura y precipitación así como de
eventos entremos de olas de calor/frío y sequías a fin de disponer de las bases
climatológicas para calcular índices bioclimáticos que puedan valorar a escala local el
impacto vitícola. Este estudio permitirá evaluar la idoneidad de la zona de estudio para el
cultivo vitícola, así como determinar qué zonas van a perder potencial vitícola y cuáles van
a ganarlo, lo que supondrá una información de gran utilidad para definir posibles medidas
de adaptación del sector vitícola frente al cambio climático.

Para ello, se han planteado los siguientes objetivos secundarios:

a) Generación de escenarios para el Siglo XXI de temperatura y precipitación para

Aragón a escala local en base a modelos pertenecientes al CMIP5 usando una
metodología de downscaling estadístico basado en un método de análogos en dos
pasos (Ribalaygua et al., 2013a). Como escenario futuro se han escogido los RCPs
4.5 y 8.5 como representación de una situación de evolución futura promedio y
extrema.

44
b) Generación de escenarios de olas de calor/frío y sequías para Aragón a partir de
las series de temperatura y precipitación generadas en el objetivo anterior.

c) Ampliar la generación de escenarios de temperatura y precipitación, así como los

escenarios de olas de calor/frío y sequías para todo el territorio español ibérico-
balear como base metodológica a la generación de índices bioclimáticos para todo
el territorio.

d) Finalmente, en base a estas proyecciones de temperatura y precipitación, la

generación de escenarios futuros de seis indicadores bioclimáticos a escala local:
índice de Huglin, Índice de sequedad, Índice de frescor nocturno, Índice
hidrotérmico de Branas, Bernon and Levadoux, Sistema de clasificación vitícola
“Geoviticulture Multicriteria Climatic Classification” y el Índice CompI. para todo el
territorio Ibérico-balear español.

45
3. METODOLOGÍAS

46
Atendiendo a los objetivos expuestos en el apartado anterior la tesis ha constado de tres
fases principales. Una primera fase en la que se han generado los escenarios de
temperatura y precipitación regionalizados para Aragón y evaluado los cambios esperados
en dichas variables y cómo afectarán al territorio. Posteriormente, y utilizando las series
regionalizadas en la primera fase, se han generado escenarios futuros de episodios de
olas de calor/frío y sequía para Aragón y los impactos asociados a dichos eventos
extremos. Finalmente, se ha procedido a calcular diversos indicadores climáticos de
interés vitícola para todo el territorio español ibérico-balear (Figura 28).

Figura 28. Fases desarrolladas a lo largo del estudio. Elaboración propia.

Tras el desarrollo de las fases 1 y 2 y como punto de partida para contextualizar la fase 3,
se han generado escenarios de temperatura y precipitación así como escenarios de olas
de calor/frío y sequía para todo el territorio español ibérico-peninsular. Esta información
supone un complemento al estudio realizado en dichas fases y permite obtener una visión
total del impacto climático y extremo en términos térmicos y pluviométricos.

3.1. Generación de escenarios de clima futuro de temperatura y precipitación

Teniendo en cuenta los puntos analizados en el apartado de introducción sobre

modelización climática y considerando las características orográficas y climáticas de la
zona de estudio así como la disponibilidad de datos observados, se ha elegido una
metodología de downscaling estadístico basada en dos pasos (Ribalaygua et al., 2013a)
como herramienta para la generación de las proyecciones climáticas de temperatura y
precipitación.

47
3.1.1. Metodología de downscaling estadístico en dos pasos

Esta metodología consta de dos pasos: estratificación analógica y funciones de

transferencia que permitan establecer relaciones entre los predictores (campos
atmosféricos de baja resolución) y los predictandos (variables en superficie). La manera
de trabajar en el segundo paso va a depender de la variable a simular.

 Primer paso: estratificación analógica

El primer paso consiste en una estratificación analógica basada en la hipótesis de que los
patrones atmosféricos "análogos” (predictores) deben causar efectos locales “análogos”
(predictores), lo que significa que se seleccionan los días más parecidos al día que se va
simular (Benestad et al., 2007; Zorita y von Storch, 1999). La similitud entre dos días
cualesquiera se midió utilizando una distancia pseudoeuclidiana entre los campos a gran
escala utilizados como predictores entre cada día análogo y el día problema. Para cada
predictor, se calculó y estandarizó la distancia euclidiana ponderada sustituyéndose por el
percentil más cercano de una población de referencia de distancias euclidianas
ponderadas para ese predictor. Este método es adecuado para reproducir las relaciones
no lineales entre los predictores y los predictandos, pero no sirve para simular valores
fuera del rango de valores observados (Imbert y Benestad, 2005). Es por esto que se
procede a realizar un segundo paso.

 Segundo paso: funciones de transferencia

El comportamiento estadístico de cada variable (temperatura y precipitación en este caso)

es muy diferente y, por tanto, el segundo paso depende de la variable.

1) Para estimar las temperaturas mínimas y máximas diarias, se realiza, por cada variable,
una regresión lineal múltiple con selección automática de predictores sobre el conjunto de
análogos seleccionado en el paso 1. Las relaciones lineales se aplican sobre un conjunto
de predictores potenciales, compuesto por los valores de las variables atmosféricas en la
vertical del punto (espesores) para el que se quiere estimar la temperatura en superficie,
un indicador de la duración de la noche en el día en cuestión y un promedio ponderado de
las temperaturas de los días anteriores. Una vez establecida la relación lineal existente
entre los predictores seleccionados y el predictando (temperatura mínima o máxima), se
aplica dicha relación a los valores de los predictores del día problema para estimar el valor
del predictando en dicho día.

2) Para estimar la precipitación, en primer lugar, cada grupo de m días problema se

regionaliza mediante el primer paso (m se elige como el número de días de un mes
concreto). A continuación, y para cada día problema de estos m días, se obtiene una
cantidad de precipitación preliminar promediando la cantidad de lluvia de sus n días más
análogos, por lo que las m fechas problema se ordenan de mayor a menor por cantidad
de precipitación preliminar. Para asignar la cantidad de lluvia final, tomamos cada una de
las cantidades de lluvia de los m×n días análogos y las ordenamos y agrupamos en m
grupos. Es decir, las m cantidades de precipitación final también se ordenan, al igual que
las m fechas problema (que se ordenan previamente a partir de la precipitación preliminar).
A continuación, ambas se vinculan: cada cantidad ordenada se asigna a cada una de las
m fechas ordenadas.

48
Finalizados ambos pasos se tiene un conjunto de 18 proyecciones climáticas (9 ESMs x
2RCPs) de temperatura mínima, temperatura máxima y precipitación en cada uno de los
observatorios considerados en el estudio. Será lo que llamemos proyecciones climáticas
“brutas”, es decir, no se les ha corregido el posible error sistemático asociado al método
empleado.

3.1.2. Procesos de verificación y validación

Una vez que se han generado escenarios climáticos de precipitación y temperatura a

escala local, se ha llevado a cabo un proceso de verificación (comparativa entre series
simuladas a partir del reanálisis frente a datos observados) y un proceso de validación
(comparativa entre series simuladas a partir del Historical de cada ESMs frente a series
simuladas a partir del reanálisis). Los datos obtenidos en ambos procesos serán de utilidad
para cuantificar y corregir el error sistemático consecuencia del método empleado.

3.1.3. Corrección del error sistemático

A las series “brutas” de proyecciones de temperatura y precipitación se les aplica un

método de corrección de error sistemático con el fin de que su uso a nivel absoluto (y no
como incremento) sea fiable y robusto.

En este caso, se ha empleado un método que permite corregir los datos de forma
proporcional, es decir, a aquellos valores que presenten un mayor error se les aplica una
corrección mayor y viceversa. La magnitud del error se obtiene: 1) comparando la función
de distribución de los valores observados frente a la función de distribución de los valores
simulados en base al ERA-40 (error 1 o de verificación) y 2) comparando la función de
distribución de los valores simulados en base al ERA-40 con la función de distribución
correspondiente a los valores simulados en base al Historical para cada modelo empleado
(error 2 o de validación). En este caso, se emplea la función de distribución acumulada
empírica (ECDF).

Al comparar las dos funciones entre sí se pueden establecer relaciones de

proporcionalidad (no lineales) entre ambas, siendo las diferencias obtenidas las que nos
permiten corregir el error entre ambos grupos. Una vez obtenidos los errores 1 y 2, estos
se combinan y se aplican sobre la variable a corregir.

Una vez finalizado todo el proceso, los escenarios de clima futuro de temperatura y
precipitación suponen un conjunto de proyecciones climáticas fiables y robustas que sirven
de base para estudios de eventos extremos e indicadores bioclimáticos.

3.2. Generación de escenarios de clima futuro de valores extremos

3.2.1. Olas de calor/frío

Teniendo en cuenta las diferentes definiciones planteadas para cuantificar las olas de calor
y frío propuestas por los diferentes organismos oficiales nacionales e internacionales
comentadas en la introducción, y considerando las características climáticas de la zona de
estudio, se ha aplicado la siguiente definición de ola de calor en este estudio (figura 29):

“al menos tres días consecutivos con temperatura máxima superior al percentil 95
obtenido de su serie de temperatura máxima observada y calculado entre los meses de
junio a septiembre durante el periodo 1980-2000”.
49
 Figura 29. Definición de ola de calor y frío empleada en este estudio. Elaboración propia.

y se ha considerado como un episodio de ola de frío como (figura 29):

“al menos tres días consecutivos con temperatura mínima inferior al percentil 5 obtenido
de su serie de temperatura mínima observada y calculado entre los meses de noviembre
a abril durante el periodo 1980-2000”

Los resultados obtenidos se han analizado en base a varias consideraciones: intensidad

media y máxima, duración y extensión espacial del evento.

3.2.2. Episodios de sequía

Después de un extensivo análisis sobre indicadores de sequía y atendiendo a las

características climáticas de la región de estudio se han seleccionado el SPI y el SPEI
como indicadores idóneos para evaluar la sequía meteorológica en Aragón. Por un lado,
la definición de ambos indicadores permite su comparativa entre regiones y periodos
temporales, por otro, el SPEI fue desarrollado y testado precisamente en Aragón por
Vicente-Serrano et al. 2010. Además, el SPEI al incluir el efecto de la temperatura permite
llevar a cabo un estudio de la importancia que tiene en el balance hídrico, punto muy
importante en el contexto actual de calentamiento global.

Para evaluar los resultados se ha tenido en cuenta la intensidad de los indicadores así
como su extensión espacial, además se ha considerado de interés calcular los valores de
ambos índices a 6, 12, 24 y 60 meses abarcando situaciones de corto a largo plazo.

3.3. Generación de escenarios de clima futuro de indicadores bioclimáticos

Una vez obtenida una idea clara del impacto del cambio climático en el clima medio y en
la ocurrencia de eventos extremos en Aragón se procede a evaluar qué implicaciones
pueden tener estas alteraciones climáticas en el cultivo de la vid. En esta fase del estudio,
se ha ampliado el área de análisis a todo el territorio Ibérico-Peninsular. El motivo de
ampliar el área se debe a que los resultados obtenidos en otras regiones del país pueden
se orientativas para los viticultores aragoneses a la hora de tomar medidas de adaptación
frente a los problemas asociados al cambio climático.

La cantidad de indicadores vitícolas a emplear es muy extensa, pero en base a la literatura

científica y al enfoque que se le quería dar al estudio (basado en el clima promedio y no
en extremos) se han elegido como indicadores bioclimáticos: índice de Huglin, Índice de
50
sequedad, Índice de frescor nocturno, Índice hidrotérmico de Branas, Bernon and
Levadoux, Sistema de clasificación vitícola “Geoviticulture Multicriteria Climatic
Classification” y el Índice CompI.

A continuación, se expone la definición de cada uno de los índices propuestos.

 Índice de Huglin (HI)

El Índice de Huglin es un índice termal basado en grados días, es decir en el concepto de

acumulación de calor. Este índice se utiliza para evaluar la demanda térmica y radiativa
básica de la uva durante el periodo de crecimiento que garantice una completa y adecuada
maduración.

Cada variedad de uva requiere una acumulación de calor determinada para que la
maduración se produzca de forma óptima (Tonietto, 1999). Según los valores obtenidos
de HI, el clima variará entre muy frío o muy cálido:

Figura 30. Índice de Huglin: definición matemática, escala de intensidades e implicaciones asociadas a cada categoría.
Elaboración propia.

 Índice de sequedad (Dryness Index DI)

El Índice de sequedad (DI) evalúa la disponibilidad de agua en el suelo proporcionando

información sobre las condiciones de estrés hídrico. Según diversos estudios (Blanco-
Ward et al., 2007; Vanderlinden et al., 2004; Brixner et al., 2014) y concretamente Fonseca
et al (2012), se eligió la fórmula de Hargreaves para el cálculo del DI en lugar de otras
formulaciones más complejas en el cálculo de la ETP.

51
Figura 31. Índice de sequedad: definición matemática, escala de intensidades e implicaciones asociadas a cada categoría.
Elaboración propia.

 Índice de frescor nocturno (Cool night Index CI)

El índice de frescor nocturno (CI) es un índice termal basado en la temperatura nocturna

durante el periodo de maduración (septiembre en el HN). Temperaturas nocturnas
moderadamente bajas junto con temperaturas diurnas altas favorecen la producción de
vinos de alta calidad ya que se favorece la síntesis de algunas componentes fenológicas.

Figura 32. Índice de frescor nocturno: definición matemática, escala de intensidades e implicaciones asociadas a
cada categoría. Elaboración propia.

 Índice hidrotérmico de Branas, Bernon and Levadoux (HyI)

El índice hidrotérmico de Branas, Bernon and Levadoux es un índice que combina el efecto
de la humedad del aire (a través de la precipitación) y la temperatura en la época de
crecimiento con el fin de evaluar el riesgo de exposición de la uva a ciertas enfermedades
como el Mildiu.

52
 Figura 33. Índice hidrotérmico de Branas, Bernon y Levadoux: definición matemática, escala de intensidades e
implicaciones asociadas a cada categoría. Elaboración propia.

 CompI Index

El CompI Index (Malheiro et al., 2010) es un índice que permite evaluar la adecuación
climática para el crecimiento de la uva. El CompI es el porcentaje de años óptimos para el
cultivo de la vid para un periodo dado. Se entiende por año óptimo aquel año en el que se
alcanzan unos umbrales críticos de los índices HI, DI, HyI y unas condiciones de
temperatura mínima.

Figura 34. CompI Index: definición matemática, escala de intensidades e implicaciones asociadas a cada
categoría. Elaboración propia.

 The Geoviticulture Multicriteria Climatic Classification (MCC) System

El sistema MCC se desarrolló para mejorar la caracterización del clima vitícola en las
regiones productoras de vino del mundo. Se trata de un sistema de clasificación climática
de las regiones vitícolas basado en la integración de las diferentes clases de tres índices
climáticos -DI, HI y CI-. De este modo, se establece el clima vitícola de cada región y se
pueden clasificar y agrupar las regiones. Así, el sistema permite la identificación de
análogos climáticos. Permite evaluar una región vitícola desde el punto de vista del clima
vitícola, del grupo climático y del clima vitícola con variabilidad intra-anual (Tonietto and
Carbonneau 2004).

53
 Figura 35. MCC System: definición matemática, escala de intensidades e implicaciones asociadas a cada
categoría. Elaboración propia.

Para el análisis de los resultados se ha optado por varios enfoques complementarios entre
sí:

 Enfoque 1: por un lado, se analizan los resultados a nivel de mapa (interpolado en

base a todos los observatorios disponibles para el estudio) y por otro, se analiza la
información agrupada por DOs. De esta manera, se puede optar por evaluar
directamente la DO que se quiere consultar y comparar con otras DOs de
características similares. También permite evaluar qué regiones del territorio serán
óptimas o dejarán de serlo para un tipo de variedad o vino en particular o bien
comparar regiones climáticas parecidas con maneras de producir similares y
establecer sinergias entre ellas.
 Enfoque 2: los valores de los indicadores se han representado de dos maneras: en
valores absolutos y categorizados. Las categorías en las que se mueven los
indicadores son bastante amplias por lo que no es lo mismo que una región se
encuentre en el límite superior de la categoría que en el inferior. Por ejemplo, si los
valores del DI se sitúan dentro del rango [-100,50] dos observatorios con valores
en ese rango pertenecerán a esa categoría pero desde el punto visto de aporte
hídrico no es lo mismo tener DI= -90 que DI= 45. De ahí la importancia de evaluar
ambas escalas simultáneamente.

3.4. Datos y área de estudio

3.4.1. Área de estudio

El estudio se ha llevado a cabo sobre dos regiones. La primera parte, en la que se ha

evaluado el impacto del cambio climático a través de las variables meteorológicas
primarias (temperatura y precipitación) y los eventos extremos asociadas a las mismas
(olas de calor/frío y eventos de sequía, respectivamente), se ha ejecutado principalmente
sobre Aragón y se ha complementado con información para toda España. En la segunda
parte se ha evaluado el impacto del cambio climático sobre los viñedos a través de
indicadores bioclimáticos, a nivel de territorio español ibérico-balear. Esto ha permitido,
por un lado, evaluar cómo se verán afectados los viñedos aragoneses y por otro, poder
comparar dichos resultados con los obtenidos en otras regiones ibérico-baleares. Disponer
de información en un territorio más extenso permite considerar qué medidas de adaptación
están tomando otras regiones con características climáticas y vitícolas similares a las
nuestras y considerar su implantación.

54
Características físicas y climáticas de Aragón

Aragón se encuentra ubicado en la zona noreste de la Península Ibérica (España) y ocupa

una extensión de aproximadamente 47,720 km2 (340 km de largo por 240 km de ancho).
Por su localización (figura 36), Aragón se encuentra en una zona con clima mediterráneo
del oeste, presentando inviernos muy fríos y veranos muy cálidos y secos así como escasa
pluviometría. Sin embargo, la gran diferencia de altitudes, más de 3000m de diferencia
entre los valles (como el valle del Ebro) y las zonas montañosas (Los Pirineos) junto con
la orografía específica de la zona (la cuenca hidrográfica del Ebro en el valle y la cadenas
montañosas, los Pirineos al norte y la Cordillera Ibérica al sur del valle, respectivamente)
hacen que las características climáticas se modifiquen localmente. Estas modificaciones
climáticas locales explican ciertas características típicas de la región como son: la
sequedad del terreno a lo largo de las riberas del río Ebro, la presencia de patrones de
lluvia aleatorios, un alto contraste térmico entre el invierno y el verano como consecuencia
de las fuertes características continentales de la región y los típicos vientos "mistral" del
noreste, que son frecuentes en la región.

Las temperaturas medias anuales de Aragón varían entre la zona del valle del Ebro cuya
temperatura media se sitúa en torno a los 15 ºC y las zonas más altas con temperaturas
medias de apenas 7 ºC. Un rasgo característico de la región es su alta amplitud térmica
anual, ya que en algunas zonas del territorio puede oscilar entre temperatura negativas en
los meses de inverno y temperaturas superiores a los 35 ºC en los meses estivales,
aunque en promedio la amplitud térmica anual se sitúa en torno a los 12 ºC.

Una característica importante de la región es la escasa precipitación, la cual se distribuye

claramente conforme al relieve como demuestran las isoyetas, que se disponen en círculos
concéntricos que van disminuyendo desde las zonas de montaña hasta el centro de la
región. Aunque la precipitación total media anual del territorio aragonés se sitúa en torno
a los 550 mm, hay regiones cuya media está por debajo de estos valores (por ejemplo, en
el sector central de la depresión del Ebro). Sólo en los Pirineos y, en menor medida, en la
Cordillera Ibérica, las precipitaciones alcanzan valores importantes, 1800-2000 mm y
muestran valores positivos de balance hídrico (considerando la diferencia entre
precipitación y la ETP).

Por otro lado, más del 60% de la región presenta valores medios de ETP superiores a
1100 mm, mostrando un balance hídrico negativo, a lo que contribuye no sólo la escasez
de precipitaciones sino también el fuerte viento ("Cierzo") característico del Valle del Ebro
(López et al., 2007). Así, el 70% del territorio aragonés se considera semiárido de acuerdo
al valor del índice propuesto por el Programa de Naciones Unidas para el Medio Ambiente
(inferior a 0,5) e incluso el 30% del territorio presenta valores del índice de 0,3. (Cherlet et
al., 2018).

55
 Figura 36. a) Temperatura media observada y b) precipitación media diaria observada en el territorio español Ibérico-
balear en el periodo 1971-2000. Elaboración propia.

Características físicas y climáticas del territorio español Ibérico-balear

El territorio español peninsular y las Islas Baleares considerado en el estudio abarca un

área de 588.294 km2 (figura 36). Debido a su ubicación próxima a grandes masas de agua
y por su compleja orografía (desde el nivel del mar hasta picos que superan los 3400 km)
el clima español es muy variado y complejo, de manera que se pueden contar hasta 13
regiones climáticas según la clasificación de Köppen, y múltiples microclimas.

El clima de la España ibérico-balear depende de su situación en el extremo suroccidental

de Europa y de su compleja orografía mientras que las islas Baleares se encuentran en el
Mediterráneo occidental cerca de la península ibérica y son relativamente montañosas.
Este espacio geográfico está afectado climáticamente por los vientos dominantes
provenientes del Atlántico y las borrascas asociadas al frente polar, las altas presiones
relacionadas con los anticiclones de las Azores y centroeuropeos, las bajas norteafricanas,
la influencia del Mediterráneo y las advecciones de masas de aire frío centroeuropeo y
ártico.

Las precipitaciones medias totales anuales en la Península disminuyen, en general, de

norte a sur y de oeste a este. Los máximos pluviométricos se presentan en el norte
peninsular (costa Atlántica) mientras que los valores mínimos se encuentran en el sureste
peninsular (litoral de Almería y Murcia). Las precipitaciones son más abundantes, en
general, en otoño e invierno y escasas en verano. La mayor parte del territorio presenta
sequía estival y son relativamente frecuentes los chubascos intensos asociados a
inestabilidad y la nubosidad convectiva, sobre todo en las regiones mediterráneas y del
suroeste peninsular.

Respecto a las temperaturas, las más bajas se producen en el mes de enero y las más
altas en julio. En invierno, en las zonas de montaña y en el interior peninsular hay
frecuentes heladas y se pueden alcanzar temperaturas mínimas muy bajas, tanto los días
anticiclónicos con fuerte irradiación nocturna como los días de advecciones frías árticas o
centroeuropeas. Independientemente de las altas cumbres, se alcanzan temperaturas

56
invernales muy bajas en valles intramontanos del Pirineo aragones-catalán, del sistema
Ibérico y de la vertiente sur de la cordillera Cantábrica. En verano las máximas
temperaturas se alcanzan en zonas del centro del valle del Guadalquivir y del sur de
Extremadura. En el litoral galaico-cantábrico las temperaturas veraniegas son
relativamente suaves y la interacción de las brisas marinas con los montes costeros origina
algunos días de llovizna; en el litoral mediterráneo el verano se caracteriza por la fuerte
insolación, la humedad relativa alta con sensación de bochorno y el largo periodo de
sequía, aunque pueden producirse algunas tormentas.

Los factores fundamentales del paisaje vegetal son el clima, el suelo y la geomorfología.
La mayor parte de Galicia, la cornisa cantábrica y los Pirineos, pertenecen a la región
biogeográfica Eurosiberiana (normalmente sin sequía estival), mientras que el resto de la
Península y Baleares pertenecen a la región mediterránea.

3.4.2. Bases de datos empleadas

Para llevar a cabo un estudio de análisis de impacto del cambio climático en base a las
metodologías que hemos seleccionado, es necesario disponer de dos grupos de datos:
datos observados en superficie y datos atmosféricos (reanálisis y modelos climáticos). En
la figura 37, se muestra, a modo resumen, en qué parte del proceso se requiere cada base
de datos.

Figura 37. Esquema de las bases de datos necesarias para la generación de escenarios de clima futuro de indicadores
bioclimáticos y fenómenos extremos así como los procesos en los que son necesarias. Elaboración propia.

Base de datos observados en superficie (campos predictandos)

 Base de datos empleados en el estudio propio de Aragón

El conjunto de datos observacionales utilizado en el presente estudio consiste en una serie

temporal de las temperaturas máximas y mínimas diarias así como de precipitación,
obtenidas de la extensa red de observatorios instrumentales de AEMET
(http://www.aemet.es) y distribuidos de forma bastante homogénea, por todo el territorio
de Aragón.

57
Estos observatorios están sometidos a un estricto control de calidad de los datos, que
incluye la detección y corrección de múltiples puntos críticos en las series temporales
climáticas a largo plazo, la cobertura de las lagunas, la evaluación de la homogeneidad de
las series climáticas y la corrección de las inhomogeneidades encontradas.

Para garantizar la calidad de los datos, se descartaron las estaciones con un gran número
de lagunas de datos o sin al menos 15 años de registros diarios. En segundo lugar, se
analizó el conjunto de datos para detectar la presencia de puntos de datos extremos que
pudieran considerarse observaciones sospechosas o errores de transcripción
(consideramos que los puntos de datos eran extremos si se separaban de la media
mensual en más de cuatro veces el valor de la desviación estándar). En tercer lugar, se
verificó que los datos no contuvieran inhomogeneidades que pudieran introducir un sesgo
en los registros que no estuvieran relacionados con el clima.

El primer conjunto de datos (compuesto por 103 observatorios) se utilizó para analizar y
elaborar los escenarios climáticos regionales de temperaturas máximas y mínimas (figura
38a, puntos rojos). De estos 103 observatorios, los que tenían al menos el 75% de los
datos de temperatura máxima recogidos de junio a septiembre (es decir, el periodo de
referencia para el cálculo de las olas de calor) entre 1980 y 2000 y los observatorios que
tenían al menos el 75% de los datos de temperatura mínima recogidos de noviembre a
abril (es decir, el periodo de referencia para el cálculo de las olas de frío) entre 1980 y
2000 configuran un segundo conjunto de datos seleccionados para el estudio de los
eventos extremos de olas de calor y frío, respectivamente. En ambos casos, el número de
observatorios fue de 71 (figura 38b).

Figura 38. a) Ubicación de los observatorios de temperatura (en rojo) y de precipitación (en rojo y negro) utilizados para la
generación de los escenarios futuros, b) localización de los escenarios utilizados en el cálculo de olas de calor y frío (en
amarillo), solo para olas de calor (en negro) y solo para olas de frío (en azul), c) localización de los escenarios utilizados
en el cálculo de los indicadores de sequía (puntos naranjas). Elaboración propia.

58
Para la simulación de futuros escenarios climáticos de precipitación, se utilizó un primer
conjunto de 263 estaciones (figura 38a, puntos rojos y negro). De estas 263 estaciones,
sólo se utilizaron para la simulación de los índices de sequía las que tenían datos para
ambas variables, temperatura y precipitación (figura 38c).

 Base de datos empleados en el estudio propio del territorio español ibérico-balear

Al igual que en el caso de estudio de Aragón, los datos empleados provienen de la red de
observatorios instrumentales de AEMET y cumple los mismo requisitos de calidad que se
describen en el apartado anterior. Esta base de datos ha sido empleada en numerosos
estudios de generación de escenarios climáticos futuros en este área (Gaitan et al., 2019;
Gaitan et al., 2020; Gomez-Martinez et al., 2021; Monjo et al., 2016; Ribalaygua et al.,
2013b). En el estudio se han utilizado un total de 1778 observatorios con datos tanto de
temperatura como de precipitación, cubriendo ampliamente todo el territorio estudiado
(figura 39).

Para el análisis de los resultados por DOs, se han elegido aquellos observatorios que se
encuentran dentro del territorio clasificado como tal (figura 39). En total, hay 789
observatorios dentro de 59 DOs situadas en el territorio ibérico-balear. Se dejan fuera del
presente estudio las DOs en las que no se encontraron observatorios con información de
temperatura y precipitación simultáneamente o con información meteorológica escasa, lo
que supuso sólo alrededor del 1,3% del total de la superficie ocupada por las DOs.

Figura 39. a) Estaciones meteorológicas empleadas en el estudio vitícola así como b) localización de las DOs
seleccionadas para su análisis climático. Elaboración propia.

Datos atmosféricos (campos predictores)

 Reanálisis Europeo ERA-40

Para estudiar el comportamiento de las simulaciones históricas del modelo CMIP5, se

utilizó el conjunto de datos de reanálisis del Centro Europeo de Previsiones
Meteorológicas a Medio Plazo (ECMWF ERA-40;
http://www.ecmwf.int/research/era/do/get/) (Uppala et al., 2005) para el periodo 1958-2000
con una resolución temporal de 6 horas y una resolución espacial de 125 km. Para la
59
verificación de la metodología, fue necesario reducir la escala temporal y espacial del
reanálisis para poder comparar tanto ERA-40 como las simulaciones del modelo climático
(Ribalaygua et al., 2013a; Ribalaygua et al., 2013b). Los límites geográficos de la ventana
atmosférica utilizada fueron las latitudes 31,5°N a 55,1°N y las longitudes 27,0°W a 14,6°E,
cubriendo no sólo el área geográfica en estudio sino también las zonas atmosféricas
circundantes que ejercen una influencia meteorológica en toda la Península Ibérica
(Ribalaygua et al., 2013a). El uso del conjunto de datos ERA-40 nos ha permitido comparar
estos nuevos resultados con los publicados por Ribalaygua et al 2013b basados en
modelos del CMIP3.

 Modelos climáticos globales pertenecientes al CMIP5

En el presente estudio, se trabajó con un conjunto de nueve modelos climáticos que

pertenecen al CMIP5 y que fueron proporcionados por los archivos del Programa de
Diagnóstico e Intercomparación de Modelos Climáticos (PCMDI) (Tabla 1). El número de
modelos elegidos se limitó a la disponibilidad de datos con frecuencias temporales diarias.
Todos los modelos eran ESM.

Para generar los campos predictores de los modelos, tomamos los valores diarios de
varios campos de interés a gran escala (es decir, altura geopotencial, humedad específica
y viento) en diferentes niveles de presión para los nueve ESMs mostrados en la tabla 1
asociados a la CMIP5 (Taylor et al., 2012). Además, utilizamos un experimento histórico
(Taylor et al., 2012) que generó simulaciones del pasado siglo XX y que fue útil para la
evaluación del rendimiento del modelo frente al clima actual. Por último, se consideraron
dos escenarios o RCPs (Moss et al., 2010), concretamente el escenario RCP8.5 "alto" y
el escenario RCP4.5 "intermedio", ambos correspondientes a diferentes rangos posibles
de forzamiento radiativo alcanzado en el año 2100 respecto a los valores de la era
preindustrial (4,5 y 8,5 W/m2, respectivamente).

60
 Resolución Centro de investigación
Modelo climático Referencias
espacial/temporal responsable
2ºx2,5º
National Oceanic and Atmospheric
GFDL-ESM2M Dunne et al. (2012)
Administration (NOAA), E.E.U.U.
daily

2,8ºx2,8º
Canadian Centre for Climate
CanESM2 Modeling and Analysis (CC-CMA), Chylek et al. (2011)
daily Canadá.

1,4ºx1,4º CNRM (Centre National de

CNRM-CM5 Recherches Meteorologiques), Voldoire et al. (2013)
daily Meteo-France, Francia.

1,4ºx1,4º Beijing Climate Center (BCC), China

BCC-CSM1-1 Meteorological Administration, Xiao-Ge et al. (2013)
daily China.

1,87ºx1,25º
Met Office Hadley Center, United
HADGEM2-CC Collins et al. (2008)
Kingdom.
daily

Japan Agency for marine-Earth

Science and Technology
2,8ºx2,8º (JAMSTEC), Atmosphere and
MIROC-ESM-CHEM Ocean Research Institute (AORI), Watanabe et al. (2011)
daily and National Institute for
Environmental Studies (NIES),
Japan.

1,8ºx1,8º Raddatz et al. (2007);

Max-Planck Institute for Meteorology
MPI-ESM-MR
(MPI-M), Germany.
daily Marsland et al. (2003)

1,2ºx1,2º
Meteorological Research Institute
MRI-CGCM3 Yukimoto et al. (2011)
(MRI), Japan.
daily

2,5ºx1,9º Bentsen et al. (2012);

Norwegian Climate Centre (NCC),
NorESM1-M
Norway.
daily Iversen et al. (2012)

Tabla 1. Información sobre los nueve modelos climáticos pertenecía al CMIP5. Los modelos fueron suministrados por el
Programa de Diagnóstico e Intercomparación de Modelos Climáticos (PCMDI).

3.4.3. Denominaciones de Origen

De las 101 DO totales se ha trabajado finalmente con 54 DOs de ellas (figura 39). Se han
seleccionado aquellas DOs de las que se tienen datos históricos de temperatura y
precipitación procedentes de observatorios ubicados en el área denominada para cada
DO. Así mismo también se han quedado fuera del estudio algunas DOs que no estaban
incluidas en el mapa de la base de datos del Ministerio de Agricultura, Pesca y
Alimentación, MAPA, en el momento de su consulta y aquellas que a pesar de tener datos
meteorológicos estos no han pasado los filtros mínimos de calidad. Las DOs canarias se
han dejado fuera del estudio ya que debido a su ubicación geográfica, se rigen por
patrones atmosféricos diferentes a los utilizados en Península y Baleares y requieren de
un estudio propio considerando sus particularidades climáticas.

61
4. RESULTADOS

62
 Resultado 1
Proyecciones climáticas futuras de temperatura y de
episodios de olas de calor y de frío para Aragón
(España) en base a las salidas de modelos pertencientes
al CMIP5

63
 Projection of temperatures and heat and cold waves for Aragón (Spain)
using a two-step statistical downscaling of CMIP5 model outputs

Abstract

Heat- and cold-wave scenarios and temperature scenarios during the 21st century were
obtained for Aragón (Spain), using, for the first time, nine Earth System Models (ESM) and
two Representative Concentration Pathway (RCP) scenarios – RCP4.5 and RCP8.5 -
belonging to the 5th Coupled Model Intercomparison Project (CMIP5).

Local climate heat-wave scenarios show an increase of its mean intensity close to 2 ºC
(reaching temperatures of up to 38.8 ºC) and an average increase of the maximum
intensity of 3.6 ºC (temperature of up to 41.5 ºC) with respect to a historic period (1971–
2000) for the RCP8.5 scenario at the end of the century. The duration of heat waves will
increase by 7 days at the end of the century (total average duration of 12 days). The future
intensity and duration of cold-wave episodes will remain stable.

Local climate change scenarios for daily maximum temperatures show a gradual increase
throughout the 21st century. The greatest increases will occur during the summer at the
end of the century, reaching values of up to 7 ºC for the RCP 8.5 scenario. The minimum
temperature increases show similar behaviours to the maximum temperatures, but with
less marked increases (3 ºC and 5.6 ºC for the RCP4.5 and RCP8.5 scenarios respectively
in summer at the end of the century).

The highest temperatures and the intensity of the heat waves will be especially intense in
the Ebro Valley, the most populated area. In addition, the Pyrenees will suffer the longest
heat waves, especially at the end of the century, and the greatest increases in maximum
temperatures.

The downscaling of the CMIP5 models, offers accurate scenarios -both spatially and
temporally- of extreme temperatures and heat and cold waves, useful for decision-making
for local adaptation to climate change but also as a reference for other European regions.

1. Introduction
There is widespread agreement in the scientific literature that the most severe effects of
global warming will be related not only to a change in the mean climate, but especially to
an increase in the frequency and intensity of extreme events, such as floods, droughts and
heat waves as is reflected in the conclusions provided by the Intergovernmental Panel on
Climate Change (IPCC) and in independent studies (Cook et al. 2013). Heat waves (i.e., a
period of consecutive days with extreme hot temperatures) are one form of extreme
weather that is likely to increase in frequency, intensity and duration under the influence of
a changing climate during upcoming years (IPCC, 2013; Field et al., 2012). Extreme
temperature events can impact many aspects of human life, such as mortality, health,
comfort, agriculture and hydrology (Garcia-Herrera et al., 2005; Roldan et al., 2016). For
example, in 2003, an extraordinary ‘‘mega heat wave’’ occurred in Central Europe, which
had important consequences on society.

One of the main problems when evaluating extreme episodes of temperature (that is, cold
or heat waves) is that there are no universal definitions of these phenomena. Some
64
definitions use a fixed term (5 days > 30 ºC or 3 days > 35 ºC in the case of a heat wave);
some include humidity, spatial extent or accumulated heat ((Anderson and Bell, 2009;
Russo et al., 2017; Suparta and Yatim, 2017); some set their threshold according to a
specific percentile (Anderson and Bell, 2009; Kent et al., 2014; Smith et al., 2013); and
some are based on combined indexes of different meteorological variables, such as World
Meteorological Organization member guidelines (WMO, 2010).

Although large-scale and small-scale processes involved in the development of European

heat waves, such as interactions between land surface and atmosphere and the influence
of large-scale atmospheric circulation, are not fully understood (Vautard et al., 2013),
climate models could be a suitable tool for providing accurate heat-wave assessments for
future adaptation strategies.

Global climate models (GCMs) are fundamental tools for the study of future climate, and
they are able to provide data to estimate large-scale aspects of climate change, to drive
regional climate models or to be used directly by impact models. GCM outputs are also
used for projections of temperature extremes around the world (Dosio, 2017; Gross et al.,
2017; Jeong et al., 2016; Saeed et al., 2017; Sillmann et al., 2014).

Recently, efforts to reduce model uncertainty have led to a new generation of global climate
models called Earth system models (ESM) that integrate the individual parts of the climate
system (atmosphere, ocean, land and sea ice) and the exchange of energy and mass
between them (Knutti and Sedlacek, 2013). They also include chemical processes, land
use, plant and ocean ecology and an interactive carbon cycle, which enables integration of
biochemical processes into the models (Heavens et al. 2013). These models are the basis
of the fifth phase of the Coupled Model Intercomparison Project (CMIP5) (Taylor et al.,
2012), constituting the most current set of coordinated climate model experiments (Brands
et al., 2013; Carvalho et al., 2017; Chen et al., 2016; Perez et al., 2014)

The CMIP5 models and their predecessors have become an important resource for climate
scientists and others investigating possible future impacts. However, the spatial resolution
of these models remains too low for many impact assessments at a local scale, on which
policy and management decisions need to be made. These models are run at coarse
spatial resolutions, which is also a major limitation in projecting extreme events because
projections of climate extremes and simulations of heat-wave characteristics on regional
scales are very sensitive to spatial resolution and the interpretation of gridded outputs
(Chen et al., 2016).

The deficiency in scale is addressed through downscaling techniques involving either

statistical approaches or dynamic approaches. In the statistical approach, high-resolution
predictands (surface variables) are obtained by applying relationships that are identified
from the observed climate data to these predictands and to large-scale predictors of GCM
outputs (Asong et al., 2016; Benestad et al., 2007; Semenov and Stratonovitch, 2010). In
dynamical downscaling, a time-varying regional climate model (RCM) of atmospheric
boundary conditions is nested within the GCM ( Giorgi and Gutowski, 2015; Giorgi and
Mearns, 1991;). Both methods have been widely used and both have advantages and
disadvantages. The main advantages of statistical downscaling techniques are that they
are computationally inexpensive, they provide local information and they allow quantifying
the uncertainty associated to the downscaling process and to the climatic models.

65
Several studies have been carried out to explore present temperature extremes in Europe
(Soares et al., 2012) and in the Mediterranean region (Kuglitsch et al., 2010; Seubert et al.,
2014), and other studies have developed scenarios of temperature extremes in the same
areas (Estrella and Menzel, 2013; Hertig et al., 2010; Lavaysse et al., 2012). However,
none of them have applied the downscaling CMIP5 models.

Although many climate models have difficulties in properly reproducing climate extremes,
such as heat-wave conditions (Stegehuis et al., 2015), a few studies have reported on heat-
wave scenarios in Europe (Fischer and Schar, 2010; Schoetter et al., 2015), or located in
in the Mediterranean region (Zittis et al., 2016), in France (Planton et al., 2008), Italy
(Tomozeiu et al., 2007) or Finland (Kim et al., 2018). Schoetter (2015) and Kim (2018) are
one of the few studies that already uses the models from the IPPC5. As a part of the
European Coordinated Regional Downscaling Experiment (EUROCORDEX) (Giorgi,
2009), the EUROCORDEX initiative provides regional climate projections for Europe.
(Vautard et al., 2013) used the EUROCORDEX project to simulate heat waves at the
European regional scale, providing a downscaling of CMIP5 simulations. These European-
scale studies only featured a few locations in Spain. Several downscaling studies have
been specifically conducted on temperature in Spain (Brands et al., 2011b; Frias et al.,
2005; Frias et al., 2010; Hervada-Sala et al., 2000; Miro et al., 2016; Turco et al., 2014),
but very few studies have explored temperature extremes in Spain (Fernandez-Montes and
Rodrigo, 2012; Fonseca et al., 2016) and they do it in past periods (1925–2006 and 1986–
2005, respectively).

Only a few studies have developed temperature scenarios in the Aragón region
(northeastern Spain) (Buerger et al., 2007; Goncalves et al., 2014; Ribalaygua et al.,
2013a), but none of them developed temperature-extreme or heat-wave scenarios. Only
(Barrera-Escoda et al., 2014) developed scenarios of extreme temperatures in the Ebro
basin. That study showed that the projected increase in the number of tropical nights and
extreme temperatures could have a negative effect on human health and comfort
conditions. Other studies have also analysed the possible impacts of climate change in this
region on human health (Roldan et al., 2016). Both studies used the climate models
associated to the 4th assessment report of the IPCC.

Aragón is characterised by a highly complex topography that causes large climate

gradients. Downscaling techniques were required to capture these orographic features,
allowing for the areas that were most vulnerable to extreme changes to be identified.
Statistical downscaling is particularly recommended for areas with complex topography
(Kattenberg and Amer Meteorol, 1996). Nevertheless, downscaling of CMIP5 simulations
has not been used to date, and heat-wave scenarios cannot be found in this sensible area
of Spain.

Therefore, the objectives of this study were:

 Generate local heat- and cold-wave scenarios of the 21st century to downscale
GCMs from CMIP5 using a statistical methodology in the area of Aragón, Spain.

 In addition, generate new climate scenarios using CMIP5 models for Aragón to
simulate the future daily maximum and minimum temperatures in order to analyse
the differences with respect to the scenarios we generated in 2013 for the 21st

66
 century using the Special Report on Emissions Scenarios, SRES (IPCC, 2000)
previous to CMIP5.

Aragón, moreover, is a representative territory of different European climates, from the low-
land areas in the centre of Aragón (the Ebro River Basin) to the mountain regions in the
north (the Pyrenees) which makes it a good indicator of future European climate changes.
Finally, an original two-step analogue / regression statistical downscaling method
developed by the Climate Research Foundation was carried out.

The information used in the present study was based on the most current data available. It
was useful for identifying the areas that were most vulnerable to extreme temperature
changes in Aragón, and for helping decision makers to design mitigation and adaptation
strategies in response to territorial environmental impacts derived from climate change.

2. Material and Methods

2.1 Study area and datasets

2.1.1 Study area

The present study was carried out in the region of Aragón (Fig.R1.1), which is located in
the northeast part of the Iberian Peninsula in Spain and which has an area of approximately
47,720 km2 (340 km length and 240 km width). Because of its location, Aragón falls within
the Western Mediterranean climate area, with cool winters and hot, dry summers. However,
the extreme altitude differences of over 3,000 m between the plains (the Ebro River Basin)
and the mountains (the Pyrenees), together with the specific topography of the Ebro River
Basin and the mountain chains (the Pyrenees to the north of the basin and the Iberian
Mountains to the south), modify the local climate. As a result, the climate characteristics of
the area are somewhat different from a standard Western Mediterranean climate, and they
typically consist of dryness of the land along the banks of the Ebro River, random rain
patterns, high thermal contrast between winter and summer as a consequence of the strong
continental characteristics of the region and the typical northeast “mistral” winds, which are
frequent in the region.

2.1.2. Surface observation dataset (predictands)

The observational dataset used in the present study consists of a time series of the daily
maximum and minimum temperatures distributed, quite homogeneously, throughout the
Aragón territory (Fig.R1.1). This dataset it is the same as the one used in a previous study
(Ribalaygua et al., 2013a) in order to facilitate comparison of the results.

Data were obtained from the extensive network of instrumental observatories owned by the
Spanish Meteorological Agency (AEMET) (http://www.aemet.es). The stations are subject
to strict data quality control, which includes detecting and correcting multiple aberrant
points in the long-term climate time series, covering the gaps, assessing the homogeneity
of the climate series and correcting any inhomogeneities found. This process was carried
out by the Aragón Government (López et al., 2007) to guarantee the quality of the data,
stations with a large number of data gaps or without at least 15 years of daily records were
discarded. Second, we analysed the dataset for the presence of extreme data points that
could be considered suspect observations or transcription errors (we considered data
points to be extreme if they were separated from the monthly average by more than four
times the value of the standard deviation). Third, we verified that the data did not contain
67
inhomogeneities that could introduce bias into any records that were not related to the
climate, (Ribalaygua et al. 2013a)

Figure R1.1. Location of the study Area: Aragon (Spain) in Europe. Points indicate the stations used in the study. a)
Stations of temperature (103) used in the generation of climate regional scenarios of maximum and minimum temperature
(verification, validation and scenarios). b) Stations used exclusively on the generation of heat waves (7, red), used
exclusively on the generation of cold waves (7, blue) and those used in both extreme events (64, green). “(Map source:
OpenStreetMap)”

The first set of data (composed of 103 stations) was used to analyse and produce the
regional climate scenarios of maximum and minimum temperatures (Fig. R1.1a). Of these
103 stations, those that had at least 75% of maximum temperature data collected from
June to September (i.e., the reference period for the calculation of heat waves) between
1980 and 2000 and stations that had at least 75% of minimum temperature data collected
from November to April (that is, the reference period for the calculation of cold waves)
between 1980 and 2000 perform a second set of data selected for the study of extreme
events. In both cases, the number of observatories was 71 (Fig.R1.1b).

2.1.3. Atmospheric dataset (predictors)

2.1.3.1 Reanalysis

As a reference dataset, we used the reanalysis from the European Centre for Medium-
Range Weather Forecasts (ECMWF ERA-40;
http://www.ecmwf.int/research/era/do/get/era-40) (Uppala et al., 2005) for the period 1958–
2000. For the atmospheric window, we used geographical limits from 31.5°N to 55.1°N
latitude and from 27.0°W to 14.6°E longitude, covering not only the geographic area under
study, but also the surrounding atmosphere areas, which exert a meteorological influence
all over the Iberian Peninsula (Ribalaygua et al., 2013a).
68
For verification of the methodology, it was necessary to reduce the temporal (six hourly)
and spatial (125 km) scale of the reanalysis to that of the different climate models in order
to compare both, ERA-40 and the climate model simulations (Ribalaygua et al., 2013a;
Ribalaygua et al., 2013b).

2.1.3.2. Climate model data

In the present study, we worked with a set of nine climate models that belong to the CMIP5
and that were provided by the Program for Climate Model Diagnosis and Intercomparison
(PCMDI) archives (Table R1.1). The number of models chosen was limited to the
availability of data with daily time frequencies. All of the models were ESMs.

To generate the model predictor fields, we took the daily values of several large-scale fields
of interest (that is, geopotential height, specific humidity and wind) at different pressure
levels for the nine ESMs showed in the table R1 associated with the CMIP5 (Taylor et al.,
2012). Moreover, we used an historical experiment (Taylor et al., 2012) which generated
simulations of the past 20th century and which was useful for the evaluation of model
performance against the present climate. Finally, the representative concentration pathway
(RCP) families were considered (Moss et al., 2010), specifically the RCP8.5 ‘high’ scenario
and the RCP4.5 ‘intermediate’ scenario, both of which corresponded to different possible
ranges of radiative forcing reached in the year 2100 with respect to values of the pre-
industrial era (4.5 and 8.5 W/m2, respectively).

2.2 Methodologies

2.2.1 Description of the downscaling methodology

The methodology of statistical downscaling used in this study was chosen according to
three main advantages:

1) Allows easier quantification of the main uncertainties associated with the generation
of future climate scenarios (van der Linden and Mitchell, 2009).

2) It is a key for the achievement of climate simulations that are consistent with
observations which, in turn, are physically coherent (Ribalaygua et al., 2013b).

3) Provides local detail, which is useful information because nearby data points in
space can evolve under different future climate conditions (Ribalaygua et al.,
2013b).

A two-step analogue / regression statistical downscaling method that was developed by

the Climate Research Foundation (FIC) (Ribalaygua et al., 2013b) was applied to obtain
future scenarios of the maximum and minimum temperatures at a local scale in the region
of Aragón. This methodology consisted of a two-step analogue / regression statistical
method, which has been used in national and international projects with good verification
results. In the present paper, we present a summary of the two-step method (more details
can be found in Ribalaygua et al., 2013b).

69
 2ºx2,5º
National Oceanic and Atmospheric
GFDL-ESM2M Dunne et al. (2012)
Administration (NOAA), E.E.U.U.
daily

Canadian Centre for Climate Modeling

2,8ºx2,8º
and Analysis (CC-CMA),
CanESM2 Chylek et al. (2011)
daily
Canadá.

1,4ºx1,4º CNRM (Centre National de Recherches

CNRM-CM5 Meteorologiques), Meteo-France, Voldoire et al. (2013)
daily Francia.

1,4ºx1,4º
Beijing Climate Center (BCC), China
BCC-CSM1-1 Xiao-Ge et al. (2013)
Meteorological Administration, China.
daily

1,87ºx1,25º
Met Office Hadley Center, United
HADGEM2-CC Collins et al. (2008)
Kingdom.
daily

Japan Agency for marine-Earth Science

and Technology (JAMSTEC),
2,8ºx2,8º
Atmosphere and Ocean Research
MIROC-ESM-CHEM Watanabe et al. (2011)
Institute (AORI), and National Institute
daily
for Environmental Studies (NIES),
Japan.

1,8ºx1,8º Raddatz et al. (2007);

Max-Planck Institute for Meteorology
MPI-ESM-MR
(MPI-M), Germany.
daily Marsland et al. (2003)

1,2ºx1,2º
Meteorological Research Institute
MRI-CGCM3 Yukimoto et al. (2011)
(MRI), Japan.
daily

2,5ºx1,9º Bentsen et al. (2012);

Norwegian Climate Centre (NCC),
NorESM1-M
Norway.
daily Iversen et al. (2013)

Table R1.1. Information about the nine climate models belonged to the 5 Coupled Model Intercomparison Project (CMIP5)
corresponding to the fifth report of the IPCC. Models were supplied by the Program for Climate Model Diagnosis and
Intercomparison (PCMDI) archives.

The first step was analogous stratification. An analogue method was applied based on the
hypothesis that ‘analogue’ atmospheric patterns (predictors) should cause analogue local
effects (predictands), which means that the number of days that were most similar to the
day to be downscaled were selected (Benestad et al., 2007; Zorita and von Storch, 1999 ).
The similarity between any two days was measured using a pseudo-Euclidean distance
between the large-scale fields used as predictors. For each predictor, the weighted
Euclidean distance was calculated and standardised by substituting it with the closest
percentile of a reference population of weighted Euclidean distances for that predictor. This
method is a good method for reproducing nonlinear relationships between predictors and
the predictands, but it could not be used to simulate values outside of the range of observed
70
values ((Imbert and Benestad, 2005). In order to overcome this problem and to obtain a
better simulation, a second step was required.

The second step focuses on temperature. To determine the temperature, a multiple linear
regression analysis for the selected number of most analogous days was performed for
each station and for each problem day. From a group of potential predictors, the linear
regression selected those with the highest correlation, using a forward and backward
stepwise approach.

About the accuracy in simulating temperatures of our downscaling technique, in our

previous work about climate change in Aragón (Ribalayagua et al., 2013a) we achieved an
average bias below 0.1 ºC” (Ribalaygua et al., 2013a).

2.2.2. Validation and climate simulation of temperature

In order to determine the ability of each ESM to simulate the predictor fields, absolute and
relative temperatures from the downscaled ESM simulation of the historical experiment was
compared with the downscaled ERA-40 simulation (previously verified against the
observations) during a common historical period (1958–2000).

Due to the characteristics of the ESMs, the validation process presents certain limitations.
First, as climate models do not reproduce day-to-day meteorology, validation cannot be
performed on a daily scale and it must be done using climate statistics over long periods
of time, resulting in a loss of information on climate variability. Second, we could not
compare the climate characteristics obtained from the ESM historical simulation with those
obtained from the observations because the latter had missing data and large gaps;
therefore, we had to compare ESMs simulations with simulations from the reanlaysis of a
dataset (previously validated in the verification process). Both sources of error should be
considered in the final uncertainty analysis.

As error measures, bias and standard deviation were analysed on a seasonal scale for
both maximum and minimum temperatures. An ensemble strategy was used to quantify
the uncertainties inherent in future climate projections (IPCC, 2013). For each scenario
(i.e., RCP4.5 and RCP8.5), an ensemble of the approved downscaled ESMs was used to
estimate the mean change (compared to 1976-2005) and to quantify the main contributions
to the uncertainty.

Future local climate scenarios for maximum and minimum temperature, for nine GCMs and
two RCPs have been produced at the daily scale. To draw the temperature maps, we used
Thin Plate Spline regression (TPS) from the R-Package "fields" (Nychka et al., 2015).

2.2.3. Definition of hot- and cold-wave episodes

There are many and varied terms used to define a heat wave. For example, the AEMET
defines a heat wave as an “episode of at least three consecutive days, in which at least
10% of the stations considered record maximums above the 95% percentile of their series
of maxima daily temperatures of the months of July and August of the period 1971–2000.”
The WMO defines a heat wave as an extreme event with marked warmed of the air or the
invasion of very warm air over a large area, it usually last from a few days to a few weeks.”
The IPCC defines a heat wave as “a period [of] abnormally and uncomfortably hot weather”.
Those definitions are not the only ones accepted in the scientific literature. Some heat-
wave definitions have been used to identify heat waves in a time series of temperature data
71
( Smith et al., 2013; ), and the choice of the heat wave definition can influence both
projected heat-wave trends (Smith et al., 2013) and estimates of health risks during events
(Anderson and Bell, 2009; Chen, et al. 2015, Kent, et al. 2014).

According to the recommendations of the WMO (WMO, 2010), a practical and qualitative
definition of a heat wave must consider marked and unusual hot weather over a region
during at least two consecutive days in the hot period of the year, based on local climate
conditions, with thermal conditions recorded above given thresholds. On the basis of these
recommendations and the characteristics of the Aragón climatology, we defined a heat
wave as follows: at least three consecutive days with a maximum temperature above the
95th percentile of the maximum temperature series and calculated between the months of
June to September during the period 1980–2000.

In broad terms, a cold wave can be defined as a meteorological event characterised by a

sharp drop of air temperature near the surface, leading to extremely low values. According
to the IPCC 2007, (Parry et al., 2007) a cold wave is an event that often causes problems
and severe impacts on the population, especially in northern altitudes. However, there is
still a lack of a clear and consistent definition for cold-wave events in the world.

In the WMO’s Meteoterm vocabulary, a cold wave is defined as "marked cooling of the air,
or the invasion of very cold air, over a large area." For the AEMET, a cold wave is “an
episode of at least three consecutive days, in which at least 10% of the considered stations
register minimums below the 5% percentile of their series of minima daily temperatures for
the months of January and February of the period 1971–2000”. The WMO’s members
guideline define a cold wave as “marked and unusual cold weather characterised by a
sharp and significant drop of air temperatures near the surface (max, min and daily
average) over a large area and persisting below certain thresholds for at least two
consecutive days during the cold season”.

Considering the climate characteristics of Aragón, we defined a cold wave as follows: at

least three consecutive days with a minimum temperature below the fifth percentile of a
minimum temperature series and calculated between the months of November to April
during the period 1980–2000.

To better identify a heat or cold wave, we evaluated average duration, maximum intensity
and average intensity as it is recommended by the WMO in the Guidelines on the definition
and monitoring of extreme weather and Climate Events. The average duration refers to the
average number of days the heat wave lasted. The average intensity represents the
average value of the temperature during the heat wave. The maximum intensity is equal to
the maximum value that the temperature reached during the heat wave.

2.2.4. Verification and simulation of heat and cold waves

In order to assess the capacity of the downscaling methodology to simulate heat and cold
waves observed in the past, we evaluated how the downscaling methodology simulated
the 95th percentile of maximum temperatures and the 5th percentile of minimum
temperatures, which are the thresholds that indicate the existence of an episode of a heat
and cold wave, respectively. In addition, we analysed the intensity of the wave (average
and maximum) and the average duration, comparing the heat and cold waves calculated
from the simulated ERA-40 temperature series with those obtained from the observed
series. Verification of the maximum and minimum temperatures can be seen in a previous
72
study by Ribalaygua (Ribalaygua et al., 2013a). The statistical measures used in the
verification processes were the bias, the standard deviation and the Pearson correlation
(Murray et al., 2009) at the daily scale. The statistical measures were calculated using R-
cran package computing software (R Development Core Team 2010).

From the ESM simulated temperature series (nine ESMs and two RCPs), we determined
the heat- and cold-wave episodes that were expected in Aragón during the upcoming
decades of the 21st century. The heat- and cold-wave scenarios were compared to a
historical period (1976–2005) to analyse the future changes with respect to the actual
situation of these extreme events.

3. Results

3.1. Validation of the CMIP5 model statistical downscaling to predict temperatures

The seasonal bias that result from a comparison between the ERA-40 temperature
simulations and the historical temperature simulations for each ESM for a common period
(1958-2000) are shown in Fig. R1.2 for absolute temperatures (that is, the difference
between simulated Historical ESM data and simulated ERA-40 data). For both the
maximum and minimum temperatures, the obtained bias was around tenths of a degree in
all months, so they were very close to zero. The error was not above half of a degree for
any of the cases. Therefore, the results showed that the ESMs were capable of adequately
simulating both the maximum and the minimum temperatures on annual and seasonal
scales.

Figure R1.2. Validation of the GCMs used for the simulation of temperatures. Absolute Bias for a) maximum and c)
minimum temperature and absolute standard deviations for b) maximum and d) minimum temperature, between the results
obtained by downscaling the Historical scenario of each GCM used in the study with those obtained by downscaling the
reanalysis ERA-40 for a common period (1958-2000). There are four boxplots, representing a season of the year (winter,
spring, summer and autumn), for each GCM.
73
3.2. Local climate scenarios to predict future temperatures

Fig. R1.3 and R1.4 show local climate-change scenarios for future daily maximum and
minimum temperatures respectively, which have been predicted on the basis of the nine
models (see Table R1.1). Fig. R1.3 and 4 allowed for the establishment of a general view
of the changes expected in Aragón. In addition, Fig. R1.5 and R1.6 represent the summer
maximum and winter minimum temperature changes expected for the periods 2041–2070
and 2071-2100 as presentation of mid- and end-century expected changes, which were
calculated according to the scenarios RCP4.5 and RCP8.5.

Figure R1.3. For the four seasons (winter, spring, summer and autumn) simulated maximum temperature for the twenty-
first century displayed as absolute increase against the value simulated for the 1976–2005 Historical period. The vertical
dotted line marks the end of the Historical data (2005). Data grouped for every RCP simulation of every global climate
model used and for the last 30 years of every station. The ensemble median (solid lines) and the 10th–90th percentile
(shaded areas) values are displayed

Expected changes in the average temperature for the rest of the seasons of the year can
be found in the supplementary material for maximum temperatures increase (Fig. S1.1–
S1.3) and minimum temperatures increase (Fig S1.4-S1.6). In addition, Fig S1.7- S1.9
showed the maximum and S1.10-S1.12, minimum temperatures expected in this century.

Fig.R1.3 shows a gradual increase in the maximum temperatures throughout the 21st
century. During the mid-century, the least significant changes are expected during the
winter and springs months (around 1.5 and 1.7 ºC for the RCP4.5 and RCP8.5,
respectively), and the most significant changes are expected for the summer months, with
74
values of 2.1 ºC for the RCP4.5 and 2.7 ºC for the RCP8.5. At the end of the century, the
expected changes are more marked and differ notably between the two RCPs. During
winter and spring months, the expected values are around 2.3 ºC and 2.4 ºC for the
RPC4.5 and 4.3 ºC and 4.7 ºC for the RCP8.5, respectively. The greatest changes are
expected for summer months, reaching values of 7 ºC for the RCP8.5 and 3.6 ºC for the
RCP4.5.

Figure R1.4. For the four seasons (winter, spring, summer and autumn) simulated minimum temperature for the twenty-
first century displayed as absolute increase against the value simulated for the 1976–2005 Historical period. The vertical
dotted line marks the end of the Historical data (2005). Data grouped for every RCP simulation of every global climate
model used and for the last 30 years of every station. The ensemble median (solid lines) and the 10th–90th percentile
(shaded areas) values are displayed

The areas most affected by the maximum temperature increases are expected to be the
Pyrenees and the areas in the southwest and north of Aragón during all the 21ª century for
the months of summer, spring and autumn, especially in the RCP8.5 scenario (Fig. R1.5).
In contrast, in the winter months, the highest temperature increases at the end of the
century are expected in the Pyrenees area and the Ebro Valley (Fig S1.1).

Scenarios of maximum temperatures of the 21st century (Fig R1.5 and Fig S1.7-S1.9) point
that expected maximum temperatures are not proportional to the expected increases
shown before (Fig R1.5 and Fig S1.1-S1.3).

75
 Figure R1.5. Geographical representation of the expected changes of maximum temperature in summer for the periods
2041-2070 and 2071-2100 respect to the reference Historical Period (1976-2005). Both emissions scenarios are
represented: RCP4.5 (figures b and c) and RCP8.5 (figures d and e). Figure 5a represents the Historical absolute
temperature for the period 1976-2005.

The evolution of the minimum temperature increases throughout the 21st century
demonstrate similar behaviours to the maximum temperatures, but with less marked
increases, especially at the end of the century (Fig. R1.6 and S1.4-S1.6). By the middle of
the century, all of the seasons of the year, except for the summer, are expected to show
increases in the minimum temperature of less than 2.0 ºC, and even in the spring, the
increases are expected to be close to 1.0–1.2 ºC. The summer months are expected to
show the greatest increases in minimum temperatures, varying from 1.8 ºC for the RCP4.5
to 2.3 ºC for the RCP8.5. At the end of the century, increases in the minimum temperatures
during the summer months are expected to range from 3.0 ºC (RCP4.5) to 5.6 ºC (RCP8.5.
Winter months are expected to show an increase between 2.8 ºC and 4.1 ºC, spring
months between 2.1 ºC and 4 ºC and autumn months between 2.6 ºC and 5 ºC for the
RCP4.5 and RCP8.5 scenarios, respectively.

The increases in the minimum temperature in are expected to be quite homogeneous

throughout the region and the seasons according to the RCP4.5. In contrast, according to
RCP8.5 (Fig. R1.6c, d), these changes will be more pronounced in the Ebro Valley at the
end of the 21st century. The highest minimum temperatures are expected to be recorded
in the central zone of Aragón during all seasons of the year (see supporting information,
S1.10-S1.12).

76
Figure R1.6. Geographical representation of the expected changes of minimum temperature in winter for the periods 2041-
2070 and 2071-2100 respect to the reference Historical Period (1976-2005). Both emissions scenarios are represented:
RCP4.5 (up, figures b and c and RCP8.5 (figures d and e). Figure 6a represents the Historical absolute temperature for the
period 1976-2005.

3.3. Verification of hot- and cold-wave episodes

To verify the simulation of both of heat waves and cold waves, we compared the wave
episodes obtained from ERA-40 simulations against the episodes recorded by the
observations.

The first step was to verify that the methodology correctly simulated both percentiles, the
95th percentile of the maximum temperature and the 5th percentile of the minimum
temperature. Both cases have been well simulated by applying the downscaling
methodology to the ERA-40 series (Fig. R1.7). Data from the simulated ERA-40 series and
the observed series were very close, and the error committed by simulating the ERA-40
series respect to the observed one was very low, especially in the 95th percentile. For
example, we found error values around 0.36 ºC for summer months and around 0.69 ºC
for the winter months in the 95th percentile, and 0.6 ºC for summer months and around
0.68 ºC for the winter months in the 5th percentile.

Regarding the verification of the identification parameters of a heat wave, Fig. R1.8 shows
the results obtained for the duration of a heat wave, its temporary occurrence and its
average intensity (maximum intensity is not shown here, but it was also taken into account
in the present study and can be seen in Support Information (S1.13)).

77
 Figure R1.7. Results of the verification process obtained for a) the 95th percentile of maximum temperature and b) the 05th
percentile of minimum temperature comparing the results obtained by downscaling the reanalysis ERA-40 (black) along the
observations (red) for a common (1980-2000). The solid lines represent the median and the shaded areas the 10-90th
percentile of the values.

As can be seen in Fig. R1.8a and b the average intensity of observed heat waves and ERA-
40 simulations are very similar in the 71 observatories. We found a high correlation
between the average intensity of a heat wave and the results from the observed and the
simulated ERA-40 scenarios (p = 0.9889). A similarly high correlation was obtained in terms
of the maximum intensity of a heat wave (p= 0.9883, not shown here).

Regarding to the average duration of heat waves episodes, Fig. R1.8c and d showed a
similar pattern of temporal and spatial distribution and a strong correlation (p = 0.8976).
For example, both datasets showed a clear signal of heat waves during the years 1982,
1987, 1990, 1994 and 1998, and they both showed no heat wave during the years 1986,
1987, 1996 and 1997.

78
 Figure R1.8. Results of the verification process for heat waves obtained comparing the simulated heat waves for the
reanalysis Era-40 along the observed heat waves for a common period (1980-2000). a) Average intensity of the heat
waves registered in the 71 stations used in the study against b) the average intensity obtained by downscaling of the
reanalysis ERA-40 in these stations for the period 1980-2000. The color of each pixel shows the average intensity that
corresponds to each heat wave ( ºC), each column represents one year of the considered period and each row one station.
The upper row shows the mean of the average intensity from the whole observatories. Down, c) and d) same as a) and b)
but for the duration of the heat waves. The color of each pixel shows the duration corresponding to each heat wave
(number of days).

Similar results are present in Fig. R1.9 for cold waves. In this case, the correlation obtained
between the average duration of the observed cold waves (Fig R1.9c) and those simulated
for ERA-40 (Fig R1.9d) was p = 0.8593, which was lower than the p-value obtained for heat
waves. The correlations of the intensities between both groups of data were strong (p =
0.8849 and p = 0.792 for maximum intensity and average intensity, respectively), but lower
than those obtained for the heat waves. The cold-wave pattern (that is, duration, average
intensities and maximum intensities) were very similar in both datasets, marking episodes
of cold waves during the same time periods.

79
 Figure R1. 9. Results of the verification process for cold waves obtained comparing the simulated cold waves for the
reanalysis Era-40 along the observed cold waves for a common period (1980-2000). a) Average intensity of the heat waves
registered in the 71 stations used in the study against b) the average intensity obtained by downscaling of the reanalysis
ERA-40 in these stations for the period 1980-2000. The color of each pixel shows the average intensity that corresponds to
each cold wave ( ºC), each column represents one year of the considered period and each row one station. The upper row
shows the mean of the average intensity from the whole observatories. Down, c) and d) same as a) and b) but for the
duration of the cold wave. The color of each pixel shows the duration corresponding to each cold wave (number of days)

3.4. Local future climate scenarios of hot- and cold-waves episodes

Fig. R1.10 and R1.11 show the climate scenarios for heat waves in Aragón according to
the RCP4.5 and the RCP8.5scenarios, respectively. The figures show the temporal
evolution (from historical data to 2100) for three main features of the heat waves: average
intensity (first row), maximum intensity (second row) and average duration (third row). The
temporal periods analysed were 1976–2005 (historical; first column), 2041–2070 (second
column) and 2071–2100 (third column). These periods were chosen as representative of
the present, mid-century and end-century periods. Fig. R1.12 and R1.13 show the same
information but for cold waves.

In addition, in Support information heat and cold waves are represented as increments in
temperature for heat waves (S1.15 and S1.16 for RCP4.5 and the RCP8.5 scenarios,
respectively) and for cold waves (S1.17 and S1.18 for RCP4.5 and the RCP8.5 scenarios,
respectively).

80
 Figure R1.10. Geographical representation of the expected evolution of the heat waves for Aragon in the periods 2041-
2070 and 2071-2100 compared to the reference Historical Period (1971-2000) in terms of absolute values according to the
RCP4.5. The rows show the three parameters analyzed in the study (average intensity, maximum intensity and duration)
and the columns the three temporal periods (Historical, 2041-2070 and 2071-2100). The maps are generated by
interpolating the available stations over the entire territory.

During the coming decades, an increase in the duration of heat waves is expected.
According to the RPC4.5, sharp increases in the average intensity of heat waves are not
expected with respect to the values corresponding to the historical period (an average of
approximately 0.8 ºC). This means that a mean intensity of 37.6 ºC is expected at mid-
century and 37.8 ºC at the end of the century. According to the RCP8.5, the average
intensity would increase an average of approximately 1–1.2 ºC during the middle of the
century (reaching 38 ºC) and close to 2 ºC at the end of the century (reaching 38.8 ºC).

In the case of the maximum intensity of the heat waves, the average increases with respect
to the expected historical period were somewhat more pronounced than those expected
for the average intensity, especially in the case of RCP8.5. According to RCP4.5, average
increases of 1.4 ºC and 1.7 ºC (reaching 39.3 ºC and 39.6 ºC) are expected in the
maximum intensity during the middle and end of the century, respectively. In the case of
RCP8.5, those average increases are expected to rise to 2 ºC and 3.6 ºC (reaching 39.9
ºC and 41.5 ºC) during the middle and end of the century, respectively.
81
 Figure R1.11. Geographical representation of the expected evolution of the heat waves for Aragon in the periods 2041-
2070 and 2071-2100 compared to the reference Historical Period (1971-2000) in terms of absolute values according to the
RCP8.5. The rows show the three parameters analyzed in the study (average intensity, maximum intensity and duration)
and the columns the three temporal periods (Historical, 2041-2070 and 2071-2100). The maps are generated by
interpolating the available stations over the entire territory.

Note that for the average (maximum) intensity, there was a difference between
observatories, such that in some of them the intensity could reach 46 ºC (48 ºC), while in
other parts of the region it would not exceed 32 ºC (34 ºC) according to the RCP8.5.

The average and maximum intensity of the heat waves will be especially intense in the
Ebro Valley in the middle of the century, and it will spread throughout the entire territory
toward the end of the century, except for the Pyrenean area further north.

According to RCP4.5, an average increase in the number of days that lasts a heat waves
(duration) compared to the historical period of about 2 days is expected during mid-century
(2041-2070), with a maximum increase of 11 days, which would mean that the average
duration of heat waves it would be around 7 days, with durations that vary between 4 and
18 days according to the station. At the end of the century, very sharp variations are not
expected with respect to half a century. Results from the RCP8.5 predict that the average
duration of heat waves will increase by 3 days in the middle of the century and by 7 days

82
at the end of the century, which means that heat waves would have an average duration
of between 8 and 12 days, respectively. Note that according to the RCP8.5 scenario, some
observatories are expected to experience sharp increases during the historic period of up
to 16 and 44 days for the period 2041-2070 and 2071-2100 respectively, so there would
be observatories where the consecutive days that exceed the “hot-day” threshold would be
23 and 52 days, for the same periods.

In this case, the Pyrenees to the north and the Iberian to the south of Aragón are the zones
that would suffer longer heat waves; although, as we have seen, they would be the least
affected by the intensity of the expected heat waves.

Figure R1.12. Geographical representation of the expected evolution of the cold waves for Aragon in the periods 2041-
2070 and 2071-2100 compared to the reference Historical Period (1971-2000) in terms of absolute values according to the
RCP4.5. The rows show the three parameters analyzed in the study (average intensity, maximum intensity and duration)
and the columns the three temporal periods (Historical, 2041-2070 and 20712100). The maps are generated interpolating
the available observatories over the entire territory.

On the other hand, the episodes of cold waves would not change either the average or
maximum intensity values (see Fig. R1.12 and Fig. S1.17 for the RCP4.5 scenario and Fig.
R1.13 and Fig. S1.18 for the RCP8.5). The variations of average intensity with respect to
the historic period are, on average, ± 0.5 ºC, and in some observatories they could reach
± 1.0–1.5 ºC, independent of the RCP.

83
A similar pattern is expected for the maximum intensity of the cold waves (Fig. R1.12b and
R1.13b), with few variations with respect to the historical period and similar patterns to
those expected for the mean intensity. The north and the south of the territory are the
regions that are expected to be the most affected by the intensity of the cold waves.

During the upcoming decades, an increase in the duration of cold-wave episodes is not
expected. Both RCP4.5 (Fig. R1.12h and Fig. S1.17h) and RCP8.5 (Fig. R1.13h and Fig.
S1.18h) predict that by the middle of the century, the average duration of cold waves will
remain around 4 to 5 days, with episodes of 3 to 7 days in the middle of the century and 11
days at the end of the century. This would mean that the average duration of cold waves
would not be modified with respect to the historical period, unlike what is expected in the
case of heat waves, and in the most severe cases, the increase would be about three days
at the end of the century. No differences of this pattern are expected in the territory.

Figure R1.13. Geographical representation of the expected evolution of the cold waves for Aragon in the periods 2041-
2070 and 2071-2100 compared to the reference Historical Period (1971-2000) in terms of absolute values according to the
RCP8.5. The rows show the three parameters analyzed in the study (average intensity, maximum intensity and duration)
and the columns the three temporal periods (Historical, 2041-2070 and 20712100). The maps are generated interpolating
the available observatories over the entire territory.

4. Discussion

In the present study, heat- and cold-wave scenarios for the 21st century were obtained for
the first time in the region of Aragón, Spain, using new ESMs belonging to the CMIP5. In
84
addition, the climate scenarios for temperatures in this region were reconstructed with the
new ESMs in order to assess differences with respect to the models used prior to CMIP5.
These results offer one of the best snapshots of future climate change based on currently
available data on the risks that extreme temperatures can cause in the Aragón region, both
spatially and temporally.

4.1. Considerations of the temperature scenarios

The new models used in the present study to obtain climate scenarios in Aragón are ESMs,
not climate models. The improvements incorporated in the ESMs allow for greater precision
in the simulation of climate variables, which provides evidence that the results of the
validation are good.

The validation of the ESMs was good for both maximum and minimum temperatures. The
differences between seasonal means were practically negligible (below a few tenths of a
degree in temperature), and the differences between seasonal standard deviations were
almost always below the reference values for every ESM (the standard deviation of the
downscaled reanalysis data).

The validation results obtained in the present study show lower error rates than those
obtained by the scenarios described in the fourth IPCC report and presented in previous
studies (Ribalaygua et al., 2013a). In that case, the error was 1 ºC, and even in some
models (for example, the CNCM3 model in the summer), it was around 1.5 ºC; in contrast,
the results presented here did not reach such high values. Likewise, the values of the
standard deviation are lower than those obtained in the previous study (5–10% in the
present study vs. 15–20% in a previous study (Ribalaygua et al., 2013a).

Another key point that justifies the generation of new climate scenarios for Aragón is the
current designation of climate scenarios provided by IPCC experts—that is, RCPs.
Therefore, obtaining local climate scenarios that are based on the most current information
is a priority.

The results obtained in the present study, both in the validation process and in the
verification process, allow for the maximum and minimum temperature scenarios to be
possibly used in subsequent studies of extreme events and bioclimatic variables. Future
climate scenarios (maximum and minimum temperatures) show an evolution toward
warmer climates throughout the Aragón region. These results agree with those obtained
from the CMIP5 models by the IPCC (IPCC, 2013) and by the AEMET (www.Aemet.es).
Previous data collected by other authors and assembled by the IPCC (IPCC, 2013) expect
that the greatest temperature increases will occur during the summer months. The results
of the present study showed that increases in temperature (both maximum and minimum)
were less marked than those published by the AEMET (for example, our results showed
changes of 3.6 and 7 ºC at the end of the century, while the AEMET results predicted
changes of 4.1 and 8.2 ºC under RPC4.5 and RCP8.5 scenarios, respectively), although
both our results and the AEMET results suggest that the summer months will be the time
during which the most abrupt changes in temperature will be observed, while the winter
months will be the time during which the least abrupt changes will be observed.

Results presented by the IPCC came from the ESMs, so they do not have the added value
of applying downscaling techniques to obtain results at a local scale. On the other hand,
one of the advantages of working with climate-change scenarios is to have the widest range
85
of future projections (using the greatest possible number of climate scenarios and
downscaling methodologies). Therefore, the scenarios generated by the AEMET are
complementary to those presented here because different statistical downscaling
methodologies, as well as different ESMs, have been applied.

Similar processes were carried out in the generation of climate scenarios for Spain
published by the AEMET (www.Aemet.es) and in the EsTcena project (Brands et al., 2013;
Brands et al., 2011a). Results that were generated by different centres were presented on
the basis of different methodologies, including the one used in the present study.

We compared the new scenarios that were based on the models of the fifth report of the
IPCC with those published by Ribalaygua (Ribalaygua et al., 2013a), which correspond to
the fourth report of the IPCC. We observed that the most recent scenarios trended toward
higher temperatures than what was expected in (Ribalaygua et al., 2013a), with greater
increases occurring during summer months compared to winter months. For example, for
the summer months, the new results expect that for the period 2071–2100, the maximum
temperature in the most extreme case will reach up to 7 ºC, while those presented by
Ribalaygua (Ribalaygua et al., 2013a) for the same period were estimated to reach 5 ºC.
However, both studies agreed that the northwest and southwest regions would be the most
affected by these variations.

In both cases, maximum and minimum temperatures, in view of the validation results and
in the face of the same reanalysis of ERA-40, it is justified that the scenarios generated for
the fifth IPCC report are more precise than those that already exist for the Aragón region.
Having a range of climate-scenario projections in Aragón facilitates the evaluation of
extreme phenomena of interest, such as heat and cold waves, which have strong impacts
in the region.

4.2. Consideration about the simulation of heat and cold waves

Our results allow for the simulation of episodes of heat and cold waves throughout the 21st
century and prove the validity of the simulations, as is evident by the good verification and
validation results. As the methodology is able to adequately simulate the 95 th percentile of
maximum temperatures and the 5th percentile of minimum temperatures, it is assumed that
the thresholds that define the heat or cold waves are being simulated in an appropriate
way.

The downscaling methodology that was used was able to successfully simulate heat and
cold waves reported by the observations, as shown by the results of the verification
process. The ESMs were also successful at simulating the maximum and minimum
temperatures during the validation process. However, in the verification process of heat
and cold waves, were identified some limitations that must be considered in the simulation
of future scenarios:

1) There were gaps in the observed series. The lack of data in the observed series reduced
the length of observed data compared to the reanalysed ERA-40 series (that is, the full
series without gaps) for the same period of time. Therefore, the number of days included
in the 95th percentile (or 5th percentile) of the maximum (or minimum) temperature was
greater in the ERA-40 series compared to the observed series.

2) In general, downscaling techniques tend to soften the simulated temperature series with
respect to the observed values, which implies that certain episodes of hot or cold waves
86
are softened in their duration or intensity compared to the observed data and would
therefore be classified differently. In the present study, it was appreciated that the average
and maximum intensities of the observed cold and heat waves was greater than that
simulated for the ERA-40 reanalysis.

In summary, the uncertainties associated with both verification and validation must be
considered when interpreting future scenarios.

The results of this study about heat and cold waves are consistent with the results
published by both the IPCC (IPCC, 2013) and the AEMET (www.Aemet.es). All models
agree that the number of days considered as “hot days” will increase in the coming decades
and, therefore, the heat waves will be more frequent. On the other hand, the number of
“cold days” will be maintained and, in some cases, will even decrease so, on the contrary
of what is expected for heat waves, the frequency of cold waves will remain at its current
frequency.

What is stated in the previous paragraph is plausible with one of the main theoretical
conclusions about the trend of temperatures reported by the IPCC (IPCC 2013), which is
based on the change in the probability of reaching certain temperature values. The
probability distribution of temperatures can either move to warmer climates without
undergoing changes, can be extended or a combination of both. The third situation is the
one that has been observed in recent decades in different studies (Hansen et al., 2012)
IPCC, 2001) and is the most plausible situation in the context of the observed data. The
new distribution of temperatures implies a less pronounced change in the colder
temperatures of the series, a much more marked change in the warmer temperatures and
a greater number of warm extreme events. These conclusions support the results
anticipated for the coming decades—i.e., the maintenance of cold waves (in number and
intensity) and an increase of heat waves.

Regarding the territory of Aragón, the scenarios obtained in the present study indicate that
the Ebro valley, the most populated area in the region, will observe the highest maximum
temperatures, especially at the end of the century and in the summer (around 40 ºC), as
well as the greatest intensity of heat waves. This can have important impacts on the health
of the population. The health risks associated with heat waves are well known, both in
terms of mortality (Robine et al., 2008; Roldan et al., 2016) and mobility (Lin et al., 2009;
Steul et al., 2018), as well as health costs (Roldan et al., 2015). Moreover, these extreme
events will cause significant socio-economic impacts because a large part of Aragonese
industry and farming is situated in this zone (Olesen et al., 2011).

In addition, the Pyrenees will suffer the longest heat waves (but not the most intense ones),
especially at the end of the century, and the greatest increases in maximum temperatures
with respect to the values recorded during the historical period (1971–2000). The northern
zone is where the greatest intensity of cold waves will also be located, although their
durations will not change much compared to the current cold waves. Therefore, it is the
heat that is the most outstanding future risk for these ecosystems.

Changes in temperature (even if they are moderate) and warming are potential risks to this
rich ecological area, which is characterised by a diversity of species and a wealth of native
plants and animals, giving rise to national parks such as Ordesa and Monte Perdido (high
mountains). Therefore, these areas are particularly vulnerable to any changes in climate

87
that could lead to radical alterations in habitats or cause losses in their rich biodiversity, as
has been previously described (Kulakowski et al., 2017).

Furthermore, the Pyrenees area decides the amount of water available for urban and
agricultural use, with a direct effect on hydrology and agriculture in the Ebro Valley.
Changes in temperature and snow accumulation could seriously threaten the sustainability
and the equilibrium between available resources and water demand. On the other hand,
abrupt temperature changes can lead to accelerated thawing that could cause flooding in
populated areas near the river, a risk that has been well evaluated in the literature (Gobiet
et al., 2014; Guerreiro et al., 2018) and that will demand new and innovative adaptation
strategies (De Martino et al., 2012).

Finally, the disappearance of snow due to a rise in temperature in the area could lead to a
negative impact on the winter tourism industry and the Pyrenean ski resorts, which are
currently an important part of the region's economy, especially in the Pyrenean valleys
(Gilaberte-Burdalo et al., 2014; Gilaberte-Burdalo et al., 2017; Lopez-Moreno et al., 2011;
Pons et al., 2015).

5. Conclusions

Our study of data from Aragón, Spain, is the first to simulate episodes of heat and cold
waves throughout the 21st century in a valid way, based on the good verification and
validation results. In addition, we used climate models to obtain more precise climate
scenarios for the temperatures in this territory compared to previously available scenarios.

The use of ESMs from the fifth report of the IPCC and a methodology of downscaling has
proven to be effective, allowing us to obtain greater precision in the simulation of the climate
variables in this territory and to predict a realistic picture of the risks that extreme
temperatures can cause in the Aragón region, both spatially and temporally. The climate
scenarios predict higher maximum temperatures compared to previous scenarios. The
greatest increases will occur during summer at the end of the century, reaching values of
up to 7 ºC in the most unfavourable scenario.

Climate heat- and cold-wave scenarios showed that heat waves will be longer and more
intense in upcoming decades. However, intensity and duration of cold waves will not
change much compared to the current cold waves. The temperatures and heat-wave
intensities will be especially high in the Ebro Valley, the most populated area. However, the
Pyrenees will suffer the longest heat waves, especially at the end of the century, and the
greatest increases in maximum temperatures.

The present study provides useful information coming from the downscaling for the first
time in Aragón of models from IPCC5 to support decision-making and in the development
of specific measures to prevent socio-economic, environmental and human health impacts
due climate change in Aragón, a territory that can be a good indicator of the impacts of
climate change in southern Europe.

88
Material suplementario Resultado 1

Figure S1.1. Geographical representation of the expected changes of maximum temperature in winter for the periods
2040-2070 and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions
scenarios, RCP4.5 (figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute maximum
temperature

Figure S1.2. Geographical representation of the expected changes of maximum temperature in spring for the periods
2040-2070 and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions
scenarios, RCP4.5 (figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute maximum
temperature.
89
Figure S1.3. Geographical representation of the expected changes of maximum temperature in autumn for the periods
2040-2070 and 2070-2100 compared to the reference HistoricalPeriod (1976-2005) according to both emissions
scenarios, RCP4.5 (figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute maximum
temperature.

Figure S1.4. Geographical representation of the expected changes of minimum temperature in winter for the periods
2040-2070 and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions
scenarios, , RCP4.5 (figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute minimum
temperature.

90
 Figure S1.5. Geographical representation of the expected changes of minimum temperature in spring for the periods
2040-2070 and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions
scenarios, , RCP4.5 (figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute minimum
temperature.

Figure S1.6. Geographical representation of the expected changes of minimum temperature in autumn for the periods
2040-2070 and 2070-2100 compared to the reference HistoricalPeriod (1976-2005) according to both emissions
scenarios, RCP4.5 (figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute minimum
temperature.

91
Figure S1.7. Geographical representation of the expected maximum temperature in winter for theperiods 2040-2070 and
2070-2100 compared to the reference Historical Period (1976- 2005) according to both emissions scenarios, RCP4.5
(figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute maximum temperature.

Figure S1.8. Geographical representation of the expected máximum temperature in spring for theperiods 2040-2070 and
2070-2100 compared to the reference Historical Period (1976- 2005) according to both emissions scenarios, RCP4.5
(figures b and c) and RCP8 (figures d and e). Figure a represents the Historical absolute maximum temperature

92
 Figure S1.9. Geographical representation of the expected máximum temperature in autumn for the periods 2040-2070
and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions scenarios, RCP4.5
(figures b and c) and RCP8(figures d and e). Figure a represents the Historical absolute maximum temperature

Figure S1.10. Geographical representation of the expected minimum temperature in winter for the periods 2040-2070 and
2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions scenarios, RCP4.5
(figures b and c) and RCP8(figures d and e). Figure a represents the Historical absolute mínimum temperature

93
 Figure S1.11. Geographical representation of the expected minimum temperature in spring for the periods 2040-2070
and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions scenarios, RCP4.5
(figures b and c) and RCP8(figures d and e). Figure a represents the Historical absolute mínimum temperature

Figure S1.12. Geographical representation of the expected minimum temperature in autumn forthe periods 2040-2070
and 2070-2100 compared to the reference Historical Period (1976-2005) according to both emissions scenarios, RCP4.5
(figures b and c) and RCP8(figures d and e). Figure a represents the Historical absolute mínimum temperature

94
 Figure S1.13. Results of the verification process for heat waves obtained comparing the simulated heat waves for the
reanalysis Era40 along the observed heat waves for a common period (1980-2000). a) maximum intensity of the heat
waves registered in the 71 observatories used in the study against b) the maximum intensity obtained by downscaling of
the reanalysis Era40 in these observatories for the period 1980-2000. The colour of each pixel shows the maximum
intensity that corresponds to each heat wave ( ºC). The upper row shows the mean of the maximum intensity from the
whole observatories.

Figure S1.14. Results of the verification process for cold waves obtained comparing the simulatedheat waves for the
reanalysis Era40 along the observed cold waves for a common period(1980-2000). a) maximum intensity of the heat
waves registered in the 71 observatoriesused in the study against b) the maximum intensity obtained by downscaling of
the reanalysis Era40 in these observatories for the period 1980-2000. The colour of each pixel shows the maximum
intensity that corresponds to each cold wave ( ºC). The upperrow shows the mean of the maximum intensity from the
whole observatories.

95
Figure S1.15. Geographical representation of the expected evolution of the heat waves for Aragon in the periods 2041-
2070 and 2071-2100 in terms of increments values respect to the reference Historical Period (1971-2000) according to
the RCP4.5. The rows showthe three parameters analyzed in the study (average intensity, maximum intensity and
duration) and the columns the three temporal periods (Historical, 2041-2070 and 2071- 2100). The maps are generated
interpolating the available observatories over the entireterritory.

96
Figure S1.16. Geographical representation of the expected evolution of the heat waves for Aragon in the periods 2041-
2070 and 2071-2100 in terms of increments values respect to the reference Historical Period (1971-2000) according to
the RCP8.5. The rows showthe three parameters analyzed in the study (average intensity, maximum intensity and
duration) and the columns the three temporal periods (Historical, 2041-2070 and 2071- 2100). The maps are generated
interpolating the available observatories over the entireterritory.

97
Figure S1.17. Geographical representation of the expected evolution of the cold waves for Aragon in the periods 2041-
2070 and 2071-2100 in terms of increments values respect to the reference Historical Period (1971-2000) according to
the RCP4.5. The rows showthe three parameters analyzed in the study (average intensity, maximum intensity and
duration) and the columns the three temporal periods (Historical, 2041-2070 and 2071- 2100). The maps are generated
interpolating the available observatories over the entireterritory.

98
Figure S1.18. Geographical representation of the expected evolution of the cold waves for Aragon in the periods
2041-2070 and 2071-2100 in terms of increments values respect to the reference Historical Period (1971-2000)
according to the RCP8.5. The rows showthe three parameters analyzed in the study (average intensity, maximum
intensity and duration) and the columns the three temporal periods (Historical, 2041-2070 and 2071- 2100). The
maps are generated interpolating the available observatories over the entireterritory.

99
Anexo R1. Escenarios de temperatura y olas de calor y frío para Península
Ibérica y Baleares

En el Anexo Resultado 1 se recogen los resultados de las proyecciones climáticas para

todo el territorio español (Península e Islas Baleares) de temperatura (máxima y mínima)
y episodios de olas de frío y calor.

En las figuras AR1.1 a AR1.4 se muestran los resultados obtenidos para temperatura
máxima y temperatura mínima en base a los escenarios RCP4.5 y RCP8.5. Suponen
una representación espacial de la evolución esperada de cada variable en los periodos
2041-2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) bajo el
escenario considerado. Las filas muestran las cuatro estaciones del año (invierno,
primavera, verano y otoño) y las columnas los tres periodos temporales (Historical,
2041-2070 and 2071-2100).

En las figuras AR1.5 a AR1.8 se muestran los resultados obtenidos para los episodios
de olas de calor y frío en base a los escenarios RCP4.5 y RCP8.5. Suponen una
representación espacial de la evolución esperada de cada variable en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) bajo el
escenario considerado. Las filas muestran las características principales (intensidad
media, intensidad máxima y duración media) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100).

Los mapas se han generado interpolando todos los observatorios disponibles en el

estudio.

En base a los resultados obtenidos y en promedio a todo el territorio (habrá zonas donde
los cambios proyectados serán más o menos intensos) se espera:

 Un aumento gradual de las temperaturas medias (entre 2 y 4 ºC en función del

escenario considerado y la zona del territorio), siendo los meses estivales los
que experimentarán los ascensos más acusados.
 Un aumento en las temperaturas máximas a finales de siglo de entre 3.8 y 7 ºC
en los meses estivales (dependiendo del RCP considerado) que son los más
críticos en cuanto a esta variable. Estos aumentos suponen temperaturas
superiores a 38 ºC en toda la mitad sur peninsular, la zona Mediterránea y las
Islas Baleares. Aunque en el resto del año los ascensos se esperan algo menos
acusados, siguen siendo elevados. En los meses de invierno se esperan
aumentos de entre 2 y 4 ºC, en los meses de primavera entre 2.8 y 5 ºC y en
los meses de otoño entre 3y 6 ºC. Estos resultados corresponden con el periodo
2071-2100 bajo el escenario RCP8.5.
 Los aumentos esperados en las temperaturas máximas tienen como
consecuencia que los episodios de olas de calor sean cada vez más intensos y
aparezcan con más frecuencia. Además de ser más persistentes. La costa
Mediterránea se verá especialmente afectada por situaciones de olas de calor,
las cuales podrán abarcar hasta 20 días a finales de siglo en el caso más
desfavorable.
 Los mayores aumentos en temperatura mínima se esperan en los meses de
verano y otoño, especialmente en la Costa Mediterránea. En promedio, los

100
 aumentos esperados oscilan entre 3 y 5.8 ºC en los meses de verano y otoño y
entre 2 y 4 ºC en los meses de invierno y primavera.
 Aunque se esperan cambios considerables en las temperaturas mínimas
medias, no se esperan grandes cambios en los extremos, por lo que no se
aprecian alteraciones importantes de los episodios de olas de frío futuros
respecto a los actuales.

Figura AR1.1. Representación espacial de la evolución esperada de la Temperatura Máxima en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP4.5. Las filas
muestran las cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

101
Figura AR1.2. Representación espacial de la evolución esperada de la Temperatura Máxima en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP8.5. Las filas
muestran las cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

102
Figura AR1.3. Representación espacial de la evolución esperada de la Temperatura Mínima en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP4.5. Las filas
muestran las cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

103
Figura AR1.4. Representación espacial de la evolución esperada de la Temperatura Mínima en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP8.5. Las filas
muestran las cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

104
 Figure AR1.5. Representación espacial de la evolución esperada de las olas de calor en los periodos 2041-2070 y
2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP4.5. Las filas muestran las
características principales (intensidad media, intensidad máxima y duración media) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

105
 Figure AR1.6. Representación espacial de la evolución esperada de las olas de calor en los periodos 2041-2070 y
2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP8.5. Las filas muestran las
características principales (intensidad media, intensidad máxima y duración media) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

106
 Figure AR1.7. Representación espacial de la evolución esperada de las olas de frío en los periodos 2041-2070 y
2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP4.5. Las filas muestran las
características principales (intensidad media, intensidad máxima y duración media) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

107
 Figure AR1.8. Representación espacial de la evolución esperada de las olas de frío en los periodos 2041-2070 y
2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP8.5. Las filas muestran las
características principales (intensidad media, intensidad máxima y duración media) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

108
 Resultado 2
Impacto del cambio climático en la ocurrencia de
eventos de sequía de Aragón (NE de España)

109
 Impact of climate change on drought in Aragon (NE Spain)

Abstract

Droughts are one of the extreme climatic phenomena with the greatest and most
persistent impact on health, economic activities and ecosystems and are poorly
understood due to their complexity. The exacerbation of global warming throughout this
century probably will cause an increase in droughts, so accurate studies of future
projections at a local level, not done so far, are essential.

Climate change scenarios of drought indexes for the region of Aragon (Spain) based on
nine Earth System Models (ESMs) and two Representative Concentration Pathways
(RCPs) corresponding to the fifth phase of the Coupled Model Intercomparison Project
(CMIP5) have been generated for the first time. Meteorological Drought episodes were
analysed from three main aspects: magnitude (index values), duration and spatial extent.
The evolution of drought is also represented in a novel way, allowing identification,
simultaneously, of the intensity of the episodes as well as their duration in different
periods of accumulation and, for the first time, at the observatory level.

Future meteorological drought scenarios based on the Standardized Precipitation Index

(SPI) hardly show variations in water balance with respect to normal values. However,
the Standardized Precipitation Evapotranspiration Index (SPEI) which, in addition to
precipitation, considers evapotranspiration, shows a clear trend towards increasingly
intense periods of drought, especially when considering cumulative periods and those at
the end of the century.

Representation of the territory of the drought indexes reflects that the most populated
areas (Ebro Valley and SW of the region), will suffer the longest and most intense drought
episodes. These results are key in the development of specific measures for adapting to
climate change.

1. Introduction

Drought is probably one of the extreme climatic phenomena with the greatest impact on
the world’s population and that can affect millions of people every year around the planet
(Bryant, 1991; Wilhite, 2000). It also has serious effects on the availability of water and
therefore on economic activities such as agriculture (Lesk et al., 2016) and tourism and
profound impacts on human health (Stanke et al., 2013) and ecosystems (Alary et al.,
2014) that may persist over time. (Dai, 2011). However, drought is a phenomenon that
is not well understood due to its complexity and lack of historical records (Wilhite, 2000)
and because it depends on numerous factors.

For this reason, the scientific community and institutions are putting a lot of effort into
understanding, identifying, documenting and monitoring this phenomenon more
exhaustively. Examples are the drought databases of the European Drought
Observatory, the National Drought Mitigation Center and, the historical database of the
Standardized Precipitation Evapotranspiration Index (SPEI)
(http://spei.csic.es/database.html).

110
1.1 Droughts types and indexes

Precipitation is the primary controlling factor of drought but other meteorological

phenomena, such as temperature (Cook et al., 2014; Hao et al., 2017; Livneh and
Hoerling, 2016), wind (McVicar et al., 2012a) and relative humidity (Willett et al., 2014),
can modulate its intensity (Bates et al., 2008). Through potential evapotranspiration
(PET), it is possible to evaluate the amount of water that would evaporate and transpire
if there was enough water available, which is very important in the evaluation of
meteorological droughts.

Because drought affects so many different aspects (environmental, economic, social,

health), a single ‘drought’ does not really exist. Drought is often classified into four types
(Wilhite, 2000; Wilhite et al., 1985): meteorological, agricultural, hydrological and
socioeconomic drought.

The main subject of the current study is meteorological drought, a type of drought
characterized by below-normal precipitation over a period of months to years and that
should be defined as a condition relative to the normal local condition (Dai, 2011;
Paparrizos et al., 2018; Wilhite, 2000).

On the other hand, to characterize droughts, standardized drought indexes are used in
the literature. These indexes are direct indicators based on climate information, defined
so that the results are comparable in time and space since droughts of the same
magnitude can have very different effects depending on the time of year and the place
where they occur (Hayes et al., 1999; Vicente-Serrano, 2016; Wilhite, 2000).

Some of these indexes are well established and have been used to monitor climatic
conditions across different locations; these include the Palmer Drought Severity Index
(PDSI; Palmer, 1965) and Standardized Precipitation Index (SPI; McKee et al., 1993),
for example. The Lincoln Declaration on Drought Indexes (Hayes et al., 2011)
determined that SPI is the only index, from the point of view of meteorological drought,
valid for any region of the world and any time scale, being one of the most used in Europe
(Spinoni et al., 2015). It is able to provide better spatial standardization than PDSI (Lloyd-
Hughes and Saunders, 2002) and indicate drought initiation and termination because
they are implicit parts of the index (Sonmez et al., 2005).

SPI, however, presents some limitations such as that it neglects the effect of temperature
increase and, therefore, the effect that an increase in PET (Vicente-Serrano et al.,
2010a) or in the atmospheric evaporative demand (AED) (Vicente-Serrano et al, 2020)
can have on droughts, which may affect prediction of the impact of global warming in
future drought conditions. It should be noted, however, that other meteorological
variables as wind speed, solar radiation and air humidity, can also affect PET changes
linked to climate change.

To avoid this problem, (Vicente-Serrano et al., 2010a) proposed a new climatic drought
index, SPEI, which considers the difference between monthly precipitation and AED.
Thus, SPEI best reflects climate change as it makes a more realistic measurement of
water availability by incorporating the effect of temperature on changes in evaporation
demand as does PDSI. On the other hand, it maintains the multi-temporal nature and
simplicity of SPI (Marcos-Garcia et al., 2017).

111
According to the latest report of the Intergovernmental Panel on Climate Change (IPCC,
2014a), analysis of the precipitation regime (Calbo, 2010; Lavaysse et al., 2012),
droughts (Burke and Brown, 2008; Lopez-Bustins et al., 2013) and the extreme
temperatures that drastically increase evapotranspiration (ET) (Rebetez et al., 2006) and
decrease soil moisture (Sheffield and Wood, 2008) suggest that drought episodes could
become more severe around the world in the 21st century (Dai, 2013).

There are some studies that emphasize that future projections of drought may
overestimate drought episodes if future soil moisture (Berg et al., 2017) and runoff (Yang
et al., 2018) simulations are not taken into account (Berg and Sheffield, 2018), which can
affect AED. In addition, recent studies highlight the need to include CO2 concentration
in the analysis of AED under climate conditions since an increase of the CO2 acts
contributing to the increase in temperatures that in turn affect the Vapour-Pressure
Deficit (VPD). On the other hand, CO2 could increase water use efficiency by plants
reducing AED and therefore mitigate the drying (Dai et al., 2018; Roderick et al., 2015).
In this context, most European areas and the Mediterranean region seem to be
prominent regional climate change hotspots where an increment in the occurrence of
extreme events is expected (Beniston et al., 2007; Skaugen et al., 2004). Specifically, a
possible rise in the intensity and frequency of extreme drought events is expected
(Forzieri et al., 2014; Hoerling et al., 2012; Iglesias et al., 2007; Marcos-Garcia et al.,
2017; Paparrizos et al., 2018), especially in the summer months (Vicente-Serrano et al.,
2010c), and will have significant environmental, social and economic impacts
(Blenkinsop and Fowler, 2007).

The global climate models used today reproduce temperature trends very well, but the
level of precision for large-scale precipitation patterns is lower than for temperature
(IPCC, 2014b). This has caused the climatic projections of droughts to show great
uncertainty and therefore we cannot know with precision the effects of climatic change
on drought severity at the regional level in the future (Burke and Brown, 2008). This is
especially problematic in areas with high precipitation variability, such as the
Mediterranean region, where the drought patterns derived from the results of global
climate models are not consistent (Vicente-Serrano et al., 2004).

1.2 Drought in NE Spain (Aragón)

In Spain, as in the rest of Europe (Feyen and Dankers, 2009), different series of major
droughts have been happening in recent decades. In addition, the literature seems to
indicate a trend towards an increase in meteorological water scarcity in the Iberian
Peninsula, either due to an increase in the frequency of drought episodes or due to a
change in the precipitation regime (Fragoso et al., 2018; Gallego et al., 2011; Garcia-
Barron et al., 2011; Machado et al., 2011; Ojeda et al., 2017; Vicente-Serrano et al.,
2004). This makes necessary studies at a local level and the development of future
scenarios of droughts which are adequate as possible for evaluating the local impacts of
climate change.

Drought scenarios in Spain are also scarce: either they are from studies conducted prior
to the Fifth Assessment Report of the United Nations Intergovernmental Panel on
Climate Change (IPCC5) and in very small areas (Lopez-Bustins et al., 2013) or they
use IPCC5 models but use dynamic downscaling information from the European
Coordinated Regional Downscaling Experiment (EUROCORDEX; (Collados-Lara et al.,

112
2018; Marcos-Garcia et al., 2017). The latter also evaluates only SPI and SPEI at the
12-month scale. However, these studies agree that the combined use of SPI and SPEI
is adequate for studying drought episodes in the future (Lopez-Bustins et al., 2013;
Marcos-Garcia et al., 2017).

The combined study of both indexes, SPEI and SPI could be an effective formula for an
adequate study of meteorological drought in territories with the climatology of Aragon
(NE Iberian Peninsula). This region of Spain is characterized by a continental
Mediterranean climate with high precipitation variability and marked by very diverse
orography throughout its territory that includes areas of high mountains, valleys and
steppes (López et al., 2007). In addition, we must consider that previous studies in
Aragon have shown that an increase in temperature is one of the variables that will be
most noticeable with climate change throughout this century (Gaitan et al., 2019;
Ribalaygua et al., 2013a).

As far as we know, drought scenarios in Aragon have not been obtained to date. As has
been seen, it is essential to have local scenarios to determine the impact of climate
change on the environmental or socioeconomic reality of each region in order to make
decisions on adaptation to climate change.

The goal of this study is to obtain, for the first time, meteorological drought scenarios for
Aragon (located in NE of Spain) for the 21st century using a statistical methodology to
downscale GCMs from CMIP5.

To achieve this goal, the capacity of the GCMs to simulate the past observed climate
was assessed (validation) and using CMIP5, precipitation scenarios for Aragon were
generated to simulate future daily precipitation.

Finally, as Aragon is a region sensitive to episodes of drought caused by a varying rate

of precipitation and high temperatures, the SPI and SPEI meteorological indexes were
calculated and the frequency of occurrence of drought and its spatial distribution were
simulated to identify the drought vulnerability of the study area. Drought indexes were
also verified.

This study provides, for the first time, scenarios of meteorological drought in the NE of
Spain according to CMIP5 models, useful for predicting the impacts of climate change
on the availability of water at a local scale and which are necessary for stakeholders to
make decisions on adaptation and mitigation of climate change. On the other hand, this
region of Spain is a good indicator of many characteristic areas of southern Europe (high
mountains, river basins, steppes, etc.).

2. Data and methodology

2.1. Study area

The present study was carried out in the region of Aragon (NE of Spain) (Fig. R2.1).
Because of its location, Aragon falls within the Western Mediterranean climate area
characterized by scarce precipitation with cool winters and hot, dry summers. Differences
in latitude between the most northern and most southern points of Aragon (340 km length
and 240 km width) along with the influence of the Cantabrian and Mediterranean Seas
and the general atmospheric circulation as well as the orographic complexity of the

113
region (extreme altitude differences of over 3000 m between the plains (the Ebro River
valley) and the mountains (the Pyrenees)), give rise to great subclimate variety, with
different thermal and pluviometric regimes that condition the local climate (López et al.,
2007).

Figure R2.1. Location of the study Area and observatories. Aragon (Spain) in Europe. Points indicate the stations used
in the study. a) Stations of precipitation (264) used in the generation of climate regional scenarios of precipitation
(verification, validation and scenarios). b) Stations used exclusively on the generation of drought indexes (43). Map
source: OpenStreetMap.

Precipitation is scarce in most of Aragon and is distributed clearly according to relief, as

the isohyets are arranged in concentric circles decreasing from mountain areas to the
centre of the region. Although the average annual total precipitation of the Aragonese
territory is around 550 mm, there are regions for which the average is below these values
(for example, in the central sector of the Ebro Depression). Only in the Pyrenees and, to
a lesser extent, in the Iberian Mountain Range, does precipitation reach important
values, 1800–2000 mm, and show positive water balance values (considering the
difference between precipitation and AED). On the other hand, more than 60% of the
region has average values of AED above 1100 mm, showing a negative water balance,
to which contributes not only the scarce rainfall but also the strong wind ("Cierzo")
characteristic of the Ebro Valley (López et al., 2007). Therefore, 70% of the Aragonese
territory is considered semi-arid (index value proposed by the United Nations
Environment Program < 0.5 and even 30% presents values of 0.3) (Cherlet et al., 2018).

114
2.2. Datasets

2.2.1. Surface observation datasets

In this study, an observational dataset (daily maximum and minimum temperature and
precipitation) belonging to the extensive network of instrumental observatories owned by
the Spanish Meteorological Agency (AEMET) (http://www.aemet.es) was used (Fig.
R2.1). This dataset is the same as the one used in previous studies (Gaitan et al., 2019;
Ribalaygua et al., 2013a) in order to work with a set of data that has been subjected to
strict quality control (inhomogeneities, gaps, outliers, transcription errors and so on)
carried out first by the Government of Aragon (López et al., 2007) and completed, in a
second phase, by (Ribalaygua et al., 2013a). As a complement to quality controls, those
stations with a large number of data gaps or less than 15 years of daily records were
discarded.

For the simulation of future climate scenarios of precipitation, a first set of 263 stations
was used (red dots in Fig. R2.1a). Of these 263 stations, just those with data for both
variables, temperature and precipitation, were used for the simulation of drought indexes
(43 stations, Fig. R2.1b).

2.2.2. Atmospheric dataset

A set of nine climate models were selected from CMIP5, supplied by the Program for
Climate Model Diagnosis and Intercomparison (PCMDI) archives.

The global climate models called Earth System Models (ESMs) from the fifth phase of
the Coupled Model Intercomparison Project (CMIP5) (Tripathi et al., 2006) have
contributed to the acquisition of both dynamic and statistical downscaling techniques with
less uncertainty. These models integrate the individual parts of the climate system
(atmosphere, ocean, land and sea ice) and the exchange of energy and mass between
them (Knutti and Sedlacek, 2013). These models also include chemical processes, land
use, plant and ocean ecology and an interactive carbon cycle, which enables integration
of biochemical processes into the models (Heavens et al., 2013), constituting a robust
set of coordinated climate model experiments (Carvalho et al., 2017; Chen et al., 2016;
Perez et al., 2014).

The climate models (Table R2.1) were selected according to the time resolution (daily)
of available predictor fields, because it is required for the downscaling method used. All
of the models were ESMs (Jones et al., 2011; Wang et al., 2009).

This study used data from two different experiment families of GCMs: the Historical
experiment (Taylor et al., 2012), which covers much of the industrial period and can be
referred to as ‘twentieth-century’ simulations, and the Representative Concentration
Pathway (RCP) family (Moss et al., 2010), which corresponds to different possible
ranges of radiative forcing reached in the year 2100 with respect to values of the pre-
industrial era. This study used future projections determined by the RCP8.5 ‘high’
scenario and RCP4.5 ‘intermediate’ scenario, the core of IPCC5 experiments.

115
 2ºx2,5º
National Oceanic and Atmospheric
GFDL-ESM2M Dunne et al. (2012)
Administration (NOAA), E.E.U.U.
daily

Canadian Centre for Climate Modeling

2,8ºx2,8º
and Analysis (CC-CMA),
CanESM2 Chylek et al. (2011)
daily
Canadá.

1,4ºx1,4º CNRM (Centre National de

CNRM-CM5 Recherches Meteorologiques), Voldoire et al. (2013)
daily Meteo-France, Francia.

1,4ºx1,4º
Beijing Climate Center (BCC), China
BCC-CSM1-1 Xiao-Ge et al. (2013)
Meteorological Administration, China.
daily

1,87ºx1,25º
Met Office Hadley Center, United
HADGEM2-CC Collins et al. (2008)
Kingdom.
daily

Japan Agency for marine-Earth

Science and Technology (JAMSTEC),
2,8ºx2,8º
Atmosphere and Ocean Research
MIROC-ESM-CHEM Watanabe et al. (2011)
Institute (AORI), and National Institute
daily
for Environmental Studies (NIES),
Japan.

1,8ºx1,8º Raddatz et al. (2007);

Max-Planck Institute for Meteorology
MPI-ESM-MR
(MPI-M), Germany.
daily Marsland et al. (2003)

1,2ºx1,2º
Meteorological Research Institute
MRI-CGCM3 Yukimoto et al. (2011)
(MRI), Japan.
daily

2,5ºx1,9º Bentsen et al. (2012);

Norwegian Climate Centre (NCC),
NorESM1-M
Norway.
daily Iversen et al. (2013)

Table R2.1. Information about the nine climate models belonged to the 5 Coupled Model Intercomparison Project
(CMIP5) corresponding to the fifth report of the IPCC. Models were supplied by the Program for Climate Model
Diagnosis and Intercomparison (PCMDI) archives.

In order to study the behaviour of the CMIP5 model Historical simulations, we used the
reanalysis dataset from the European Centre for Medium-Range Weather Forecasts
(ECMWF ERA-40; http://www.ecmwf.int/research/era/do/get/ ) (Uppala et al., 2005) for
the period 1958–2000 at 6-hourly time resolution and 125 km spatial resolution. For
verification of the methodology, it was necessary to reduce the temporal and spatial scale
of the reanalysis in order to compare both ERA-40 and the climate model simulations
(Ribalaygua et al., 2013a; Ribalaygua et al., 2013b). The geographical limits of the
atmospheric window used were latitudes 31.5°N to 55.1°N and longitudes 27.0°W to
14.6°E, covering not only the geographic area under study but also the surrounding

116
atmosphere areas which exert a meteorological influence all over the Iberian Peninsula
(Ribalaygua et al., 2013a). The use of the ERA-40 data set has allowed us to compare
these new results with those published by Ribalaygua et al 2013a.

2.3. Methodologies

2.3.1. Validation and generation of future precipitation scenarios

A two-step analogue/regression statistical downscaling method developed previously

(Ribalaygua et al., 2013b) was applied to obtain future scenarios of precipitation and
drought. This method has been used in national and international projects, with good
verification results (Gaitan et al., 2019; Monjo et al., 2016; Moutahir et al., 2017;
Ribalaygua et al., 2018; Rodriguez et al., 2014; Santiago et al., 2017). This methodology
offers some advantages: it is computationally inexpensive, provides local information
and allows quantifying the uncertainty associated with the downscaling process (Van der
Linden and Mitchell, 2009). Other advantages are the application of future simulations
consistent with observations (physically coherent between them) and using local scale
(because nearby data points in space are not subjected to different climate change
conditions) (Ribalaygua et al., 2013b).

Through the validation process we can, on the one hand, evaluate the ability of each
ESM to simulate the predictor fields (comparing the downscaled Historical experiment
simulation for each model with the downscaled ERA-40 simulation for a common period,
1958–2000) and, on the other, quantify the uncertainties inherent to future climate
projections through an ensemble strategy (Monjo et al., 2016).

Bias and standard deviation at seasonal scale have been used as error measures. This
validation process presents some limitations related to the observational data available
to be considered in the final uncertainty analysis. More information about the validation
process can be consulted in (Ribalaygua et al., 2013b).

Future local climate scenarios at local and daily scale for precipitation were produced for
nine ESMs (see Table R2.1) and two RCPs (RCP4.5 and RCP8.5) as a previous step to
calculate the drought indices. As precipitation is an essential variable in the analysis of
drought, these scenarios are a starting point providing initial information on future
pluviometric conditions.

The local climatic projections of precipitation belonging to CMIP5 were obtained in this
study using the same methodology as that used for temperature scenarios, previously
described (Gaitan et al., 2019).

2.3.2. Drought indexes

SPI was developed by McKee et al. (1993) and is based on two assumptions: 1) that the
variability of precipitation is greater than that of temperature and AED, and 2) that the
rest of the variables are stationary over time. The SPI value is defined as a numerical
value that represents the number of standard deviations of precipitation, over the
accumulation period in question, with respect to the average, once the original
distribution of precipitation has been transformed into a normal distribution (i.e., mean of
zero and standard deviation of 1). The SPI values can be interpreted as the number of
standard deviations by which the observed anomaly deviates from the long-term mean.

117
SPEI developed by (Vicente-Serrano et al., 2010a) and revisited by (Begueria et al.,
2014) is a variant of the widespread SPI; it has greater potential as a drought index since
it considers the climate balance (through the difference between monthly precipitation
and AED). SPEI values can be interpreted in the same way as SPI values (number of
standard deviations by which the observed anomaly deviates from the long-term mean).

Both indexes were calculated using the R package ‘SPEI’ (Version 1.7). The SPI was
calculated using Gamma distribution to fit the original precipitation series (Organization
WMO, 2012) and the SPEI was calculated using log-logistic distribution (Vicente-Serrano
et al., 2015; Vicente-Serrano and Beguería, 2016). The parameters of these distributions
were obtained by the method of unbiased probabilistic weighted moments (Vicente-
Serrano and Beguería, 2016). The scale of SPI and SPEI values used in the study can
be seen in Table R2.2.

The period 1976-2005 was used as a reference period, which represents the last 30
years of the Historical period. Based on this reference period, both the SPI and the SPEI
were calculated for the period 2006-2100. The choice of the reference period was made
to evaluate the future hidroclimatic conditions of the region with respect to the average
conditions of the last 30 years of the Historical experiment.

To obtain the AED values used in the calculation of SPEI, both the Hargreaves-Samani
(1985) and Thornthwaite (1948), formulas have been used, denominated SPEI-Har and
SPEI-Thor, respectively. These formulas were chosen to calculate AED because they
are recommended within the SPEI package and they also depend only on temperature
and precipitation, unlike other more complex methods such as the Penman–Monteith
(Smith M et al., 1998) and Jensen–Haise methods (Jensen and Haise, 1963). Both
methods only take into account the temperature, so it is assumed that the calculation of
AED trends could have certain limitations (Irmak et al., 2012; McVicar et al., 2012b;
Sheffield et al., 2012). For a certain increase in the temperature, the change in the
obtained result can be higher than the one really expected according a complete method
like Penman-Monteith. Therefore, the role of AED on drought severity would be
overestimate and this would have some effect on the drought indices obtained for future
scenarios.

The way in which the indexes have been analysed follows the guidelines of the WMO
(WMO, 2017) which recommends analysis of a drought episode from three main aspects
– magnitude (index values), duration (alternation between positive and negative values)
and spatial extent – and all these aspects configure the severity of the episode.

In order to assess the capacity of the downscaling methodology to simulate SPI and
SPEI, we analysed the intensity and duration of the different drought episodes shown by
both indexes, comparing the SPI and SPEI values calculated from the simulated ERA-
40 temperature and precipitation series with those obtained from the observed series for
a common period (1970-2000). Verification of the maximum and minimum temperature
and precipitation can be seen in a previous study (Ribalaygua et al., 2013a). The
statistical measures used in the verification processes were the bias, standard deviation
and Pearson correlation. The statistical measures were calculated using R computing
software (R Development Core Team, 2010).

118
From the ESM simulated temperature and precipitation series (nine ESMs and two
RCPs), we determined the drought episodes that are expected in Aragon during the
upcoming decades of the 21st century. The SPI and SPEI scenarios were compared to
a historical period (1976–2005) to analyse the future changes with respect to the actual
situation of these extreme events.

To draw future local climate scenario maps, we used Thin Plate Spline (TPS) regression
from the R package ‘fields’ (Nychka et al., 2015).

3. Results

3.1. Validation and precipitation scenarios

The results of the validation process (comparison between the ERA-40 precipitation
simulations and the historical precipitation simulations for each ESM for a common
period (1958–2000)) are shown in Fig. R2.2, for both absolute (mm) and relative
precipitation (%). The results are variable depending on the model and the seasonal
period; however, all the models are able to reproduce the annual cycle of precipitation
as well as the differences between seasonal periods (maximum values in autumn and
spring, followed by winter and summer). The obtained bias and standard deviation are
less than ± 1 mm/day, which in relative terms supposes a difference of less than or
around ± 10% in the worst of the cases.

In general terms, a big variation in the Aragon precipitation regime is not expected.
According to scenario RCP8.5, mean variations in the amount of precipitation are
expected to be around ± 10% for all seasons of the year, except for the summer where
no precipitation change is expected. Scenario RCP4.5 shows no precipitation
fluctuations throughout the 21st century with respect to current values (see support
information, Fig. S2.1 to S2.4.

SPEI/SPI
≥2 extremely wet

1.5 a 2 severely wet

0.5 a 1.5 moderately wet

-0.5 a 0.5 normal values

-1.5 ≤ -0.5 moderately dry

-1.5 ≤ -2 severely dry

≤ -2 extremely dry

Table R2.2. SPEI/SPI Intensities Scale (Vicente-Serrano et al 2010a)

119
 Figure R2.2. Validation of precipitation. Comparison between the precipitations obtained using the downscaled
Historical data of the global climate models and the downscaled reanalysis data, for every seasonal period. Two graphs
at the top: seasonal comparative between the precipitation simulated using the downscaled Historical data (colour bars)
and that of the downscaled reanalysis data (black lines) for each global climate models (see Table R2.1) and for the four
seasons: winter (December-February; first bar of each group of four), spring (March–May, second bar), summer (June–
August; third bar) and autumn (September–November; four bar). Two graphs at the bottom: relative seasonal
differences between the simulated data using the downscaled Historical data and that of the downscaled reanalysis
data. Seasonal precipitation amounts are shown on the left columns and seasonal values of the standard deviation on
the right

3.2. Generation of future local climate scenarios of drought indexes

3.2.1. Verification of drought index simulation

To verify the simulation of drought indexes, the first step was to compare the SPI and
SPEI values obtained from the observations with those calculated from the simulated
series of ERA-40.

Fig. R2.3 shows the verification results corresponding to SPI at time scales from 1 month
(SPI-1M) to 12 months (SPI-12M) for the period 1970–2000 (Fig. R2.3a and 2.3b). This
process allows the identification of episodes of deficit or excess precipitation recorded
and simulated from ERA-40. In addition, the number of months in the period 1970–2000
in which SPI values were obtained within different intensity ranges (Table R2.2) for SPI-
1M, SPI-3M and SPI-6M are shown (Fig. R2.3c, R2.3d and R2.3e).

120
As can be seen in Fig. R2.3a and R2.3b, the time series of the simulated SPI for ERA-
40 shows, in an acceptable way, the same values presented by the observed SPI, with
a correlation of p = 0.75 in the case of SPI-1M, p= 0.72 for SPI- 3M, p = 0.64 for SPI-6M
and p = 0.61 for SPI-12M.

Figure R2.3. SPI verification. Results of the verification process for the SPI. a) time series of the SPI index calculated
from observed data at the time-scales from 1 to 12 months for the period 1970-2000, b) time series of the SPI index
calculated from downscaled ERA-40 at the time-scales from 1 to 12 months for the period 1970-2000. c), d) and e)
number of months within the 1970-2000 period corresponding to each interval of the SPI intensities scale based on
observed data (blue columns) and on downscaled ERA-40 (red columns) the time-scales of 1, 3 and 6 months. Average
of the all the stations used in the simulations.

The simulated and observed SPI values show dry episodes (negative SPI) in similar
periods, for example the periods 1970–1972, 1978, 1981–1982, 1989, 1994–1995 and
1998. The same can be seen for wet episodes (positive SPI) as in, for example, the
periods 1976–1977, 1988 and 1996–1997.

Figs. R2.4 and S2.5 show the results of the verification process for SPEI based on SPEI-
Har and SPEI-Thor calculations, respectively.

Similar results were obtained for calculation of SPEI based on the Hargreaves method
(Fig. R2.4a and R2.4b) although in this case the correlation obtained between the
observed and simulated time series of SPEI is slightly higher (0.80 for SPEI-1M, 0.78 for
SPEI-3M, 0.72 for SPEI-6M and 0.73 for SPEI-12M).

When the Thornthwaite method is used for calculating AED in SPEI (see Fig. S2.5), the
temporal correlations are lower than those obtained with SPEI based on Hargreaves.

121
On the other hand, for about 65–70% of the period considered, water balance conditions
in Aragon was considered normal (SPI/SPEI between −0,5 and 0,5), suffering extreme
wet or dry episodes for only 2–4% of the period 1970–2000.

The error (bias) made in the simulation of SPI (Fig. R2.3c to R2.3e) and SPEI (Fig. R2.4c
to R2.4e) is quite small for all of the classes considered (< ± 2 months).

Figure R2.4. SPEI Har verification. Results of the verification process for the SPEI based on Hargreaves
Evapotranspiration. a) time series of the SPEI index calculated from observed data at the time-scales from 1 to 12
months for the period 1970-2000, b) time series of the SPEI index calculated from downscaled ERA-40 at the time-
scales from 1 to 12 months for the period 1970-2000. c), d) and e) number of months within the 1970-2000 period
corresponding to each interval of the SPI intensities scale based on observed data (blue columns) and on downscaled
ERA-40 (red columns) for the time-scales of 1, 3 and 6 months. Average of the all the stations used in the simulations.

3.2.2. Local climate scenarios to predict drought indexes

Figs. R2.5 to R2.9 (complemented with Figs. S2.6 and S2.7) show the results obtained
for the simulation of SPI and SPEI throughout the 21st century from different
perspectives.

Fig. R2.5 shows local climate change scenarios for future SPEI (a, c and d) and SPI (b,
d and f) at 3-, 6- and 12-month scale, which have been predicted on the basis of the nine
models (see Table R2.2) and here are used to obtain a general vision of the changes in
water balance for the Aragon region as a whole. The future projections of SPI and SPEI
for the period 2006-2100 have been made based on the reference period (Historical
1976-2005). When working with normalized indexes, the future values of the SPI and
SPEI represent anomalies with respect to the average values of the reference period,
which allows to evaluate the future evolution of the hydric conditions in Aragon with
respect to the average of the last 30 years of the Historical experiment.

122
The SPI values obtained are hardly modified with respect to the Historical period so,
according to these results, the water balance characteristics of the region as a whole
would remain similar to the current ones. On the other hand, the SPEI climate change
scenarios, considering the effect of AED, show a marked tendency towards increasingly
negative values of the index with respect to the Historical period, especially at the end of
the century.

Both RCPs show a similar evolution until 2060, with changes of SPEI with respect to the
Historical period of −0.6 for SPEI-3M, −0.9 for SPEI-6M and −1.3 for SPEI-12M. For the
final period of the century, the variation begins to be more pronounced under the
conditions of RCP8.5, with changes of −1.2 for SPEI-3M, −1.8 for SPEI-6M and −2.8 for
SPEI-12M, while under scenario RCP4.5, SPEI values vary slightly from those reached
in 2060.

Figure R2.5. Simulated SPEI and SPI for the twenty-first century. Values are displayed as absolute increase compared
to the amount simulated for the 1976–2005 Historical period for the time scales 3 months (a and b), 6 months (c and d)
and 12 months (e and f). The vertical dotted line marks the end of the Historical data (2005). Data grouped for every
RCP simulation of every global climate model selected and for the last 30 years of every station. The ensemble median
(solid lines) and the 10th–90th percentile (shaded areas) values are displayed.

123
These results are very well reflected in the time-scale evolution maps, where the
simulated time series from 1976 to 2100 are represented, both for SPEI (Fig. R2.6) and
SPI (Fig. R2.7) and under both scenarios, RCP4.5 (Figs. R2.6a and R2.7a) and RCP8.5
(Figs.R2.6b and R2.7b) with respect to different time scales (from 1 to 12 months).

Figure R2.6. SPEI Time series under RCP4.5 and RCP8.5 along the 21st century at time-scales from 1 to 12 months.
Data grouped for every RCP simulation of every global climate model and for every station. Both emissions scenarios
are represented: RCP4.5 (figure a) and RCP8.5 (figure b)

Fig. R2.6 shows a tendency towards more and more extreme SPEI values, especially in
the longer time scales. For time scales of up to 4 months, an alternation between periods
considered normal and dry periods is expected (SPEI values between −1,5 and 0,5). For
longer time scales, there is a tendency towards more intense and prolonged periods of
drought, with SPEI values of up to −3 at the end of the century. The pattern obtained is
similar under both RCPs, being more pronounced in the case of RCP8.5.

In the time-scale map corresponding to SPI (Fig. R2.7), the same pattern as that
obtained for SPEI is not appreciated; in this case, alternating dry and wet periods are
observed for all time scales, these being somewhat more extensive as we move along
the time scales. The same pattern is observed under both RCPs, the signal being slightly
stronger in the case of RCP8.5.

124
 Figure R2.7. SPI Time series under RCP4.5 and RCP8.5 along the 21st century at time-scales from 1 to 12 months.
Data grouped for every RCP simulation of every global climate model and for every station. Both emissions scenarios
are represented: RCP4.5 (figure a) and RCP8.5 (figure b).

As a complement to the previous results, which allowed the extraction of results for the
water regime of Aragon as a whole, the spatial maps of both indexes are shown. Figs.
R2.8 and R2.9 show the climate scenarios for mean SPEI according to RCP4.5 and
RCP8.5, respectively. These figures show the temporal evolution for four time scales: 1
month (first row), 3 months (second row), 6 months (third row) and 12 months (fourth
row). The temporal periods chosen were 2011–2040 (present), 2041–2070 (mid-century)
and 2071–2100 (end-century). Figs. S2.6 and S2.7 show the same information but for
mean SPI.

The results for SPEI vary considerably between different points in the Aragon region.
Coinciding with what was said before, it is observed how the SPEI values become more
extreme as the 21st century and time scales advance. The Ebro Valley area is the one
that will be subject to more intense episodes of precipitation shortage at the end of the
21st century, with SPEI values from −1 at 3 months to −2 at 12 months according to
RCP4.5 and considerably more intense under RCP8.5 with values from −1.8 at 3 months
to −4 at 12 months. The north-west area of the region, which is expected to be most
affected by drought episodes, deserves special attention. The Pyrenees zone is the one
that will clearly suffer the fewest expected drought episodes; under RCP4.5 it is expected

125
to remain in normal water balance conditions while under RCP8.5, at most, SPEI will
reach values of −1.5 (at the end-century and at 12-month time scale).

Figure R2.8. Time-scales SPEI maps under RCP4.5. Geographical representation of the expected evolution of the SPEI
for Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period (1971–2000) in terms
of absolute values according to the RCP4.5 at different time-scales. The rows show the four time-scales analysed in the
study (1 months, 3 months, 6 months and 12 months) and the columns, the three temporal periods (2011-2040, 2041–
2070 and 2071–2100). The maps are generated by interpolating the available stations over the territory.

The SPI spatial maps (Figs. S2.6 and S2.7) show how the region will remain under
normal water balance conditions, highlighting the Ebro basin at the end of the 21st
century and under RCP8.5, where more negative values of SPI (around −1) are
appreciated, but which are still within the range considered normal for the region.

126
Figure R2.9. Time-scales SPEI maps under RCP8.5. Geographical representation of the expected evolution of the SPEI
for Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period (1971–2000) in terms
of absolute values according to the RCP8.5 at different time-scales. The rows show the four time-scales analysed in the
study (1 months, 3months, 6 months and 12 months) and the columns the three temporal periods (2011-2040, 2041–
2070 and 2071–2100). The maps are generated by interpolating the available stations over the territory.

It is important to emphasize that if the average value of the SPEI/SPI tends to

increasingly negative values and if this is a constant trend in the future, the conditions
considered normal today will evolve towards new values considered normal (Vicente-
Serrano et al. 2020).

This study has been carried out for each of the observatories used in the study and for
each of the climatic models, which reveals that the entire region is going to be affected
by episodes of drought despite its location and height. As an example, the temporal

127
evolution of both indexes obtained according to the MPI-ESM-MR climate model and
under both RCPs is shown for the observatories of Zaragoza (Figs. R2.10 and R2.11)
and Cedrillas-Huesca (Figs. S2.8 and S2.9).

Figure R2.10. Time series for Zaragoza under MPI-ESM-MR RCP4.5. Evolution of the SPEI (first column) and the SPI
(second column) based on the MPI-MR-SM model and under the RCP 4.5 at different time-scales - 1 month (first row), 3
months (second row), 6 months (third row) and 12 months (fourth row) - for Zaragoza.

The choice of these observatories was based on the Climate Atlas of Aragon (López et
al., 2007), since they are two of the reference points used in the climatic characterization
of the region. The choice of these observatories was also made based on their location;
the Zaragoza observatory is located in the Zaragoza airport station at a height of 263 m
while the Cedrillas-Huesca observatory is located in the northern area of the region at a
height of 1347 m. In addition, the Zaragoza airport station is considered representative
of the variability of temperatures in Aragon (Roldan et al., 2011).

The expected temporal evolution of SPEI throughout the 21st century is consistent with
that explained above, but as it is a single climatic model and uses a single observatory,
the alternation between wet and dry periods can be seen more clearly at a time scale of
1 to 3 months. Also, as we move forward in the time scales, this alternation softens,
resulting in periods of more intense and prolonged precipitation shortage while, for SPI,

128
the alternation between wet and dry periods is observed for all time scales. This
highlights, as for SPI, how periods with positive SPI for the Cedrillas-Huesca observatory
are more intense and prolonged than those predicted for Zaragoza.

Figure. R2.11. Time series for Zaragoza under MPI-ESM-MR RCP8.5. Evolution of the SPEI (first column) and the SPI
(second column) based on the MPI-MR-SM model and under the RCP 8.5 at different time-scales - 1 month (first row), 3
months (second row), 6 months (third row) and 12 months (fourth row) - for Zaragoza.

4. Discussion

These results offer the possibility of having future climate projections based on recently
updated data, allowing the evaluation of how drought could affect the region of Aragon,
both spatially and temporarily, and can be taken as a reference to analyse its impact on
multiple sectors. Temporally, drought increases to the end of the century; at the territory
level, the area most affected will be the central area of the Ebro Valley, where most of
the population in the area is concentrated.

The difficulty of developing impact studies and quantifying their damage as a result of
periods of water scarcity comes mainly from the lack of observed values and studies at
a local level with future projections, hence the need to publish studies of these
characteristics.

129
In this study, climate change scenarios of drought indexes for the region of Aragon,
Spain, based on nine ESMs corresponding to CMIP5 have been generated for the first
time.

The evolution of two indexes, SPI and SPEI, has been obtained throughout this century
and also over the territory, which has allowed us to observe that while SPI, which only
considers precipitation, shows few changes, SPEI, that considers temperature and
incorporates the effects of AED, shows a tendency towards periods of increasingly
intense drought, especially when considering accumulated periods of longer duration
and those at the end of the century. Therefore, in the current climate change context it
is essential to take into account the effect of temperature in the study of droughts.

Figs. R2.6 and R2.7 represent a novel representation of the evolution of drought,
allowing identification, simultaneously, of the intensity of the episodes and their duration
in different periods of accumulation.

One of the strengths of this study is the use of local climate scenarios (at the observatory
level) to generate future drought indexes. Having this information will facilitate decision-
making in the face of expected changes based on what is expected to occur at each
observatory and not in the region a whole. As an example of the study at local level, the
results of future climate scenarios for Zaragoza (representative observatory of Aragon,
Roldan et al., 2011) and Cedrillas-Huesca (support information) are shown.

4.1. Precipitation scenarios used for the simulation of drought indexes.

For the simulation of precipitation, ESMs have been used instead of climatic models.
ESMs are the most powerful climatic models to date and incorporate significant
improvements (Flato et al., 2014) that allow better accuracy in climate simulation, as can
be seen in the good results obtained in the validation process.

Validation of the ESMs has shown good results for simulating precipitation. Both the
obtained bias and standard deviation are less than ± 1 mm/day, which in relative terms
supposes differences of less than or around ± 10% in the worst cases; however, those
values are within the order of natural variability of precipitation. These results are better
than those obtained for the generation of scenarios of the fourth IPCC report published
by (Ribalaygua et al., 2013a) particularly in the summer months, a particularly critical
time in Aragon.

The results obtained for the processes of verification of the methodology (Ribalaygua et
al., 2013a) and validation of the ESMs are good enough to allow the use of local climatic
scenarios generated under these conditions in impact studies and analysis of extreme
episodes such as periods of precipitation shortage.

Future precipitation scenarios show, under RCP8.5 conditions, a slight decrease in

precipitation throughout the 21st century for all seasons of the year, except for the
summer months where there is hardly any variation compared to current values of
precipitation in the region. Under RCP4.5 conditions, less pessimistic than the previous
one, barely any precipitation changes are expected at any time of the year.

These results are consistent with those published by AEMET (www.aemet.es) and
directly by the IPCC (Mukherjee et al., 2018), although the latter show the direct outputs
of the ESMs and do not carry the added value of applying downscaling techniques.

130
4.2. Consideration of the simulation of drought indexes

SPI is considered by experts in this field as one of the few indexes applicable in any
region of the world for any time scale (Hayes et al., 2011) and with multiple advantages
of application compared to other indexes of widespread use such as PDSI (Dracup et
al., 1980; Guttman, 1998; Hayes et al., 2011; Hayes et al., 1999; Vicente-Serrano et al.,
2010b). In the context of climate change with significant temperature variations (Gaitan
et al., 2019), SPEI has been chosen; its formulation is similar to that of SPI and allows
the comparison of both indexes and evaluation of the future behaviour of drought
episodes considering the effects of future temperature changes. Both indexes have been
verified and used previously in Aragon (Vicente-Serrano et al., 2010a). We have only
used the temperature in the calculation of AED because the absence of observed
historical data of variables such as radiation or humidity does not allow us a correct
validation process of certain indices such as Penman that include these variables.

4.2.1. Verification results

In general, the results of the verification process show good correlations between the
observed and simulated time series for both indexes for the period 1970–2000, higher
ones being obtained for SPEI. This is consistent with the results published by Vicente-
Serrano et al. (2012); they obtained higher correlations for the calculation of SPEI than
SPI, especially for the summer months, which are the most critical in the region of
Aragon.

The temporal series based on observations are satisfactorily represented by the

temporal series based on simulations, recreating almost all dry and wet episodes of
importance. It is observed how both the simulated SPI and SPEI tend, for the majority of
times, to present dry and humid periods of greater intensity than those observed,
especially for longer time scales, as occurred in 1976–1977 for positive values of the
indexes and in 1981–1982 for negative ones. In general, the number of months of the
period 1970–2000 located within each of the classes defined for SPI/SPEI has been
simulated very satisfactorily.

The dry and wet periods detected in this study coincide with or are approximate to those
published previously (Vicente-Serrano and Lopez-Moreno, 2005) based on SPI (dry
episodes: 1986–1987, 1989 and 1994–1997; wet episodes: 1976–1980), in the Climate
Atlas of Aragon (López et al.,2007) based on the precipitation regime (dry episodes:
1970, 1985, 1993 and 1995), by Spinoni (Spinoni et al., 2015) based on a combined 12-
month index (dry episodes: 1979–1980 and 1995–1998) and by Tselepidaki (Tselepidaki
et al., 1992) from a European study (dry episode: 1989), among others. In some cases,
the years are not exactly the same because of the use of different drought and temporal
scale indexes.

4.2.2. Future scenarios

The uncertainties associated with both processes, verification and validation, should be
considered when interpreting future scenarios. For drought projections the focus should
be on changes in the frequency and magnitude of cases located at the lower tail of the
distribution as was recommended by Vicente-Serrano et al. (2019).

131
Future meteorological drought scenarios based on SPI barely show water balance
variations with respect to normal values, regardless of the time scale considered and the
region of Aragon, except for the Ebro Valley where there is a slight sign of drought at the
end of the 21st century and under the conditions of RCP8.5.

These results were expected due to precipitation scenarios barely showing changes
throughout the 21st century.

When considering other climatic variables, such as temperature, the drought scenarios
based on SPEI show a clear trend towards increasingly dry periods and longer droughts,
especially in the Ebro area and south-west of the region. According to the trends shown
by the temperature and precipitation scenarios obtained for Aragon, the results obtained
were expected. The fact that the results obtained at the 12-month scale are more intense
than those of 1-3 months is partly a result of the way in which drought indices are
formulated and the autoregressive component of its metric so that when the timescale
increases, changes in the frequency of drought conditions increase more in comparison
to changes in the mean state. Although, recently, Vicente-Serrano et al (2019) showed
that these changes are independent of the metric with which these indices have been
calculated, changes in the frequency of drought conditions increase more in comparison
to changes in the mean state.

The lack of consideration of variables such as temperature, wind or humidity in the

calculation of SPI means that this index presents certain limitations under global warming
conditions(Mishra and Singh, 2010; Mishra and Singh, 2011; Vicente-Serrano et al.,
2010a) and it is for this reason that, when considering AED in the calculation of SPEI,
such different results are obtained, especially at the end of the century and not only under
the conditions of RCP8.5, that some authors consider less realistic (Hausfather and
Peters, 2020), but also of RCP4.5.

Some studies recommend the use of PET and add value against global warming (Hu
and Willson, 2000; Vicente-Serrano et al., 2010a), Tsakiris and Vangelis, 2005). Recent
studies (Vicente-Serrano et al. 2019; Vicente-Serrano et al., 2020) suggest using AED
in the future study of droughts, as well as analyzing the impact caused by the increase
in CO2 (Yang et al., 2019). Probably, considering the response that vegetation could
have to an increase in CO2 and, therefore, in the evapotranspiration process, could
provide some variation in the future evolution of drought episodes that should be
explored in future studies.

The results of future drought scenarios presented here show results in line with those
obtained in other studies where it is concluded that the Mediterranean regions will
experience an increase in the severity and frequency of droughts (Stagge et al., 2015)
as a result of a slight decrease in precipitation and an abrupt increase in temperatures
(European Environment Agency, 2010; (Stagge et al., 2015) and which represent an
increase in water scarcity (Estrela et al., 2012). More specifically in the region of Aragon,
the ECCE project, based on dynamic downscaling and scenarios of the fourth IPCC
report (Ministerio de Medio Ambiente, 2011), showed a future decline of the Ebro runoff,
and Cook (Cook et al., 2014), based on scenarios of the fifth IPCC report but without
downscaling, obtained an increase in drought episodes based on SPEI.

132
4.2.3. Impact on the territory

Although there have been studies on the Aragon area, none present as complete a
picture as this study, combining drought evaluation with SPI and SPEI (that is,
considering the effect of global warming) based on scenarios of the fifth IPCC report and
providing the added value of working at the local scale by applying a downscaling
technique.

The scenarios obtained in this study indicate that the Ebro Valley, the most populated
area in the region that includes the largest city, Zaragoza with more than 650.000
people, will be most susceptible to future periods of extreme drought and will suffer
periods of drought of greater intensity and duration, especially at the end of this century,
which will have consequences in sectors such as health, water management, economy
and society in general (Lee et al., 2017).

It is remarkable that, in previous publications (Ribalaygua et al., 2013a), we detected

that the highest values of maximum temperature, especially at the end of the century
and in summer (around 40 °C) as well as the greatest intensity of heat waves will also
take place in this area, so it will be especially vulnerable and these data should be
considered in the development of specific measures for adapting to climate change.

Adaptation to climate change in each region requires studies applied to the climatic
dynamics of each territory, so downscaling quality studies are essential for this. However,
these results at the local level are also useful for the whole of the southern Iberian
Peninsula and central Europe, since Aragon brings together geographical and climatic
features representative also of these other areas.

5. Conclusions

The generation for the first time of climate change scenarios of drought indexes for the
region of Aragon (Spain) based on nine ESMs and two RCPs from CMIP5 has allowed
us to obtain simultaneously the most accurate representation to date of the magnitude,
duration and intensity of meteorological drought episodes and their duration in different
periods of accumulation in this area of Spain. The use of different drought indices and
drought time-scales and its graphic representation is a relevant novelty in the scientific
literature.

This has allowed the detection of a clear trend towards increasingly intense periods of
drought, especially at the end of the century when cumulative periods of longer duration
are considered. This trend is detected only in the future drought scenarios based on
SPEI (which in addition to precipitation, considers AED), while in the SPI-based
scenarios it is softened. These results reinforce the need to study these extreme
phenomena in a context of climate change, considering the temperature.

At the territory level, spatial representation allowed us to discover that the area that will
be most affected by longer and more intense periods of drought, but also the greatest
decrease in precipitation (around 10%), is the Ebro Valley, the area that concentrates
most of the population as well as the main economic activities of the zone. The results
have also allowed, for the first time, the study of future drought indexes at the observatory
level, specifically for the most populous city, Zaragoza.

133
To cope effectively with the impacts of these extreme events that are expected in the
present century, it is essential to be able to generate local scenarios that accurately
describe climate change at the territory level. On the one hand, our results not only
confirm a trend already described in the Mediterranean area of an increase in the severity
and frequency of droughts but can also serve as a model and sentinel for similar areas,
since it has very varied climatic and orographic conditions.

134
Material suplementario Resultado 2

Figure S1.1. Absolute seasonal precipitation in Aragón (mm/day). Simulated precipitation for the twenty-first century
displayed as absolute increase compared to the amount simulated for the 1976–2005 Historical period, for the four
seasons (winter DJF, spring MAM, summer JJA and autumn SON) The vertical dotted line marks the end of the
Historical data (2005). Data grouped for every RCP simulation of every global climate model selected and for the last
30years of every station. The ensemble median (solid lines) and the 10th–90th percentile (shaded areas) values are
displayed

135
 Figure S1.2. Seasonal precipitation maps under RCP4.5 Geographical representation of the expected changes of
absolute precipitation in winter, spring, summer and autumn for the periods 2041-2070 and 2071-2100 respect to the
reference Historical Period (1976-2005) regarding to the scenario RCP4.5 (ESMs Ensemble mean).

136
 Figure S1.3. SPEI Thor Verification Results of the verification process for the SPEI based on Thornthwaite
Evapotranspiration. a) time series of the SPEI index calculated from observed data at the time-scales from 1 to 12
months for the period 1970-2000, b) time series of the SPEI index calculated from downscaled ERA-40 at the time-
scales from 1 to 12 months for the period 1970-2000 .c), d) and e) number of months within the 1970-2000 period
corresponding to each interval of the SPI intensities scale based on observed data (blue columns) and on downscaled
ERA-40 (red columns) for the time-scales of 1, 3 and 6 months. Average of the all the stations used in the simulations.

137
Figure S1.4. Time-scales SPI maps under RCP4.5. Geographical representation of the expected evolution of the SPI for
Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period (1971–2000) in terms of
absolute values according to the RCP4.5 at different time-scales. The rows show the four time-scales analysed in the
study (1 months, 3months, 6 months and 12 months) and the columns the three temporal periods (2011-2040, 2041–
2070 and 2071–2100). The maps are generated by interpolating the available stations over the entire territory.

138
Figure S1.5. Time-scales SPI maps under RCP8.5. Geographical representation of the expected evolution of the SPI for
Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period (1971–2000) in terms of
absolute values according to the RCP8.5 at different time-scales. The rows show the four time-scales analysed in the
study (1 months, 3months, 6 months and 12 months) and the columns the three temporal periods (2011-2040, 2041–
2070 and 2071–2100). The maps are generated by interpolating the available stations over the entire territory.

139
Figure S1.6. Time series for Cedrillas-Huesca under MPI-ESM-MR RCP4.5. Evolution of the SPEI (first column) and the
SPI (second column) based on the MPI-MR-SM model and under the RCP 4.5 at different time-scales - 1 month (first
row), 3 months (second row), 6 months (third row) and 12 months (fourth row) - for Cedrillas-Huesca.

140
Figure S1.7. Time series for Cedrillas-Huesca under MPI-ESM-MR RCP8.5. Evolution of the SPEI (first column) and the
SPI (second column) based on the MPI-MR-SM model and under the RCP 8.5 at different time-scales - 1 month (first
row), 3 months (second row), 6 months (third row) and 12 months (fourth row) - for Cedrillas-Huesca.

141
Anexo resultado 2. Escenarios de precipitación sequía para Península
Ibérica y Baleares

En el Anexo Resultado 2 se recogen los resultados de las proyecciones climáticas para

todo el territorio español (Península e Islas Baleares) de precipitación y episodios de
sequía a partir del SPI y del SPEI.

En las figuras AR1.1 a AR1.2 se muestran los resultados obtenidos para precipitación
en base a los escenarios RCP4.5 y RCP8.5. Suponen una representación espacial de
la evolución esperada en los periodos 2041-2070 y 2071-2100 comparados con el
periodo de referencia (1976-2005) bajo el escenario considerado. Las filas muestran las
cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres
periodos temporales (Historical, 2041-2070 and 2071-2100).

En las figuras AR1.5 a AR1.8 se muestran los resultados obtenidos para los episodios
de olas de calor y frío en base a los escenarios RCP4.5 y RCP8.5. Suponen una
representación espacial de la evolución esperada de cada variable en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) bajo el
escenario considerado. Las filas muestran las características principales (intensidad
media, intensidad máxima y duración media) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100).

Los mapas se han generado interpolando todos los observatorios disponibles en el

estudio.

En base a los resultados obtenidos y en promedio a todo el territorio (habrá zonas donde
los cambios proyectados serán más o menos intensos) se espera:

 Las proyecciones de precipitación para el territorio peninsular y balear apenas

muestran ligeros cambios en la cantidad de precipitación (alrededor de un ±5%)
acumulada anualmente, aunque con eventos de precipitación más concentrados
en el tiempo. Se espera un descenso más acusado en los meses de otoño,
pudiendo ser del 5 al 8% a finales de siglo en el caso más desfavorable.
 Según los resultados obtenidos para el SPI (y de acuerdo a los cambios
esperados en las precipitaciones) todo el territorio se mantendrá bajo episodios
normales de alternancia de episodios secos y húmedos (valores de SPI entre ±
0.5)
 Según los resultados obtenidos para el SPEI, la situación es completamente
diferente. En este caso se espera que la región se mueva hacía episodios de
sequía cada vez más extremos (SPEI inferior a-3). Intensificándose la sequía
según aumentan los periodos considerados. La sequía será especialmente
extrema en la meseta central a finales de siglo, con valores de SPEI inferiores a
-4, aunque el resto del territorio también se espera que se encuentre en sequía
severa o extrema.

Una situación diferente se obtiene al evaluar la sequía meteorológica a partir de

escenarios basados en el SPEI (combinando el efecto de la precipitación y la ETP). Los
resultados obtenidos muestran que bajo el escenario RCP8.5 y a finales de siglo, la
situación se va agravando según se aumentan los periodos considerados. De esta
manera, en acumulado a 1 o 3 meses, la mayoría del territorio sufrirá sequía moderada

142
a severa, mientras que en acumulado de 12 meses, casi todo el territorio se encontrará
en situación de sequía severa a muy extrema. A mediados de siglo, la mayor parte del
territorio se encontrará en situación de sequía moderada a severa.

Figura AR2.1. Representación espacial de la evolución esperada de la Precipitación absoluta en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP4.5. Las filas
muestran las cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

143
Figura AR2.2. Representación espacial de la evolución esperada de la Precipitación absoluta en los periodos 2041-
2070 y 2071-2100 comparados con el periodo de referencia (1976-2005) según el escenario RCP4.5. Las filas
muestran las cuatro estaciones del año (invierno, primavera, verano y otoño) y las columnas los tres periodos
temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado interpolando todos los
observatorios disponibles en el estudio.

144
 Figura AR2.3. Representación espacial de la evolución especial de la evolución esperada del SPEI en los periodos
2041-2070 y 2071-2100 comparados con el periodo de referencia (1971-2000) según el escenario RCP4.5. Las filas
muestran los cuatro periodos acumulados de sequías considerados (6 mes, 12 meses, 24 meses y 60 meses) y las
columnas los tres periodos temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado
interpolando todos los observatorios disponibles en el estudio.

145
 Figura AR2.4. Representación espacial de la evolución especial de la evolución esperada del SPEI en los periodos
2041-2070 y 2071-2100 comparados con el periodo de referencia (1971-2000) según el escenario RCP8.5. Las filas
muestran los cuatro periodos acumulados de sequías considerados (6 mes, 12 meses, 24 meses y 60 meses) y las
columnas los tres periodos temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado
interpolando todos los observatorios disponibles en el estudio.

146
 Figura AR2.5. Representación espacial de la evolución especial de la evolución esperada del SPI en los periodos
2041-2070 y 2071-2100 comparados con el periodo de referencia (1971-2000) según el escenario RCP4.5. Las filas
muestran los cuatro periodos acumulados de sequías considerados (6 mes, 12 meses, 24 meses y 60 meses) y las
columnas los tres periodos temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado
interpolando todos los observatorios disponibles en el estudio.

147
 Figura AR2.6. Representación espacial de la evolución especial de la evolución esperada del SPI en los periodos
2041-2070 y 2071-2100 comparados con el periodo de referencia (1971-2000) según el escenario RCP8.5. Las filas
muestran los cuatro periodos acumulados de sequías considerados (6 mes, 12 meses, 24 meses y 60 meses) y las
columnas los tres periodos temporales (Historical, 2041-2070 and 2071-2100). Los mapas se han generado
interpolando todos los observatorios disponibles en el estudio.

148
 Resultado 3
Evaluación del impacto del cambio climático en el
sector vitícola español mediante indicadores
bioclimáticos

149
 Future climate change impacts for the Spanish wine sector through bioclimatic
indicators

Abstract

Grapevine cultivation is an ancestral practice in Mediterranean regions such as Spain.

The current climatic characteristics of this region make it a particularly optimal area for
its cultivation, and the climatic changes expected in the coming decades may jeopardise
this climatic suitability. Therefore, accurate studies of future projections at a local level
are essential.

Local climate change scenarios of six bioclimatic indicators (absolute values together
with their categorisation) related to vineyards for the Spanish region based on nine Earth
System Models (ESMs) as well as two Representative Concentration Pathways (RCPs)
corresponding to the fifth phase of the Coupled Model Intercomparison Project (CMIP5)
were generated for the first time. These indicators are the Huglin Index (HI), the Cool
Index (CI), the Dryness Index (DI) and the Hidrotermic Index. As a complement, two
combined indicators were calculated: the Multicriteria Climatic Classification System
(MCC System) and the Composite Index (CompI). The whole territory was analysed as
well as the areas involved in the Spanish Denominations of Origin.

Our results show that Thermal indicators (HI and CI) will tend to increase through the
twenty-first century, while water scarcity (DI) will be more pronounced. The trends found
do not have the same repercussions throughout the territory. In the south of the
peninsula, with HI values exceeding 3500°C and CI above 20°C and DI below –200 mm,
the continuity of the wine-growing sector in its current state is seriously endangered, with
a decrease in climatically optimal years as shown by the CompI values. On the contrary,
the northern peninsula and mountainous areas, despite the expected increases, with HI
below 2500°C, cool nights (CI below 15°C) and sufficient water supply (DI above 150
mm) considerably improve their climatic suitability (CompI) although the risk of mildew
disease remains due to the increase in temperature and humidity.

1. Introduction
Over the last decades, changes directly related to the heliothermal and hydric
requirements that grapevines need for optimal growth have been observed because of
climate change. Among others, increases in temperatures, alterations in the precipitation
regime, alterations in potential evapotranspiration or increases in CO2 concentrations are
affecting vineyards worldwide (Alonso and O'Neill, 2011; Battaglini et al., 2009; France
and Dubourdieu, 2016).

Grapevine is very sensitive to climate (White et al., 2006, Winkler 1974) and weather
conditions over a wide range of time scales (Santos et al., 2020b). Changes in the
weather/climate patterns due to climate change are causing numerous impacts on the
cultivation of grapes (Jones et al., 2005) with economic repercussions, especially in
warmer areas such as Spain. .

Average climate and climatic variability are the environmental factors that most influence
wine quality and production (Santos et al., 2011). The results shown by climate
projections seem to show a trend towards stronger and stronger impacts (Meehl et al.,

150
2007) and a shift of the optimal areas for vine cultivation towards the Poles by about 20°
by 2050 (Kenny and Harrinson, 1993; Tate, 2001). Many authors have highlighted how
climate change will alter current wine-growing regions and the need to act accordingly
(Jones and Alves, 2012; Lazoglou et al., 2018; Schultz and Stoll, 2010; White et al.,
2006, Olatt et al., 2016).

Europe, as one of the world's major wine-growing regions, is one of the areas most
affected by climate change, where extreme events are expected to become more
pronounced (Gaitan et al., 2020; Porter and Semenov, 2005). This will lead to increased
irrigation demand (Doll, 2002), increased diseases and pests (Alig et al., 2002) and
changes in viticultural zoning (Malheiro et al., 2010).

Spain is one of the main wine producers and exporters as well as the world's leading
vineyard (with 949,565 ha of vineyards, 13% of the world total, FEV, 2021). Due to its
location in southern Europe, it is expected to be one of the regions most affected by the
effects of climate change, especially rising temperatures and water stress (Gaitán et al.,
2019, Gaitán et al., 2020). Indeed, these impacts are detectable today. In Spain, along
with the aforementioned impacts, the area of vineyards has decreased in the north-east
of the peninsula as a result of water stress (Odo Camps and Ramos, 2012), an increase
in the demand for irrigation (Alonso and O'Neill, 2011) and a reduction in the life
expectancy of vines by 30% (Expansión, 2016). According to a study by the University
of La Rioja (Expansión, 2019), 90% of professionals associated with a Designation of
Origin have felt the effects of climate change and 56% consider that these impacts are
affecting them considerably. Among the climatic risks that most affect them are frost,
hail, drought and heat waves (Climate change and vineyards in Spain report, 2016).
Therefore, determining the relationship between climate and vineyard and assessing its
future evolution is of particular interest in regions such as Spain, where the wine sector
is not only important in terms of biodiversity but also socio-economic terms.

It is possible to evaluate the relationship between climate/weather and the different

factors affecting vines and wine production as a whole by using bioclimatic indices
(Fregoni 2003), which make it possible to determine the climatic suitability of a region for
growing vines, the most suitable variety or the possibility of the occurrence of certain
pests and/or diseases.

Classical studies use individual indices calculated and derived from temperature and
precipitation (Carbonneau and Tonietto, 1999; Tonietto 1999, Fraga and Santos, 2017)
to assess climate-vineyard relationships (Bindi et al., 1996, Jones 2006) and their impact
from different perspectives (Schultz, 2000, Combris et al., 1997).

More recent studies have highlighted the need to work with combined indices as they
represent more complete viticultural classification and discrimination and allow wine
quality to be characterised (Huglin, 1978; Maglhaes, 2008). This way of working is
included in the concept of viticultural zoning and is the first step to evaluate the viticultural
potential of a region (Malheiro et al., 2010).

Despite many indications and reports on the effects that climate change has already had
on the wine sector, the efforts of the scientific community to identify the relationships
between weather variables and vines and the impact of these changes in the future,

151
there are still few studies focused on how wine producers and growers can adapt to
these changes (Holland et al., 2010).

In addition, most studies use dynamic climate projections (which have not taken into
account local climatology), direct outputs from climate models or a reduced number of
Spanish locations, as they are part of studies covering larger geographical areas. To
date, no study assesses the impact of climate change on vineyards in the Ibero-balear
Spanish territory using bioclimatic indices calculated based on regionalised climate
projections on a local scale with a statistical downscaling technique (considering local
climatology) generated from climate models belonging to the fifth phase of the CMIP5.

Therefore, this study aims to generate local future climate scenarios for the twenty-first
century for four individual bioclimatic indices of viticultural impact: Huglin Index (HI),
Dryness Index (DI), Cool night Index (CI) and Branas, Bernon and Levadoux
Hydrotermal Index (HyI), a combined index (CompI) and a viticultural zonation (MCC
System) for the Iberian-Peninsular Spanish territory. As a starting point, local daily
climate projections generated through a statistical downscaling technique fed with
CMPI5 scenarios will be used.

This study will make it possible to assess the suitability of the study area for wine-
growing, as well as to determine what areas are going to lose or gain wine-growing
potential, which will be very useful information for defining possible adaptation measures
and decision making for the wine-growing sector in the face of climate change.

2. Materials and Methods

2.1. Study area
This study was carried out on the Spanish peninsular territory and the Balearic Islands
(Fig.R3.1) covering an area of 588,294 km2. Due to its location close to large bodies of
water and its complex orography (from sea level to peaks exceeding 3400 km), the
Spanish climate is very varied and complex, so that up to 13 climatic regions can be
counted according to the Köppen classification (Köppen and Geiger, 1936), and there
are multiple microclimates. The climate of the Iberian Peninsula depends on its location
in the extreme southwest of Europe and its complex orography, while the Balearic
Islands are located in the western Mediterranean close to the Iberian Peninsula and are
relatively mountainous.

The Spanish mainland and the Balearic Islands have very optimal climatic conditions
that favour the cultivation of grapevines, as reflected in the almost 950,000 hectares of
Spanish territory dedicated to its cultivation.

The Canary Islands have not been included in the study because, due to their climatic
characteristics, a consequence of their location in tropical areas, they differ from those
of the rest of the Spanish territory and they deserve an individual study that
encompasses these differences.

2.2. Spanish Denominations of Origin (DOs)

The Spanish DOs are the system used in Spain for the recognition of a differentiated
quality, which is the result of specific and distinguishable characteristics due to the
geographical environment in which the raw materials are produced and the products are

152
made as well as the influence of the human factor involved. In Spain, there is a wide
network of recognised quality according to the Government of Spain (2021): 101
Denominations of Origin (occupying an area of more than 900.000 ha), 42 Protected
Geographical Indications and 26 “Vinos de Pago” (MAPA, 2022 Ministry of Agriculture,
Fisheries and Food).

Figure R3.1. Location of the study Area. a) Shows the points corresponding to the observatories used in the complete
study with available temperature and precipitation data. b) Shows the location of the denominations of origin “(Map
source: OpenStreetMap)”

2.3. Datasets
2.3.1. Surface dataset
The observed data set used in the study consists of a set of time series of daily maximum
and minimum temperature and daily precipitation data homogeneously distributed
throughout the territory and belonging to the network of observatories of the Spanish
Meteorological Agency (AEMET, www.aemet.es).

The selected dataset is the same that has been used in previous studies in the
generation of local future climate scenarios for the study area (Gaitan et al., 2019; Gaitan
et al., 2020; Gomez-Martinez et al., 2021; Monjo et al., 2016; Ribalaygua et al., 2013b),
which have been subjected to strict quality control (p.e. inhomogeneities, gaps and
outliers, Lopez et al., 2007).

A total of 1778 observatories with data of both temperature and precipitation were used
in the study, covering extensively the entire territory under study (Fig.R3.1a). For the
analysis of the results by Denominations of Origin (DOs), we have chosen those
observatories that are located within the territory classified as such (Fig.R3.1b). In total,
there are 789 observatories within 59 DOs located in the iberian-balearic territory (the
DOs where no observatories with information of temperature and precipitation
simultaneously or with poor meteorological information were found are left out of the
present study).

2.3.2. Local future climate scenarios

153
A set of daily local future climate projections of temperature (maximum and minimum)
and precipitation obtained by applying a two-step analogue/regression statistical
downscaling methodology developed by the Climate Research Foundation (FIC) was
used (Ribalaygua et al., 2013a; Ribalaygua et al., 2013b). This methodology offers some
advantages: it is computationally inexpensive, provides local information at observatory
scale and allows quantifying the uncertainty associated with the downscaling process
(Van der Linden and Mitchell, 2009). Other advantages are the application of future
simulations consistent with observations (physically coherent between them) and using
local scale (because nearby data points in space are subjected to different climate
change conditions) (Ribalaygua et al., 2013b).The generation of daily future climate local
scenarios was based on nine global climate models (Table R3.1), called Earth System
Models (ESMs, (Wang et al., 2010), belong to the fifty phase of the Coupled Model
Intercomparison Project (CMPI5, Tripathi et al., 2006) and supplied by the Program for
Climate Model Diagnosis and Intercomparison (PCMDI) archives. This generation of
models has contributed to the acquisition of both dynamical and statistical downscaling
techniques with less uncertainty in integrating the individual parts of the climate system
(atmosphere, ocean, land and sea ice) and the exchange of energy and mass between
them (Knutti and Sedlacek, 2013).

2ºx2,5º
National Oceanic and Atmospheric
GFDL-ESM2M Dunne et al. (2012)
Administration (NOAA), E.E.U.U.
daily

Canadian Centre for Climate Modeling

2,8ºx2,8º
and Analysis (CC-CMA),
CanESM2 Chylek et al. (2011)
daily
Canadá.

1,4ºx1,4º CNRM (Centre National de

CNRM-CM5 Recherches Meteorologiques), Voldoire et al. (2013)
daily Meteo-France, Francia.

1,4ºx1,4º
Beijing Climate Center (BCC), China
BCC-CSM1-1 Xiao-Ge et al. (2013)
Meteorological Administration, China.
daily

1,87ºx1,25º
Met Office Hadley Center, United
HADGEM2-CC Collins et al. (2008)
Kingdom.
daily

Japan Agency for marine-Earth

Science and Technology (JAMSTEC),
2,8ºx2,8º
Atmosphere and Ocean Research
MIROC-ESM-CHEM Watanabe et al. (2011)
Institute (AORI), and National Institute
daily
for Environmental Studies (NIES),
Japan.

1,8ºx1,8º Raddatz et al. (2007);

Max-Planck Institute for Meteorology
MPI-ESM-MR
(MPI-M), Germany.
daily Marsland et al. (2003)

154
 1,2ºx1,2º
Meteorological Research Institute
MRI-CGCM3 Yukimoto et al. (2011)
(MRI), Japan.
daily

2,5ºx1,9º Bentsen et al. (2012);

Norwegian Climate Centre (NCC),
NorESM1-M
Norway.
daily Iversen et al. (2013)

Table R3.1. Information about the nine climate models belonged to the 5 Coupled Model Intercomparison Project
(CMIP5) corresponding to the fifth report of the IPCC. Models were supplied by the Program for Climate Model
Diagnosis and Intercomparison (PCMDI) archives.

This study uses data from two different experiment families of GCMs: the Historical
experiment (Taylor et al., 2012), which covers much of the industrial period and can be
referred to as “twentieth-century” simulations and the representative concentration
pathway (RCP) family (Moss et al., 2010), which corresponds to different possible ranges
of radiative forcing reached in the year 2100 for values of the pre-industrial era. This
study uses future projections determined by the RCP8.5 ‘high’ scenario and the RCP4.5
‘intermediate’ scenario.

In total, there is a set of 18 daily climate projections for two emission scenarios, RCP4.5
and RCP8.5 (9 projections for each RCP).

The methodology employed for generating temperature and precipitation projections has
been used in national and international projects, with good verification (Gaitan et al.,
2019; Monjo et al., 2016; Moutahir et al., 2017; Rodriguez et al., 2014; Santiago et al.,
2017, Gutierrez et al., 2019, Ribalaygua et al., 2013a, 2013b, 2018) and validation
results (Ribalaygua et al., 2013a, 2013b, Gaitan et. al, 2019, 2020; Monjo et al., 2016).
Verification results (goodness of the methodology used) obtained in the above-
mentioned studies showed good results for both, temperatures and precipitation. In the
case of the temperature, the average bias achieved was below 0.1 ºC while for
precipitation, an error of 10-20% was obtained. Validation results obtained in the above-
mentioned studies for both the maximum and minimum temperatures, showed a bias of
around tenths of a degree in all months, so they were very close to zero. The error was
not above half of a degree for any of the cases. Therefore, the results showed that the
ESMs were capable of adequately simulating both the maximum and the minimum
temperatures on annual and seasonal scales. In the case of the precipitation, the results
are variable depending on the model and the seasonal period; however, all the models
are able to reproduce the annual cycle of precipitation as well as the differences between
seasonal periods (maximum values in autumn and spring, followed by winter and
summer). The obtained bias and standard deviation are less than ±1 mm/day, which in
relative terms supposes a difference of less than or around ±10% in the worst of the
cases.

From the simulated temperature series, future heat and cold wave episodes have been
calculated following Gaitán et al., 2019. Heat Waves have been defined at least three
consecutive days with a maximum temperature above the 95th percentile of the
maximum temperature series calculated between the months of June to September
during the period 1971–2000 and at least three consecutive days with a minimum
temperature below the fifth percentile of a minimum temperature series and calculated
between the months of November to April during the period 1971–2000.

155
2.4. Bioclimatic indices
A set of four bioclimatic indices (Huglin Index (HI), Dryness Index (DI), Cool night Index
(CI) and Branas, Bernon and Levadoux Hydrotermal Index (HyI)) was used to assess
the impact that climate change may have on the suitability of a region for growing
grapevines and/or certain grape varieties. In addition, two combinations of these indices
were analysed: MCC System and CompI. For a complete explanation of indices’
definition see Table R3.2.

The Dryness Index (DI) assesses soil water availability by providing information on water
stress conditions. In the absence of information on future land use and other variables,
it was decided to use a simplified formula proposed by Tonietto and Carbonneau, 2004
and based on the calculation of the potential evapotranspiration (ETP). According to
various studies (Blanco-Ward et al., 2007; Vanderlinden et al., 2004) and specifically
Fonseca et al 2012, the Hargreaves formula was chosen for the calculation of ETP
instead of other more complex formulations.

156
Table R3.2. Description of analysed Bioclimatic indicators

157
3. Results
3.1. Observed bioclimatic indices and their verification
The observed (categorised) absolute average values of the indices used in the study (HI,
DI, CI, HyI, MCC System and CompI) for the period 1971–2000 can be seen in Fig. R3.2
and Support Information R3.1. The results clearly show a north-south and west-east
spatial distribution.

Within the Iberian-Balearic territory, we can find HI values (Fig.R3.2a and S3.1a) that
cover all the categories defined for this index, from 300 to about 3300°C, accumulated
in the period from April to September.

Some regions with climatic characteristics not suitable for growing grapes were detected,
either because they are too cold with HI < 1000 (Pyrenees) or too warm with HI > 3000
(some points in the Spanish Southwest).

The areas of the northern peninsula as well as most of the valleys of mountainous areas
have characteristics of cold climates with HI values between 1500 and 1800 (category
2). In a smaller proportion, there are regions with temperate climates (HI 1800–2100).

In the mountainous regions of the Central System, the Iberian System and Betic System,
we find HI values < 1500 (category 1, very cold), which places them in the lower thermal
limit for grapevine. However, most of the territory of Sierra Morena (the mountain range
that runs from east to west in the south of the Iberian Peninsula) presents IH values that
oscillate between 2700 and 3000 accumulated degrees (category 6), being areas with
very warm climates.

The coastal areas of the Mediterranean and the plateau as well as Cádiz (SW Iberian
Peninsula) are characterised by being warm areas (category 5) with a high heliothermic
potential (HI between 2400 and 2700).

The values observed for the CI (Fig.R3.2b and S13.b) in the ripening month (average
minimum temperature in September) range between 4 and 20°C, which is a great
difference between regions within the study area. Most of the northern half of the
peninsula and the highest points of the Betic system present very cold CI values below
12°C (category 1). The rest of the northern zone and part of the central plateau present
cold CI values between 12 and 14°C (category 2). The Mediterranean coast and the
south and west areas of the peninsula, as well as the Balearic Islands, are characterised
by mild nights (CI between 14 and 18°C, category 3). Very few areas have high minimum
temperatures in September (CI > 18, categories 4 and 5).

The DI values obtained (Fig. R3.2c and S3.1c) vary between –260 mm (very dry areas)
and 500 mm (super-humid areas). The Cantabrian coast (N) presents the most humid
conditions with DI values > 150 mm (category 5), while the Cantabrian mountain range
presents DI values higher than 50 mm (category 4), which implies humid and sub-humid
characteristics with an absence of drought and a high level of water availability. Most of
the northern plateau and the Balearic Islands have DI values between 50 to –100 mm
(category 3) and are considered moderately dry areas. The rest of the peninsula is
characterised by traditionally dry regions (DI between –100 and –200 mm).

158
The observed HyI values (Fig.R3. 2d and S3.1d) show that the southern regions have
the lowest risk of incidence of diseases such as mildew (which depend on humidity and
temperature to a great extent) and that the risk gradually increases towards the north
where the precipitations are more abundant during the period of growth of grapevine.

Figure R3. 2. Geographical representation of the observed values of the a) Huglin Index (HI), b) Cool night Index (CI),
c) Dryness Index (DI), d) Branas, Bernon and Levadoux Index (HyI), e) Composite Index (CompI) and f) MCC System
for the periods 1971-2000

159
Considering the aforementioned values, there will be years more suitable from the
climatic point of view than others will. Fig.R3. 2f shows the CompI index based on
thresholds of some of the commented indices (Tmin, HI, DI, CI and HyI, see Table R3.2),
which reveals that the southwest regions of the peninsula have an optimal percentage
of years, climatically speaking, lower than those of the northern half and the Balearic
Islands.

By combining these indices, we can establish within which values of the MCC System
climatic classification the study area falls, since each of the indices separately is not a
guarantee of viticultural climatic suitability. In total, we defined 120 combinations (see
Table S3.1). The northern zone covers the classifications with categories between 1 and
20, the northern plateau belongs to those classifications with categories between 45 and
50, while the Mediterranean and eastern peninsular zones belong to the 70 and90
classes and, finally, the western and southern zones are included in the classes with
categories between 90 and 100.

In the verification results (Fig. S3.2), the indices calculated based on observed data are
compared with the indices calculated based on data from the temperature and
precipitation series of the ERA-40 reanalysis to which downscaling was performed. The
verification shows very acceptable bias values for all the indices analysed. In the case
of HI the mean Bias is around 104 degrees-day (it supposes at relative error of 4%), for
the CI is had been appreciated a mean Bias of 0.16 ºC (corresponding to a relative error
of 1.2%), the DI mean Bias is around -10mm (with a relative high error of 45% due to the
simulation in very aridity places) and finally, the HyI has showed a mean Bias of -10
ºC*mm (it means a relative error of 0.2%).

3.2. Local future climate scenarios to predict bioclimatic indices

Figures R3.3, R3.5, R3.7 and R3.9 show the expected future evolution for HI, CI, DI and
HyI, respectively, in absolute terms starting with the historical reference period (1976–
2005) and followed by correlative periods of 30 years from 2011 to 2100 as presentation
of short, mid and end-century expected values, which have been predicted based on the
nine models (see Table R3.1) and according to the RCP4.5 (top row) and RCP8.5
(bottom row) scenarios.

In addition, the simulated results for the present (Historical period) are displayed to see
the expected changes relative to the current state. In the supplementary material, the
same information is represented, but in a categorised way (Figs S3.3, S3.5, S3.7 and
S3.9).

To examine how the different DOs will be affected, Figures R3.4, R3.6, R3.8 and R3.10
show the expected future evolution for HI, CI, DI and HyI, respectively, considering
exclusively those observatories that are located within some DO (Fig. R3.1). The DOs
have been represented following a geographical order (north at the bottom of the figure-
south on the top). The complementary material represents the same information but in a
categorised manner (Figs. S3.4, S3.6, S3.8 and S3.10)

160
Figure R3.3. Geographical representation of the expected values of the Huglin Index (HI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are represented: RCP4.5
(figures b, c and d) and RCP8.5 (figures e, f and g). Fig. R3.3a represents the Historical absolute temperature for the period 1976–2005

161
In general, all the indices analysed showed a main north-south spatial distribution and a
secondary west-east distribution, which, although it fades, tends to remain throughout
the twenty-first century.

The entire Iberian-Balearic territory tends towards warmer climates, so that a large part
of the territory will show movements towards increasingly warmer HI categories (Fig.
R3.3 and S3.3), especially in the last section of the twenty-first century and in the most
extreme case of RCP8.5. The highest areas of the territory will go from too-cold climates
to optimal climates for any type of grape variety. All ODs show progressive increases in
HI throughout the twenty-first century (Fig. R3.4 and S3.4) as evidenced by the gradation
of the red colours in the graphs. Although in terms of categorisation it seems that the
different territories will not undergo heliothermic variations (as is the case of the DO of
the southern territories such as Andalusia or Murcia, see Fig. R3.3 and S3.3), the
absolute values reflect these changes perfectly (see Fig. R3.4 and S3.4).

Figure R3.4. Evolution of the expected values of the Huglin Index (HI) for the defined Denominations of origin. Historical
period 1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are
represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure R3.1 for number
identification) and each column represents a considered period.

162
Figure R3.5. Geographical representation of the expected values of the Cool Index (CI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are represented: RCP4.5
(figures b, c and d) and RCP8.5 (figures e, f and g). Fig. R3.5a represents the Historical absolute temperature for the period 1976–2005

163
Under the RCP4.5 scenario, the DOs of the País Vasco are those that start with the
lowest HI values, remaining at the end of the century with average HI values. It is followed
by the DOs of Galicia and Castilla y León, which gradually become warmer, and they will
be the ones that change the most in categorisation throughout the twenty-first century.
Under conditions of RCP8.5, the evolutions are much more pronounced, so that almost
all DOs will be at the end of the twenty-first century under climates that are too hot,
heliothermally speaking, for the cultivation of grapevine.

FigureR3. 6. Evolution of the expected values of the Cool Index (CI) for the defined Denominations of origin. Historical
period 1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are
represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure R3.1 for number
identification) and each column represents a considered period.

164
Figure 3.7. Geographical representation of the expected values of the Dryness Index (DI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are represented: RCP4.5
(figures b, c and d) and RCP8.5 (figures e, f and g). Fig. R3.7a represents the Historical absolute temperature for the period 1976–2005

165
It is expected that the CI will increase throughout the twenty-first century in the entire
territory studied by at least 1°C (Fig. R3.5 and S3.5), so that those regions that are at
the upper limit of any of the categories pass to be included in the category immediately
above. In general, all DOs are expected to vary between 3 and 4°C between current CI
values and those expected at the end of the twenty-first century in the RCP4.5 scenario
(Fig. R3.6a and S3.6b) and between 6 and 8°C under RCP8.5 conditions (Fig. R3.6b
and S3.6b). The absolute values of the CI show how this index is expected to increase
progressively throughout the twenty-first century. Under the conditions of RCP4.5, the
DOs of Andalucía, Murcia and Cataluña are those that are expected to have higher CI
values. Cooler nights in September are expected on the other side, for Euskadi, Castilla
y León and Valencia.

Figure R3.8. Evolution of the expected values of the Dryness Index (DI) for the defined DOs. Historical period 1976-
2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are represented:
RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure R3.1 for number identification) and
each column represents a considered period.

166
In the future, it is expected that the DI tends to become increasingly dry values because
of the increase in temperatures, and therefore of evapotranspiration, and that most of
the peninsula and the Balearic Islands are at risk of water stress (Fig.R3.7 and S3.7).
The northern and Mediterranean areas will be the ones that least accrue these changes,
staying in humid or not very dry climates. The expected impact on the water balance
(Fig. R3.8 and S3.8) shows that only the DO located in the País Vasco will maintain the
status of a humid region in the coming decades. Most of the DO will remain at similar
hydrological regimens, although the scenarios show a trend towards drier
characteristics.

Although the HyI values increase in the coming decades, they remain within the range
of medium risk of the presence of diseases such as mildew (Fig. R3.9 and S3.9). Most
of the DOs (Fig. R3.10 and S3.10) will be at a low average risk of mildew presence
throughout the twenty-first century. The DOs of Andalucía and Murcia are expected to
present a low risk of the presence of mildew since their survival is not favoured due to
the low precipitation. The DOs of the País Vasco, mainly, and those of Galicia will have
the highest risk of suffering from mildew because of the increase in temperatures
combined with the increase in the precipitation regime in these regions.

Finally, it is expected that in the coming decades the percentage of climatically optimal
years will decrease throughout the territory (Figs. R3.11 and R3.12) a consequence
mainly of variations in HyI and DI.

The analysed indices represent the expected changes in climatic characteristics

associated with average variables and not with extreme events, such as extreme rainfall,
drought or Heat/Cold Waves episodes. The latter, have a strong impact on the vine
depending on the phenological stage of the vine at which they occur.

The increase in maximum temperatures will lead to a greater occurrence of heat wave
episodes, as well as an increase in their duration, their average intensity and the
maximum intensity reached within each heat wave episode (figures S3.13 y S3.14).

One of the most affected area will be the Mediterranean coast, where the average
duration of a heat wave episode is expected to increase from 9-12 days to more than 18,
increasing the average intensity and maximum intensity by 3-4°C. In the northern part of
the Iberian Peninsula, although it will also suffer an increase in heat waves, this will be
less pronounced than in the rest of the territory, with average increases in duration of 2-
3 days and increases in intensity of 1-2 ºC. In the peninsular plateau and southern zone,
the average duration of heat waves is also expected to increase by about 6 days (from
9 to 15 days in duration) with the increase in average and maximum intensity reached
during these episodes (between 3-5 ºC) (see supporting information). These results are
in line with the conclusions obtained by Molina et al., 2020 in the Mediterranean area,
Torres et al., 2021 in the Balearic Islands and by Abaurrea et al., 2018 for the Iberian
Peninsula.

The expected increases in minimum temperature will not prevent the occurrence of cold
wave episodes (figures S3.15 y S3.16), although the average duration of cold wave
episodes is increasingly shorter.

167
Figure R3.9. Geographical representation of the expected values of the Branas, Bernon and Levadoux Index (HyI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios
are represented: RCP4.5 (figures b, c and d) and RCP8.5 (figures e, f and g). Fig.R3. 9a represents the Historical absolute temperature for the period 1976–2005

168
 Figure R3.10. Evolution of the expected values of the Branas, Bernon and Levadoux Index (HyI) for the defined DOs.
Historical period 1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions
scenarios are represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure R3.1 for
number identification) and each column represents a considered period.

4. Discussion
This study analyses the evolution of climatic suitability for vine cultivation on the Spanish
Mainland and the Balearic Islands based on a set of individual and combined bioclimatic
indices using, for the first time, local future climate scenarios based on ESMs from the
fifth IPCC report. In addition, the nine climate models available under two RCPs, provide
a set of future climate projections of 18 possible future evolutions, which allows taking
into account uncertainties, as recommended by various authors (Christensen et al.,
2010; Fraga et al., 2014; Weigel et al., 2010).

169
Figure R3.11. Geographical representation of the expected values of the Composite Index (CompI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are
represented: RCP4.5 (figures b, c and d) and RCP8.5 (figures e, f and g). Fig. R3.11a represents the Historical absolute temperature for the period 1976–2005

170
Moreover, there are strong differences in assessing climate impact at the regional or
local level (Santos et al., 2012). Local studies allow us to establish the origin of the main
differences between grape types grown in neighbouring regions as suggested by Ramos,
Yones and Juste (2017) and which is evident in the results obtained when considering
the future climate scenarios of the bioclimatic indices by denominations of origin. These
results reinforce the importance that climatic conditions have on the genuineness and
unique character of each designation of origin.

The generation for the first time of future scenarios of bioclimatic indicators of great
interest for the wine sector for the 21st century at local scale (considering local
microclimatic characteristics) with a wide set of future climate projections using ESMs
from the fifth IPCC report, brings novelty to the studies existing so far in the sector.

Therefore, these results offer one of the best snapshots of future climate change, based
on currently available data, and the risks that changes in temperature and precipitation
regimes could cause in the way grapevines are cultivated.

4.1. Considerations about the simulation of bioclimatic indices

The future climate scenarios at the local scale used as a basis to generate the future
scenarios of bioclimatic indices were developed by the FIC with a two-step analogue
methodology (Ribalaygua et al., 2013a), which has been verified and validated in various
studies in Spanish territory with very good results (Gaitan et al., 2019; Gaitan et al., 2020;
Gutierrez et al., 2019; Monjo et al., 2016; Ribalaygua et al., 2013b). Therefore, the future
temperature and precipitation scenarios on which the bioclimatic indicators are obtained
are robust and reliable. Moreover, it should be noted that there are studies that reinforce
the idea that daily changes in atmospheric conditions play an important role in plant
phenology (Jones and Davis, 2000), and these changes can be more or less significant
depending on the region where they occur. These aspects are considered intrinsically in
the type of downscaling’s methodology used in this study.

Another factor to consider is the benefits that an increase of CO2 under future climate
conditions plays an important role in the development of the vine (Bindi et al., 2001;
Goncalves et al., 2009; Moutinho-Pereira et al., 2009). Although this study does not
directly analyse this point, it is taken into account when considering different RCPs.

4.2. Local future climate scenarios of bioclimatic indices

In general terms, there is a positive trend in all thermal indicators (HI and CI) and a
negative trend in the water index (DI). This reflects the twenty-first century with
progressive increases in temperatures, both maximum and minimum, throughout the
Iberian Peninsula, while hardly any changes in precipitation patterns are expected. The
combination of the expected changes in both variables will have a strong impact on the
vineyard. Similar results have been obtained for areas such as Portugal (Fraga et al.,
2012), Italy (Bonfante et al., 2017, 2018), Germany (Neumann and Matzarakis, 2011;
Stock et al., 2004), France(Duchene and Schneider, 2005, Duchene 2016, Garcia-
Cortazar et at., 2017) and Spain (Gomez-Gesteira et al., 2011, Ramos 2017).

171
Figure R3.12. Evolution of the expected values of the CompI Index (CompI) for the defined DOs. Historical period 1976-
2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are represented:
RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO and each column represents a considered
period.

Consequently, although the northern regions will see their climatic suitability for growing
grapes favoured, certain regions of the south and southwest of the Peninsula as well as
the Mediterranean coast, that are at the limits of climatic suitability, may be negatively
affected. These results confirm those obtained by (Fraga et al., 2012; Resco et al., 2016),
among others.

The tendency of the HI values to increase is already in itself a determining factor of the
variety of grape that can be grown in each zone since they determine the requirements
for heat accumulation so that the ripening of the grape occurs optimally. The gradual
change of the entire territory towards warmer climates will cause the necessary levels,
in terms of heat, for the ripening of the grapes to occur at earlier times, which translates
into an advancement of the ripening date (Molitor and Junk, 2019). This can subject
vines to heat stress episodes in many regions. In addition, this index has a strong

172
correlation with the different phenological states associated with warm conditions(Bock
et al., 2011; Jones et al., 2005; Santos et al., 2012) so its increase would imply overtaking
of certain phenological properties, which would alter the phenological cycle of the vine.

In this way, in areas that are too cold (according to our results, the Pyrenees area will be
the only one that presents these characteristics), only very early varieties could reach
maturity, usually white varieties. These regions should opt for a hybrid or American
varieties, more resistant than the Vitis vinífera, while in the cool regions (high elevation
mountainous regions, (category 3) both white and red can be grown. As the climate
becomes more temperate (categories 4 and 5, some regions of the northern Peninsula
under RCP4.5 and very few areas of the North under RCP8.5), almost any type of grape
such as ‘Garnacha’ or ‘Moscatell’ can be grown in the first case and ‘Pinot Blanc’, ‘Pinot
Noir’ or ‘Chardonnay’ in the second.

The greatest impact will occur in hot or very hot regions (categories 6 and 7), most of the
Peninsular territory and the Balearic Islands, where the minimum requirements that the
different grape varieties need to ripen, including those with late-ripening, will be
exceeded. In very hot climates, there is a high risk of stress due to heat accumulation
that can be detrimental to grapevine (most of the study territory, especially under the
RCP8.5 scenario). HI values in the northern peninsular plateau are expected to be lower
than those expected under RCP4.5, while the southern plateau and the Balearic Islands
will reach very high HI values, regardless of the scenario considered.

In addition, it must be considered that within the same category it may be that each
variety has different heat requirements to reach maturity (Tonietto and Carbonneau,
1999) and that despite belonging to the same category they could not be cultivated. For
example, the ‘Cabernet Franc’ variety requires an HI of 1800 cumulative degrees while
the ‘Cabernet Sauvignon’ variety requires a HI of 1900 and the ‘Ugni Blanc’ variety needs
a HI of 2000. Although their heliothermic requirements are different, both would be in HI
Category 4. Therefore, if we only consider categorised values, we can run the risk of
selecting varieties in areas that do not reach the necessary calorific requirements for
said variety, hence the need to work with absolute values.

Regarding the CI, the expected increase in minimum temperatures will cause the CI to
increase so that in some regions the night coolness necessary for optimal grape ripening
will not be achieved. In addition, the advancement of the ripening dates suggests the
need to evaluate this index on dates before September (Ramos et al, 2021).

Changes towards very cold night temperatures can have a positive effect on certain
varieties of grapes as long as a sufficient heliothermic contribution is guaranteed to
guarantee a good level of ripening of the berries. It is not expected that any study region
will experience decreases in CI under RCP8.5, but in the case of RCP4.5, areas of the
Pyrenees and the Cordillera Cantabria will remain cold at night throughout the twenty-
first century.

In those regions with fresh or medium CI values (such as what is expected to occur in
most of the Southern Plateau, according to RCP4.5, and in the Northern Plateau, under
both RCPs), the results can be both positive as well as negative depending on the
cultivated variety, as the late varieties ripen in colder conditions than the early ones.

173
Finally, if the CI values are higher than 18°C (as will be the case of the Atlantic and
Mediterranean coast and the Balearic Islands if RCP4.5 is considered or of the entire
Southern Plateau plus the regions previously mentioned according to RCP8.5), the vines
can suffer an excess of heat that affects the colour and aromatic potential of the grape.

The changes in the expected DI values condition the region's water supply and,
therefore, determine the decisions to be made regarding irrigation. In the entire Iberian-
peninsular territory except for the Cantabrian coast and the Pyrenees, regardless of the
RCP considered, the DI values will decrease to the lower limit of 50 mm, so these
territories will be at the lower limit of water supply, which may give rise to certain
restrictions, especially in the summer months. If DI values < –100mm, the region will be
excessively dry, requiring an extra water supply.

On the contrary, values higher than 150 mm (as could occur in the areas of the Pyrenees
or the Bay of Biscay) can reduce the quality of the wines and have higher quality grapes
in wet years.

Intermediate DI values (between 50 and –100 mm) that are expected in regions such as
Galicia or Asturias, will suffer certain periods of drought that can become favourable
during ripening.

The joint assessment of these indices through the MCC System allows establishing a
more complete climatic vision of the suitability of the region. Of all the possible
combinations, those that are more suitable for grapevine cultivation are those that
combine HI values (categories HI-3, HI-2, HI + 1), CI (categories, CI + 1; CI +2) and DI
(categories DI-1, DI + 1). While the least optimal have turned out to be those with CI
(categories CI-2, CI-1) and DI (categories DI + 2, DI + 3). In the particular case of this
study, it is expected that the optimal regions under this criterion will be found in the north
of the peninsula, such as the Cantabrian Mountains and the Pyrenees, as well as in
almost the entire Northern Plateau (according to both RCPs in the middle of the century).
However, the Northern Plateau will only maintain these conditions under RCP4.5. These
results are in line with those obtained by other authors (Fraga et al., 2013; Resco et al.,
2016).

The combination of moderately low night temperatures with high daytime temperatures
in these areas will favour the production of high-quality wines since the synthesis of some
phenological components is favoured.

Regarding Hyl, no major changes are expected in the risk of certain diseases such as
mildew, but there may be many differences between regions with very different
precipitation patterns. The southern regions have a lower risk, which gradually rises to
the north where precipitation will be more abundant during the vine growing season,
similar to what other authors found (Fraga et al., 2013; Lazoglou et al., 2018).

The results obtained in the CompI index may seem contradictory since the largest
Spanish wine-growing regions, such as Andalusia and Castilla-La Mancha, have a very
low percentage of optimal years for growing vines. Similar situations have been found in
other studies (Guido, 2015). This is because they are dry regions or with strong periods
of drought with DI values < –100 mm, while one of the conditions to consider a
climatically optimal year is that the DI > –100mm. The results of this index must be
interpreted in the light of multiple socioeconomic factors that determine the success of a

174
vineyard plantation beyond the climatic conditions. For example, in these regions, this
situation is solved by viticultural producers through different management strategies and
water management, which allow solving this “climatic problem” and taking full advantage
of the rest of the climatic characteristics that favour grapevine. Other factors such as the
type of cutting used, the orientation of the vineyard, the field management tasks, the type
of soil, among others, will be keys to adapting to the challenges posed by climate change
(Alexandrei et al., 2013).

It should also be kept in mind that rising temperatures due to climate change may have
indirect effects on these crops, as they are expected to cause an increase in tropospheric
ozone concentrations and are also likely to affect the chemistry of ozone precursors
(NOx, CO, CH4, NMHC) (Isaksen and Wang, 2002). This modification of atmospheric
pollutant generation can be very detrimental to vineyards. For example, it has been
suggested that ozone can cause a loss of productivity and a reduction in the sugar
content of grapes (Ascenso et al., 2021). Increased exposure to SO2, NO2 can cause a
severe reduction in photosynthetic rate, transpiration and stomatal conductance in shoot
growth (Popescu et al., 2012).

To all these effects must be added the impact of extreme phenomena, especially heat
waves. Heat waves (see figures AR1.5 and AR1.6) combined with summer drought will
be the most common abiotic stress combination in the Mediterranean area (Hannah et
al., 2013). The response of grapevines to increased temperatures (acceleration of their
key phenological stages affecting grape quality and the properties of grape organoleptic
components, such as sugar accumulation, pH, acidity, color, aroma and flavor) (Ramos
et al., 2008, Leoni et al., 2019), is likely to increase with heat waves and will also depend
on its coincidence with grape ripening (Sgubin et al., 2018).

In more humid areas, as the northern part of the Peninsula, new growing areas may
become viable, for example in areas of higher altitude (Ramos and Martinez de Toda,
2021) or closer to the coast (Santos et al., 2020a), while low elevation areas would
probably be suitable for lower quality varieties, producing wines of high alcohol content
(Moriondo et al., 2007).

In almost all the territory (where a significant intensification of heat wave episodes is
expected, see support information) all adaptation options must be considered if current
crops are to be maintained, such as water application, row orientation or canopy cover.
Water availability is the factor, along with high temperatures, that most affects vine
development (Fraga et al., 2018, Fraga et al., 2019). In these areas where irrigation
water is not available or is too warm, such as the Guadalquivir Valley or Extremadura, it
will not be possible for the vines to mature normally, so it will be necessary to make
substitutions towards varieties more tolerant to the new climatic conditions (Ramos et
al., 2008). The search for adapted grapevine (Vitis vinifera L.) varieties will be a priority
in the coming years, either through germplasm collections or genetic improvement
processes (Duchene et al., 2012).

At the other extreme, cold wave episodes can be especially damaging to primary buds
(Gu et al., 2002), although there are not expected to be considerable variations in the
average and/or maximum intensities of such episodes in Spain (figures AR1.7 and
AR1.8).

175
Finally, an adaptive evolution with physiological modifications of the wineyard could be
expected (Ramos and Martinez de Toda, 2021) especially in those areas where climatic
extremes are not so intense or so frequent, as in the northern part of the Peninsula. The
adaptation of each variety under the same climatic conditions depends on the
peculiarities of each genotype to heat, light or water deficit and the temperature and
humidity conditions needed for ripening. The varieties with earlier phenology will be
probably the most affected as was described for the Spanish variety “Tempranillo”
(Ramos and Martinez de Toda, 2020).

The literature highlights different biochemical, physiological and molecular acclimation

mechanisms that grapevine is able to develop in order to adapt to climatic stresses.
Changes in photosynthetic efficiency and in the control of electrolyte loss through
stomata appear to be frequent mechanisms of adaptation (Zha et al., 2018). For
example, adjustment of photosynthesis to elevated temperature has been detected in
some cases (Gallo et al., 2021; Kizildeniz et al., 2021). Site-specific stomata and vein
traits modulation have been suggested as an acclimation strategy that may influence
photosynthetic yield (Damiano et al., 2022). Increasing the heat dissipation capacity by
changing the response of its stomata has also been proposed as an adaptation to
warmer climatic conditions, since keeping them open allows heat dissipation by
evaporative cooling (Costa et al. 2012). Significant changes in the expression pattern of
metabolic pathways related to metabolism and hormones have also been detected
(Duchene et al., 2012; Kovaleski and Londo, 2019). This opens up opportunities to
identify the genetic and physiological traits that make a variety more or less resistant and
select varieties which would be suitable to replace the more sensitive ones even in the
most affected areas such as those in the south or in the central zone of the country. In
Ronda (Málaga, South of the Spain), Petit Verdot (originally from Bordeaux), whose long
phenological cycle is perfectly adapted to warm climates, is producing good results in
recent experiences. In Extremadura (Southwest of the Spain), Portuguese varieties such
as Trincadeira or Touriga National are also being successfully cultivated. In Ribera del
Duero (central area of Spain) there is a growing interest in growing Malbec (traditionally
grown in areas of Castilla-La Mancha and Castilla y León above 1000 meters) to
accompany Tempranillo (traditionally accompanied by Cabernet).

Another option is to include new sub-varieties within a variety rooted in the area, as is
the case in the region of Murcia (South of the Spain), where the cultivation of four new
Monastrell grape varieties (Gebas, Myrtia, Calnegre and Calblanque), which are more
resistant to the new climatic conditions, has recently been approved.

However, the process of incorporating new varieties is complex, not only because it
depends strongly on local characteristics (to the meteorological-climatic context must be
added the type of soil, orientation, slope of the land and investments in irrigation or other
adaptive technologies, among others), but also because it is subject to the legislation in
force in each area and the long process involved in incorporating a new variety in a
region from which it does not originate.

4.3 Relevance of the results

The results obtained in this study show how the Iberian peninsular territory is one of the
wine-growing regions most sensitive to the impact of climate change, not only because
of the significant variations in the values obtained in the study, especially in those

176
dependent thermal indices, but also because these variations place the region within the
optimal thermal limits for growing vines. Other European regions such as France and
Italy are also expected to suffer variations in their climatic conditions with implications for
the vineyard, but the expected impact is not estimated to be as marked (Fraga et al.,
2013), and the regions in Northern Europe will even benefit from the expected climate
changes (Hannah et al., 2013).

Alterations in climatic requirements (heliothermic and hydric) have a strong impact on

the final organoleptic characteristics (sugar, acidity, colour, etc.) of the grape and the
characteristics of the wine. Spain has a long viticultural tradition, its wines being
recognised worldwide precisely for the characteristics of the wines belonging to each
DO, maintaining those own characteristics is of vital importance to guarantee the
continuity of the DOs.

Currently, there is a shortage of agricultural models for grapevine (Bindi et al., 1996) or
for the quality of grapevine (Webb et al., 2008) as well as tools to support decision-
making (Iglesias et al., 2012; Santos et al., 2012). There are some soil suitability models
(Escariz et al., 2007) and cereal modelling (STICS, BRIN, WANG04) that combined with
a good observed phenological database (Mosedale et al., 2016) and climate projections
such as those presented in this study, which would considerably facilitate the adaptation
of the wine sector to climate change.

The results obtained can be a starting point to review and update certain factors that
under new climatic conditions may be altered. As an example, currently in Spain, only
those varieties that are in the register of Commercial Varieties of the Vine of Spain can
be cultivated and for the control of plantations the List of Authorised, Recommended and
Plant Conservation Varieties is used, both listings could be altered due to new climatic
conditions.

The results of this study should be complemented with other limiting factors (White et al.,
2006) such as orientation, latitude, longitude, altitude, topography and proximity to water
areas as well as orientation and exposure; characteristics that, together with the
properties of the soil, the management practices and the iterations between all the
factors that make up the system provide grapevine and wine with unique qualities that
differentiate them, even within the same DO where different grape varieties, soil types,
and field characteristics may coexist. Although these factors play an essential role in the
wine creation process, they do not pose as great a challenge as the climate (van
Leeuwen et al., 2004), an aspect on which this study was focused.

Finally, the great variety of grape types that are grown in Spain because of the great
climatic diversity of the territory, and its long viticultural tradition makes it possible for the
studies to be replicated in other regions with very similar climates.

This study has focused on presenting the average climatic conditions that the wine sector
will face in the coming decades, allowing winegrowers to have a snapshot of the new
changes they will have to face. Therefore, the impacts that some extreme events (such
as extreme rainfall, hail, droughts, among others) may have on the sector in the coming
decades have not been included. Following this study, it is necessary to go a step further
by assessing the impact of extreme events as well as the impact on the phenological
stages of the vine through relationships between these and meteorological variables.

177
On the other hand, future studies should be carried out in the small parts of the island
territory not included in this study, such as some iberian-balearic DOs as well as the
Canary Islands.

4.4. Conclusions
Our results offer a precise and rigorous picture of the impacts that climate change will
cause on the grapevine crop in the Iberian Peninsula based on currently available data
applying a set of six bioclimatic indices and using, for the first time, local future climate
scenarios based on ESMs from the fifth IPCC report and working locally.

The Iberian-peninsular territory will be the wine-growing region most sensitive to the
impact of climate change in the twenty-first century as reflected by the results obtained
for the different indices, especially those dependent on thermal and hydric conditions.
Progressive increases in temperatures are expected, both maximum and minimum, as
indicated by the thermal indicators (HI and CI), while the water shortage (DI) will be more
pronounced. The combination of the expected changes of both variables will have a
strong impact on the vineyard, modifying the final organoleptic characteristics in the best
of cases, but also leaving a large part of the Iberian Peninsula within the optimal thermal
limits for growing vines, which it may have dramatic results for grapevine growing

The northern and mountainous areas of the peninsula, having climatic characteristics
colder and wetter than in the south, are expected to benefit from climatic variations that
will allow them to adapt to climate change in a positive way, either by changing the grape
variety or by regulating the water supply, among other measures. The central and
southern areas of the peninsula (with strong periods of water scarcity and high
temperatures) will have problems in maintaining certain types of grape and tillage
techniques. They will not reach the minimum water requirements and will exceed the
thermal requirements that the different grape varieties need to ripen. Therefore, these
territories will be forced to rethink vineyard management techniques that allow them to
counteract these negative effects or assess the economic viability of continuing to
cultivate vineyards in the same regions.

These results provide valuable information for decision making in this sector to develop
adaptation measures to climate change at the local scale.

178
Material suplementario Resultado 3

Figure S3.1. Geographical representation of the observed categorised values of the a) Huglin Index (HI), b) Cool night
Index (CI), c) Dryness Index (DI) and d) Branas, Bernon and Levadoux Index (HyI) for the periods 1971-2000.

Figure S3.2. Results of the verification process for the a) Huglin Index (HI), b) Cool night Index (CI), c) Dryness Index
(DI) and d) Branas, Bernon and Levadoux Index (HyI) for the periods 1971-2000. Boxplots in red correspond to the
calculation of the indices based on observed data and boxplots in blue corresponds to the calculation of the indices
based on Era40 downscaled data. Boxplots containing data from all observatories used in the study.

179
Figure S3.3. Geographical representation of the expected values of the Huglin Index Categories (HI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are
represented: RCP4.5 (figures b, c and d) and RCP8.5 (figures e, f and g). Fig.S3a represents the Historical absolute temperature for the period 1976–2005

180
Figure S3.4. Evolution of the expected values of the Huglin Index Categories (HI) for the defined DOs. Historical period
1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are
represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure 1 for number
identification) and each column represents a considered period.

181
Figure S3.5. Geographical representation of the expected values of the Cool Index Categories (CI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are
represented: RCP4.5 (figures b, c and d) and RCP8.5 (figures e, f and g). Fig. S5a represents the Historical absolute temperature for the period 1976–2005

182
Figure S3.6. Evolution of the expected values of the Cool Index Categories (CI) for the defined DOs. Historical period
1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are
represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure 1 for number
identification) and each column represents a considered period.

183
 25

Figure S3.7. Geographical representation of the expected values of the Dryness Index Categories (DI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are
represented: RCP4.5 (figures b, c and d) and RCP8.5 (figures e, f and g). Fig. S7a represents the Historical absolute temperature for the period 1976–2005

184
Figure S3.8. Evolution of the expected values of the Dryness Index Categories (DI) for the defined DOs. Historical
period 1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are
represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure 1 for number
identification) and each column represents a considered period.

185
Figure S3.9. Geographical representation of the expected values of the Branas, Bernon and Levadoux Index Categories (HyI) for the periods 2011-2040, 2041–2070 and 2071–2100. Both
emissions scenarios are represented: RCP4.5 (figures b, c and d) and RCP8.5 (figures e, f and g). Fig. S9a represents the Historical absolute temperature for the period 1976–2005

186
Figure S3.10. Evolution of the expected values of the Branas, Bernon and Levadoux Index Categories (HyI) for the
defined DOs. Historical period 1976-2005 and 7 correlative periods of 30 years starting on 2011 are showed. Both
emissions scenarios are represented: RCP4.5 (figure a) and RCP8.5 (figure b). Each row represents one DO (see
Figure 1 for number identification) and each column represents a considered period.

187
Figure S3.11. Geographical representation of the expected values of the MCC System for the periods 2011-2040, 2041–2070 and 2071–2100. Both emissions scenarios are represented: RCP4.5
(figures b, c and d) and RCP8.5 (figures e, f and g). Fig. S10a represents the Historical absolute temperature for the period 1976–2005

188
 Figure S3.12. Evolution of the expected values of the MCC System for the defined DOs. Historical period 1976-2005
and 7 correlative periods of 30 years starting on 2011 are showed. Both emissions scenarios are represented: RCP4.5
(figure a) and RCP8.5 (figure b). Each row represents one DO (see Figure 1 for number identification) and each column
represents a considered period

189
 Table S3.1. List of MCC System Categories

MCC System Huglin Index Dryness Index Cool Night

category category (HI) category (DI category (CI)
1 6 5 4
2 6 4 3
3 6 3 2
4 6 2 1
5 6 1 4
6 6 5 3
7 6 4 2
8 6 3 1
9 6 2 4
10 6 1 3
11 6 5 2
12 6 4 1
13 6 3 4
14 6 2 3
15 6 1 2
16 6 5 1
17 6 4 4
18 6 3 3
19 6 2 2
20 6 1 1
21 5 5 4
22 5 4 3
23 5 3 2
24 5 2 1
25 5 1 4
26 5 5 3
27 5 4 2
28 5 3 1
29 5 2 4
30 5 1 3
31 5 5 2
32 5 4 1
33 5 3 4
34 5 2 3
35 5 1 2
36 5 5 1
37 5 4 4
38 5 3 3
39 5 2 2
40 5 1 1
41 4 5 4
42 4 4 3
43 4 3 2
44 4 2 1
45 4 1 4
46 4 5 3

190
47 4 4 2
48 4 3 1
49 4 2 4
50 4 1 3
51 4 5 2
52 4 4 1
53 4 3 4
54 4 2 3
55 4 1 2
56 4 5 1
57 4 4 4
58 4 3 3
59 4 2 2
60 4 1 1
61 3 5 4
62 3 4 3
63 3 3 2
64 3 2 1
65 3 1 4
66 3 5 3
67 3 4 2
68 3 3 1
69 3 2 4
70 3 1 3
71 3 5 2
72 3 4 1
73 3 3 4
74 3 2 3
75 3 1 2
76 3 5 1
77 3 4 4
78 3 3 3
79 3 2 2
80 3 1 1
81 2 5 4
82 2 4 3
83 2 3 2
84 2 2 1
85 2 1 4
86 2 5 3
87 2 4 2
88 2 3 1
89 2 2 4
90 2 1 3
91 2 5 2
92 2 4 1
93 2 3 4
94 2 2 3
95 2 1 2
96 2 5 1

191
97 2 4 4
98 2 3 3
99 2 2 2
100 2 1 1
101 1 5 4
102 1 4 3
103 1 3 2
104 1 2 1
105 1 1 4
106 1 5 3
107 1 4 2
108 1 3 1
109 1 2 4
110 1 1 3
111 1 5 2
112 1 4 1
113 1 3 4
114 1 2 3
115 1 1 2
116 1 5 1
117 1 4 4
118 1 3 3
119 1 2 2
120 1 1 1

192
Como resultado de la presente tesis, se han publicado los siguientes artículos. Los
cuales se adjuntan como Anexo.

 Artículo 1:

Gaitan, E., Monjo, R., Pórtoles, J., Pino-Otin, M.R., 2019. Projection of temperatures
and heat and cold waves for Aragon (Spain) using a two-step statistical downscaling
of CMIP5 model outputs. Sci. Total Environ. 650, 2778–2795.

Factor de impacto de la revista (2019): 6.551

Factor de impacto actual: 10.754
Factor de impacto de los últimos 5 años: 10.237.
Factor de citación de la revista (2022): 10.753
Cuartil: Q1 (26/279) en Environmental Sciences

 Artículo 2:
Gaitán, E.; Monjo, R.; Pórtoles, J.; Pino-Otín, M.R. (2020). Impact of climate change
on drought in Aragon (NE Spain). / Science of the Total Environment 740.
https://doi.org/10.1016/j.scitotenv.2020.140094

Factor de impacto de la revista (2020): 7.963

Factor de impacto actual: 10.754
Factor de impacto de los últimos 5 años: 10.237.
Factor de citación de la revista (2022): 10.753
Cuartil: Q1 (26/279) en Environmental Sciences

 Artículo 3 (enviado):

Gaitán, E. and Pino-Otín, M.R. (2023). Future climate change impacts for the
Spanish wine sector through bioclimatic indicators.

Factor de impacto de la revista (2021): 5.965

Factor de impacto actual: 5.965
Factor de impacto de los últimos 5 años: 5.97
Factor de citación de la revista (2022): 8.5
Cuartil: Q1 (18/94) en Meteorology and atmospheric sciences

193
5. DISCUSIÓN GENERAL

194
En este estudio se han obtenido escenarios climáticos para el Siglo XXI que permiten
analizar diferentes impactos del cambio climático sobre Aragón y el territorio español
peninsular y balear. Por un lado, se ha analizado cómo las variables meteorológicas,
temperatura y precipitación, se van a ver afectadas en las próximas décadas así como
la capacidad y fiabilidad de los modelos climáticos y de las técnicas de downscaling
estadístico para simular el clima pasado y proyectar el clima futuro. En segundo lugar,
se ha estudiado cómo los cambios esperados de temperatura y precipitación inciden en
la frecuencia e intensidad de eventos de olas de calor y frío así como en la ocurrencia
de episodios de sequía meteorológica. Finalmente, se ha analizado dicho impacto en un
sector fuertemente ligado a las condiciones climáticas medias de una región y
condicionado a las condiciones meteorológicas que se suceden año tras año, como es
el sector vitícola, en todo territorio español ibérico-balear.

5.1. Consideraciones generales sobre la generación de escenarios de

clima futuro

Se obtienen por primera vez escenarios de clima futuro a escala local basados en ESMs
del CMIP5 para todo el siglo XXI de episodios de olas de calor y frío así como de
indicadores de sequía (SPI/SPEI) para Aragón. Estos escenarios ofrecen uno de los
mejores “Snapshot” de escenarios de cambio climático sobre el riesgo que las altas y
bajas temperaturas así como alteraciones del régimen pluviométrico pueden causar en
la región de Aragón, tanto espacial como temporalmente.

Los escenarios de variables extremas se han generado a partir de las proyecciones

futuras de temperatura y precipitación generadas en base a los ESMs anteriormente
citados. Para su generación se ha aplicado una técnica de downscaling estadístico en
dos pasos (selección de análogos y funciones de transferencia) desarrollada por la FIC
y aplicada en múltiples estudios nacionales e internacionales con excelentes resultados
(Ribalaygua et al., 2013a, 2013b, 2018; Monjo et al., 2016, 2018; Gaitán et al., 2019,
2020; Gutiérrez et al, 2019, Velasco et al., 2020).

Los modelos empleados en el estudio han sido ESMs, considerados como los modelos
más potentes que existen en la actualidad. Una de las principales ventajas de los ESMs
frente a los MCs es que los primeros presentan una mayor precisión en la simulación de
variables climáticas como demostraron los resultados de validación obtenidos en el
presente estudio. Los óptimos resultados obtenidos en el proceso de validación de los
escenarios de temperatura máxima y mínima hacen posible su uso en futuros estudios
relacionados con eventos extremos e indicadores bioclimáticos.

Otro factor clave que justifica el uso de esta generación de escenarios, es la designación
por parte del IPCC de los escenarios climáticos más actuales en el momento de la
elaboración del presente estudio, los denominados RCPs. Siendo una prioridad trabajar
con la información más actual disponible en el momento de la realización de los cálculos.

Aragón se caracteriza por una topografía muy compleja que provoca grandes gradientes
climáticos. Las técnicas de reducción de escala fueron necesarias para captar estas
características orográficas, permitiendo identificar las áreas más vulnerables a los
cambios extremos. La reducción de escala estadística se recomienda especialmente
para zonas con topografía compleja (Kattenberg y Amer Meteorol SOC, 1996). Además,
Aragón es un territorio representativo de diferentes climas europeos, desde las zonas

195
bajas del centro de Aragón (la cuenca del Ebro) hasta las regiones montañosas del norte
(los Pirineos), lo que lo convierte en un buen indicador de los futuros cambios climáticos
europeos.

5.2. Consideraciones sobre los escenarios de clima futuro a escala local

en la región de Aragón: temperatura y olas de calor/frío

Los resultados obtenidos para la temperatura muestran una clara tendencia hacía climas
cada vez más cálidos en el área estudiada, especialmente durante los meses de verano.
Los mayores aumentos de temperatura en Aragón se producirán durante el verano a
finales de siglo, alcanzando valores de hasta 7 ºC para el escenario RCP8.5. Los
aumentos de las temperaturas mínimas muestran comportamientos similares a los de
las máximas, pero con incrementos menos acusados (3 ºC y 5,6 ºC para los escenarios
RCP4.5 y RCP8.5 respectivamente en verano a finales de siglo).

Los escenarios obtenidos de olas de calor muestran un aumento de su intensidad media

cercano a los 2 ºC (alcanzando temperaturas de hasta 38,8 ºC) y un aumento medio de
la intensidad máxima de 3,6 ºC (temperatura de hasta 41,5 ºC) respecto al periodo
histórico (1971-2000) para el escenario RCP8.5 a finales de siglo, mientras que la
duración de las olas de calor aumentará en 7 días (alcanzando duración media total de
12 días). La intensidad y duración futuras de los episodios de olas de frío se mantendrán
estables.

Los resultados obtenidos para la temperatura van en línea con aquellos obtenidos en
otros estudios de características similares pero basados en otras técnicas de
downscaling. Los escenarios futuros de temperatura publicados por el IPCC (IPCC,
2013), AEMET, EURO-CORDEX (https://www.euro-cordex.net/) o Copernicus (www.
https://cds.climate.copernicus.eu/) en base al CMIP5 coinciden con los nuestros en que
los meses estivales serán los que sufran los mayores incrementos y los meses
invernales los que se verán afectados de forma más suave. Nuestros resultados
(aumentos de entre 3.6 y 7 ºC a finales de siglo) muestran un calentamiento menos
acusado que los publicados por AEMET (entre 4.1 y 8.2 ºC) (www.aemet.es). Respecto
a los resultados publicados por el IPCC o Copernicus, éstos provienen de la salida
directa de los ESMs, es decir, no tienen el valor añadido de los escenarios
regionalizados, condición que sí cumplen los escenarios publicados por AEMET. Uno
de los puntos a tener en cuenta en estudios de impacto climático, es el uso de un amplio
abanico de proyecciones climáticas (usando el mayor número de modelos climáticos y
RCPs así como de metodologías de regionalización). En base a esta hipótesis, surgieron
los proyectos EsTcena (Brands et al., 2013, Brands et al., 2011a) o EURO-CORDEX. El
primero de ellos, a pesar de basarse en downscaling estadístico, se realizó en base a
modelos climáticos del CMIP3 y no en ESMs, mientras que el segundo está enfocado
en downscaling dinámico. Todos estos resultados (incluidos los nuestros) son
complementarios entre sí y contribuyen a mejorar el conjunto de proyecciones climáticas
disponibles para Aragón.

Estos escenarios se compararon con aquellos publicados por Ribalaygua et al., 2013b,
en base al IPCC4, usando la misma metodología, datos observados y reanálisis pero
distintos grupos de modelos climáticos. Por un lado, se garantiza el buen
comportamiento de la metodología y por otro, los avances proporcionados por los
modelos CMPI5 frente a los CMIP3. Los resultados muestran cómo los nuevos

196
resultados tienden a incrementos mayores de temperatura. En ambos estudios, el patrón
espacial se mantiene (mayores incrementos en las regiones noroeste y suroeste).
Nuevamente, los resultados de validación ponen de manifiesto que los nuevos
escenarios son más precisos que los previos.

En base a los buenos resultados obtenidos en la simulación de temperaturas máximas

y mínimas, es posible partir de ellos en el análisis de la evolución futura de los episodios
de olas de calor y frío. El cálculo de los episodios de olas de calor y frio se realiza en
base a los percentiles 95 de temperatura máxima y 5 de temperatura mínima,
respectivamente. Los buenos resultados obtenidos en el proceso de verificación de
ambos percentiles permiten asumir que el umbral en el que se basa el cálculo de los
episodios de olas se está simulando de forma correcta.

Todos los modelos utilizados en el estudio coinciden en que el número de días de olas
de calor aumentará en las próximas décadas y por tanto, el número de episodios de olas
de calor. Existen casi tantas maneras de evaluar los episodios extremos relacionados
con la temperatura (especialmente los relacionados con temperaturas elevadas) como
estudios publicados. La mayoría de los estudios se centran en el análisis de eventos
pasados a partir de umbrales de temperatura máxima (Royé et al., 2020), noches cálidas
(Royé et al., 2021) o indicadores térmicos (D'Ippoliti et al., 2010). Una línea similar
siguen los pocos estudios que evalúan los eventos extremos de calor en el futuro. Por
ejemplo, basados en indicadores térmicos como el Heat Wave Magnitude Index (Russo
et al., 2014; Molina et al., 2020) o el Excess Heat Factor, EHF (Lorenzo et al., 2021) o
umbrales de temperatura (Lorenzo and Álvarez, 2022). Nuestros resultados (aumento
esperados de 7 días en la duración de una ola de calor) van en línea con los obtenidos
por muchos de ellos, como los publicados por AEMET (un aumento de entre 5 y 20 días
en la duración de las olas de calor y entre un 20 y un 40% en el número de días de calor)
o por Lorenzo et al., (2021) con un aumento del 104% de la intensidad de dichos
episodios. Los estudios mencionados se basan en modelos del CMIP5 pero o bien
interpolan directamente la salida de los ESMs o bien usan modelos dinámicos y en la
mayoría de ellos no se contempla nuestra área de estudio. Además, las definiciones
empleadas no son exactamente las mismas que las que se han utilizado en este estudio
y por tanto no podemos comparar los valores absolutos directamente pero si las
tendencias encontradas. Y a pesar de las diferentes maneras de evaluar el problema
todos ellos concluyen en que la tendencia esperada es hacía una mayor intensidad,
duración, frecuencia y extensión espacial de dichos episodios.

Lo contrario ocurre con los episodios de olas de frio, los cuales se espera que se
mantengan similares en frecuencia e intensidad. Son muy pocos los estudios que hacen
hincapié en su determinación y análisis, pero los que hay llegan a las mismas
conclusiones cualitativas (Carmona et al., 2016; Linares et al., 2015).

Estos resultados se pueden explicar fijándonos en las conclusiones teóricas mostradas

por el IPCC (IPCC, 2013) sobre los cambios en la probabilidad de alcanzar ciertos
valores de temperatura. La distribución de probabilidad de las temperaturas puede, o
bien moverse hacía climas más cálidos (variaciones en la media), puede extenderse
(incrementar su varianza) sin variar los valores promedio o puede ser una combinación
de ambos. Esta última distribución es la que más se ajusta a los datos observados en
las últimas décadas (Hansen et al., 2012, IPCC 2001). Por lo tanto, la nueva distribución

197
de temperaturas implicaría un cambio menos pronunciado en las temperaturas más frías
de la serie y un cambio mucho mayor en las más altas.

Los datos obtenidos para Aragón, muestran que los impactos por altas temperatura y
olas de calor más importantes estarán localizados en el Valle del Ebro. Resultado de
gran importancia ya que es la zona más poblada y de mayor concentración
socioeconómica de la región lo que puede suponer un gran riesgo para la salud, la
mortalidad, la movilidad y el bienestar socioeconómico entre otros factores. Aunque el
Valle del Ebro se espera que sufra las temperaturas más altas y las olas de calor más
intensas, será la zona de los Pirineos la que experimente los mayores incrementos de
temperatura y el mayor aumento en intensidad de las olas de calor respecto a los valores
actuales. Esta zona también será donde se dé la mayor intensidad de olas de frio,
aunque como se ha comentado con anterioridad no sufrirán alteraciones con respeto a
los datos actuales. Estos resultados son coincidentes con Lorenzo et al. (2021) que han
calculado el EHF a partir de modelos dinámicos pertenecientes a EURO-CORDEX y
coinciden en que la zona de los Pirineos será una de las regiones donde más se sentirán
los incrementos en los episodios de olas de calor/frío.

La zona de estudio es un área rica desde el punto de vista ecológico, con una alta
diversidad de flora y fauna y en la que se encuentran ubicados varios parajes protegidos,
como el Parque Nacional de Ordesa y Monte Perdido y los Parques Naturales de los
Valles Occidentales y de Posets, de gran riqueza ecológica. Esto la convierte en una
zona particularmente vulnerable a los efectos del cambio climático ya que su hábitat
puede verse alterado drásticamente a causa de las consecuencias que podría suponer
para la biodiversidad aragonesa (OPCC2, 2018). Entre algunas de las complicaciones
derivadas del cambio climático destacan: alteraciones en los patrones de migración de
ciertas especies (Sanz et al., 2003; Gordo and Sanz, 2006), pérdida de la cubierta
vegetal (OPCC2, 2018) y aumento de los incendios forestales (Cardil et al.,2013; Resco
de Díos et al.,2021), alteraciones del suelo causadas por la erosión y la desertificación
(Magrama, 2016) o cambios en la estratificación en altura de la vegetación (Revuelto et
al., 2022).

Mención especial merece la situación en el Pirineo Aragonés, la cual es realmente

alarmante. Estudios recientes llevados a cabo por organismos como el Instituto
Pirenaico de Ecología y diversas universidades (López-Moreno et al., 2016; Vidaller et
al., 2021; OPCC2, 2018) son contundentes y no dejan lugar a duda de la desaparición
paulatina del hielo glacial en el Pirineo. La fecha en la cual se pueden extinguir
completamente los glaciares pirenaicos se ha ido acortando como consecuencia de las
altas temperaturas y periodos cada vez más cálidos. Un ejemplo claro es el año 2022,
año en el que ha desaparecido el 20% de la masa de hielo glaciar del Aneto. Esta
situación no solo tiene repercusión en la biodiversidad de montaña sino que impacta
directamente, a través del deshielo, en el caudal de los ríos de la vertiente izquierda del
Valle del Ebro con las consecuentes implicaciones en el aporte hídrico (López-Moreno
et al., 2011) o en los episodios de inundaciones (Lastrada et al., 2020), entre otros
fenómenos.

Por lo tanto, los cambios en las temperaturas y aumentos en los eventos extremos en
las magnitudes calculadas en este estudio, pueden suponer un enorme riesgo para esta
área.

198
5.3. Consideraciones sobre los escenarios de clima futuro a escala local
en la región de Aragón: precipitación y sequía meteorológica

Uno de los principales problemas a los que nos enfrentamos en el análisis de estudios
de impacto de sequías meteorológicas es la falta de datos observados y de estudios a
nivel local que incluyan escenarios futuros y que sirvan de guía para afrontar dicha
problemática. En general, los expertos consideran el SPI como uno de los pocos índices
aplicable en cualquier región del mundo para cualquier escala (Hayes et al 2011)
ofreciendo múltiples ventajas en comparación con otros indicadores usados
ampliamente (Dracup et al., 1980; Hayes et al., 2011). En contexto de cambio climático,
además, se recomienda el análisis del SPEI- que incluye la ETP- y cuya formulación es
similar a la del SPI, lo que permite la comparativa entre ambos. Ambos índices se han
aplicado con anterioridad en Aragón (Vicente-Serrano et al., 2010a; Beguería et al.,
2014) lo que refuerza su uso en la zona de estudio. Es por ello que en este estudio,
tanto el SPI como el SPEI se han calculado en base a las proyecciones climáticas futuras
de precipitación (obtenidas en el segundo bloque de resultados) y temperatura
(obtenidas en el primer bloque de resultados).

Los escenarios futuros de precipitación muestran un leve descenso (bajo el escenario

RCP8.5) a lo largo del siglo XXI y todas las estaciones del año, excepto en verano donde
apenas se esperan variaciones con respecto a los valores actuales. Bajo el RCP4.5
apenas se esperan variaciones en el régimen pluviométrico. Estos resultados van en
concordancia con los obtenidos en otros proyectos de características similares.
Además, los resultados de validación para precipitación muestran valores de ± 1mm/día
(alrededor del 10%), estos valores se encuentran dentro de la variabilidad natural de la
precipitación por lo que se pueden considerar buenos resultados de validación. Los
nuevos escenarios de precipitación, basados en el CMIP5, muestran mejores resultados
de validación que los obtenidos en base al CMIP3, especialmente en los meses de
verano que es un mes crítico para Aragón (un bias de ±10% para CMIP5 frente a un
bias de ±20% en CMIP3).

En el análisis de los episodios de sequía meteorológica calculados a partir del SPI/SPEI,

se aprecia que la serie temporal de SPI/SPEI en base a datos observados es bien
representada por ambos indicadores calculado en base a datos simulados. Los
periodos secos/húmedos son bien recogidos aunque tienden a dar episodios de más
intensidad que los observados, especialmente a escalas largas. Estos resultados están
en línea con aquellos presentados en otros estudios para la zona y las pequeñas
diferencias que se aprecian se deben a la diferencia de metodologías e indicadores
empleados así como el periodo histórico observado. Por ejemplo, los periodos húmedos
y secos identificados en este estudio coinciden en gran medida con los identificados por
Vicente-Serrano and López-Moreno, (2005) basado en SPI (años húmedos: 1976-1980,
años secos: 1986-1987,1989, 1994-1997), Spinoni et al., (2015) basado en un índice
combinado a 12 meses (años secos:1979-1980,1995-1998) y López et al., (2007)
basado en el régimen pluviométrico (años secos: 1970, 1985, 1993 y 1995).

Respecto a los escenarios futuros, en los resultados de SPI no se aprecian cambios en

el balance hídrico en ninguna zona de Aragón ni en ninguna de las escalas temporales
consideradas, a excepción del RCP8.5 y finales de siglo en la zona del valle del Ebro
donde se aprecia una leve tendencia hacía climas más secos. Aunque hasta donde

199
sabemos no existen estudios similares en nuestra área de estudio, estos resultados van
en línea con lo obtenido en los escenarios de precipitación como era de esperar al
tratarse de la única variable sobre la que se calcula el SPI. Al considerar otras variables
como la temperatura, los escenarios de SPEI muestran una clara tendencia hacía
periodos más secos y largos, especialmente en el valle del Ebro y en el suroeste de la
región. Algo que se esperaba en base a los escenarios de temperatura y precipitación
con los que se han generado. Que los escenarios sean más intensos a escalas de 12
meses que a 1-3 meses es consecuencia de la manera en que los indicadores están
formulados. Por lo tanto, en el contexto actual de calentamiento global es esencial tener
en cuenta el efecto de la temperatura en los estudios de sequía.

Los resultados muestran que el Valle del Ebro (zona más poblada, incluyendo Zaragoza)
será la más susceptible a futuros periodos de sequía extrema con mayor intensidad y
duración, especialmente a finales de siglo, con implicaciones en sectores como la salud
(Roldán et al., 2016; Carmona et al., 2016), manejo del agua, economía (Pérez and
Barreiro-Hurlé, 2009) y sociedad en general (Lee et al., 2017). Estos resultados van en
línea con las zonas donde se detectaron los valores más elevados de temperatura y los
mayores descensos de precipitación como mostraron los resultados del ECCE Project
(Ministerio de Medio Ambiente, 2005) entre otros (Cook et al., 2014; Zambrano-Birgiarini
et al., 2010) de manera que la zona del Ebro seguirá siendo una de las más secas y
calurosas (Jiménez-Donaire et al., 2020)

Los resultados obtenidos son la representación más precisa hasta la fecha de la

magnitud, duración e intensidad de los episodios de sequía meteorológica y su duración
en diferentes periodos de acumulación. El uso de diferentes índices de sequía y escalas
temporales y su representación gráfica es una novedad relevante en la literatura
científica y pueden ser una referencia para el estudio de impactos de sequía en múltiples
sectores, especialmente en zonas con climas y orografías complejas como Aragón, algo
extrapolable a otras regiones de Europa con características similares.

5.4. Consideraciones sobre los escenarios de indicadores bioclimáticos

de interés vitícola a escala local en España

Se han generado escenarios de clima futuro para cuatro índices bioclimáticos

individuales de impacto vitícola (HI, CI, DI y HyI), un índice combinado (CompI) y una
zonificación vitícola (MCC System).

En este caso, el estudio se ha ampliado a toda la península Ibérica con el objetivo de

poder comparar los resultados obtenidos para la región vitícola aragonesa con aquellos
obtenidos para otras regiones vitícolas españolas. Además, disponer de información
climática en todo el territorio español ibérico-balear supone una herramienta de apoyo
en la adaptación al cambio climático. Por un lado, permite comparar medidas llevadas
a cabo por regiones climáticas similares (tipos de portainjertos y variedades plantadas,
manejo del regadío, etc.) y por otro lado, determinar regiones climáticamente aptas para
la vid que actualmente no se usan para tal fin y que pueden ser una alternativa futura a
las regiones actuales.

Para poder afrontar el estudio a nivel peninsular y balear, se han tenido que generar
proyecciones futuras de temperatura y precipitación para todo el territorio. Estas
proyecciones se basan en 9 ESMs del CMIP5 y dos RCPs, siguiendo la misma

200
metodología empleada para la región de Aragón. Se completa el estudio simulando
escenarios futuros de olas de calor/frío y escenarios de sequía mediante los indicadores
SPI/SPEI.

Tal como se desprende de los escenarios obtenidos en este estudio (véase Anexo R1),
se espera que en las próximas décadas la temperatura media anual ascienda en todo
el territorio español entre 2 y 4 °C, siendo los meses estivales los que experimentarán
los ascensos más acusados. Para estos meses críticos, se esperan aumentos de hasta
7 ºC a finales de siglo bajo el escenario RCP8.5 en el caso de las temperaturas máximas
y 5 ºC para las temperaturas mínimas. Toda la mitad sur peninsular, la zona
Mediterránea y las Islas Baleares experimentaran temperaturas máximas estivales por
encima de los 32 ºC a mediados de siglo, llegando a superar los 38 ºC a finales del
mismo. Estos aumentos de temperatura suponen una intensificación en los episodios
de olas de calor en todo el territorio, especialmente en toda la costa Mediterránea dónde
se esperan episodios de calor con una duración superior a 20 días (bajo el RCP8.5 y a
finales de siglo). A nivel vitícola, el aumento en los episodios de olas de calor supondrá
un gran riesgo, no solo de estrés térmico por exceso de calor, sino también de estrés
abiótico.

Respecto a las temperaturas mínimas (véase Anexo R1), los mayores aumentos se
esperan en los meses estivales, siendo la Costa Mediterránea la zona que
experimentará las temperaturas mínimas más elevadas. Aunque se esperan cambios
en las temperaturas mínimas medias, no se esperan grandes cambios en los extremos,
por lo que no se aprecian alteraciones importantes de los episodios de olas de frío
futuros respecto a los actuales. Hay dos aspectos importantes relacionados con la
temperatura mínima: las heladas y la temperatura mínima durante la maduración. Estas
variables han de ser evaluadas con precaución durante las tareas de mantenimiento del
viñedo.

Las proyecciones de precipitación para el territorio peninsular y balear apenas muestran

ligeros cambios en la cantidad de precipitación acumulada anualmente, aunque con
eventos de precipitación más concentrados en el tiempo (véase Anexo R2). Estas leves
variaciones de precipitación ratifican los resultados arrojados por los escenarios de
sequía meteorológica basados en SPI (dependiente exclusivamente de la precipitación),
los cuáles no muestran una señal significativa de cambio respecto a valores actuales.

Una situación diferente se obtiene al evaluar la sequía meteorológica a partir de

escenarios basados en el SPEI (combinando el efecto de la precipitación y la ETP). Los
resultados obtenidos muestran que bajo el escenario RCP8.5 y a finales de siglo, la
situación se va agravando según se aumentan los periodos considerados. De esta
manera, en acumulado a 1 o 3 meses, la mayoría del territorio sufrirá de sequía
moderada a severa, mientras que en acumulado de 12 meses, casi todo el territorio se
encontrará en situación de sequía severa a muy extrema. A mediados de siglo, la mayor
parte del territorio se encontrará en situación de sequía moderada a severa.

Los resultados obtenidos para las proyecciones de temperatura y precipitación así como
los de los eventos extremos concuerdan con aquellos publicados de forma oficial por
AEMET (www.aemet.es) o el IPCC (estos últimos basados en salidas directas de los
MCs) así como los publicados en estudios similares que contemplan o bien, la

201
información directa de los ESMs (Cos et al., 2022) o bien, el uso aplicado de
proyecciones climáticas (Lorenzo et al., 2021; Monjo et al., 2016; Miro et al., 2021)

Las variaciones esperadas en valores medios y extremos de temperatura y precipitación

tendrán implicaciones en los resultados obtenidos para los indicadores bioclimáticos.

Los resultados obtenidos ponen de manifiesto cómo los indicadores térmicos (HI y CI)
tenderán a aumentar a lo largo del siglo XXI, mientras que la escasez de agua (DI) será
más acusada. Las tendencias encontradas no tienen la misma repercusión en todo el
territorio, de manera que en el sur de la península, con valores de HI que superan los
3500°C y de CI por encima de los 20°C y de DI por debajo de los -200 mm, la continuidad
del sector vitivinícola en su estado actual se ve seriamente amenazada, con una
disminución de los años climáticamente óptimos como muestran los valores de CompI.
Por el contrario, el norte peninsular y las zonas montañosas, a pesar de los aumentos
previstos, con HI inferiores a 2500°C, noches frescas (CI inferior a 15°C) y un aporte
hídrico suficiente (DI superior a 150 mm) mejoran considerablemente su aptitud
climática (CompI) aunque se mantiene el riesgo de enfermedad de mildiu debido al
aumento de la temperatura y la humedad.

Estos cambios puede favorecer de forma positiva a la zona del norte de la península
pero regiones como el sur y el suroeste peninsular como el Mediterráneo (que ya se
encuentran en el límite de idoneidad vitícola) pueden verse afectados de forma negativa.
Estas últimas tendrán que hacer frente a problemas de estrés térmico e hídrico que
pueden poner en peligro el cultivo de las vides y por ende la calidad del vino. En aquellas
regiones con DO puede ser un problema de gran envergadura ya que son,
precisamente, esos factores los que dotan a una región de dicha distinción. Por otro
lado, si los cambios climáticos son muy acusados, es posible que muchas regiones
deban plantearse cambios a nivel del viñedo (fechas de laboreo, regadío, sombreado,
etc.) (Felxas et al., 2010), cambios a nivel de finca (ubicación, variedad de uva,
orientación, etc.) viñedo (Hall and Jones 2009, Lobell et al 2006) o cambios en la gestión
del (Renée and Thach 2014).

Las principales consecuencias de cambios hacía valores de HI más altos van a estar
relacionadas con la variedad de uva a cultivar. Este indicador hace referencia a los
requerimientos de acumulación de calor necesarios para que cada tipo de variedad
alcance la maduración de forma óptima. Debido a que se espera que se alcancen los
valores de requerimiento de calor con anterioridad, la maduración se completará antes
de lo que ocurre actualmente. Esto puede llevar a episodios de estrés térmico en
muchas regiones. Este índice tiene una fuerte correlación con las etapas fenológicas en
las que las condiciones cálidas juegan un papel importante, lo que se traduce en un
adelantamiento de las etapas fenológicas y una alteración del ciclo actual de la vid.

Debido al aumento esperado en las temperaturas mínimas se espera que el CI alcance

valores que se sitúen fuera de los requerimientos de frescor nocturno necesarios en el
periodo de maduración para ciertas variedades de uva. No se espera bajo ningún RCP
que los valores de CI desciendan en las próximas décadas. Por otro lado, si se alcanzan
los requerimientos térmicos de maduración antes de lo que se hace actualmente, habría
que replantearse si el cálculo del CI debe seguir haciéndose en base al mes de
septiembre. En el caso de que se garantice que se alcanzan los requerimientos térmicos

202
necesarios para la maduración, un descenso de las temperaturas nocturnas podría ser
beneficioso para ciertas variedades de uva.

Los cambios esperados para el DI van a condicionar el aporte hídrico necesario y por
consiguiente las necesidades de riego. En toda la región, a excepción de la cornisa
cantábrica y de los Pirineos, e independientemente del RCP considerado, los valores de
DI van a descender hasta valores por debajo de los 50 mm (escenario justo en el límite
de aporte hídrico), lo que puede suponer la necesidad de tomar medidas de restricción,
especialmente en los meses de verano. Si el DI se sitúa por debajo de los -100mm, la
región se volverá excesivamente seca y necesitará de aporte hídrico extra para cubrir
los requerimientos de la vid. Por el contrario, en aquellas regiones donde se espera que
el DI alcance valores por encima de los 150 mm (Pirineos y El Golfo de Vizcaya) se
puede comprometer la calidad de la uva y el vino, especialmente en años muy húmedos.
En regiones como Galicia y Asturias, donde el DI se moverá en valores intermedios
(entre 50 y -100mm) pequeñas épocas de déficit hídrico pueden ser beneficiosas,
especialmente durante la maduración.

Bajo las nuevas condiciones climáticas no se espera que aumente el riesgo de presencia
del Mildiu, pero sí se aprecia que hay diferencias de riesgo entre regiones con diferentes
patrones de precipitación. De esta manera, la zona sur será la que menos riesgo
presente, aumentando de forma progresiva hacía el norte, donde el régimen
pluviométrico es más abundante, especialmente durante la etapa de crecimiento.

Los resultados obtenidos van en línea con los publicados en otros estudios aplicados
sobre algunas de las regiones vitícolas más importante del mundo, como en Portugal
(Fraga et al., 2012), Italia (Bonfante et al., 2017), Alemania (Neumann and Matzarakis,
2011), Francia (Duchene and Schneider, 2005) y España (Gomez-Gesteira et al., 2011).
Y aunque estos estudios no se basen en las mismas metodologías (no aplican
downscaling, usan otros indicadores bioclimáticos, son regiones con características
climáticas y orográficas diferentes, etc.) obtienen conclusiones similares respecto a
cómo el sector vitícola se va a ver afectado por el aumento de las temperaturas y la
escasez hídrica.

Las características climáticas definen la idoneidad de una región para cultivar uvas que
den vinos de calidad. Se definió el CompI para evaluar si se dan dichas condiciones.
Los resultados obtenidos para este índice pueden parecer contradictorios, ya que zonas
como Andalucía o Castilla- La Mancha con fuerte tradición vitícola de calidad, presentan
un porcentaje muy bajo de años óptimos para el cultivo de la vid (principalmente debido,
a la escasez hídrica). En este caso se puede solventar con estrategias que compensen
dicha “falta de idoneidad climática”.

Los impactos derivados del cambio climático van a ser importantes a nivel de
aclimatización bioquímica, fisiológica y molecular de la uva. Diversos estudios destacan
el potencial adaptativo de la vid a nivel molecular para hacer frente al estrés climático
(Zha et al., 2018), por ejemplo, mediante un ajuste fotosintético (Gallo et al., 2021;
Kizildeniz et al., 2021), un aumento de la capacidad de disipación de calor a través de
las estomas (Costa et al. 2012) o alteraciones metabólicas (Duchene et al., 2012;
Kovaleski and Londo, 2019). Estos descubrimientos abren una vía para identificar los
rasgos genéticos y fisiológicos que hacen que una variedad sea más o menos resistente
y seleccionar las variedades que serían adecuadas para sustituir a las más sensibles.

203
5.5. Consideraciones sobre posibles medidas de adaptación y mitigación

Los cambios observados ponen de manifiesto que existe la posibilidad de que ciertas
regiones tradicionales puedan modificarse pasando de ser climáticamente óptimas para
el cultivo de la vid a no óptimas y viceversa (Bock et al., 2011; Santos et al., 2012; Fraga
et al., 2013; Jones et al., 2006; Jones and Webb, 2010)

Es importante establecer acciones prácticas y relevantes basadas en un análisis que

incluya un amplio rango de territorios y diferentes tipos climáticos y vitícolas. Si la
industria vitícola, en cualquier región del mundo, quiere sobrevivir debe adaptarse a los
retos del cambio climático, es decir, evaluar si es viable en su estado actual o si es
necesario hacer cambios (Graça et al 2016). Graça et al 2016 sugieren establecer un
marco de trabajo preliminar que permita replicar esfuerzos de planificación, reducir el
riesgo y los costes a la hora de implementar medidas de adaptación, aumentar la
capacidad de la industria para adaptarse al cambio climático y crear un sistema de
estandarización que permita comparar modelos, escenarios y opciones tecnológicas
entre regiones. Además es necesario tener en cuenta todos los aspectos que intervienen
en la cadena de valor del vino: características biológicas y físicas que se ven afectadas
por el cambio climático (Maclaren et al., 2015), prácticas de manejo (Howden et al.,
2007; Greenwood et al 2016), factores socioeconómicos y culturales (Grothmann and
Patt, 2005; Adger et al. 2008) así como las transformaciones que la industria vitícola ha
experimentado en las últimas décadas.

En función del aspecto que se quiera modificar serán necesario aplicar medidas de
adaptación inmediatas (medidas a corto plazo) o medidas que perduren en el tiempo
(medidas a largo plazo). Además hay que evaluar a que nivel se quieren aplicar. Las
medidas a nivel bodega son más fáciles, baratas e inmediatas pero tienen un bajo nivel
de adaptación. A nivel viñedo son más caras y presentan mayor dificultad de
implementación pero tienen un nivel mayor de adaptación.

En función de la escala temporal en la que se quiere actuar hablaremos de medidas a

corto, medio o largo plazo (Santos et al., 2020; Nauellau et al., 2021).

Medidas a corto plazo

Las medidas a corto plazo están centradas en amenazas específicas para optimizar la
producción y repercuten en la calidad y el estilo del vino. Estas medidas suponen
cambios en las medidas enológicas, mejoras en la calidad del vino, disminución de los
efectos en la variabilidad interanual o implicaciones en el seguro agrario. Destacan:

1) Medidas pensadas para retrasar la fecha de maduración: retraso en la fecha de poda

que puede retrasarla entre 3 a 5 días, selección clonal que puede retrasarla entre 3 a 8
días, aumentar la altura del tronco, entre 3 a 8 días, la elección del portainjerto entre 3
a 6 días y la elección de la variedad de la uva entre 10 a 25 días.

2) Reducir el estrés debido a la falta de agua, es decir: riego (si el estrés es muy alto
alto) (Felxas et al., 2010), variedad de injerto (estrés medio), especies en la cubierta
vegetal (estrés débil), labores del suelo y acolchado (estrés débil) y sistema de
espalderas (estrés medio).

204
3) Para realizar durante la época de cosecha destacan la adaptación de las técnicas de
cosecha y la adaptación del lagar (por ejemplo, control de la temperatura durante el
posproceso para evitar alteraciones bioquímicas).

5) Para aplicar en el manejo del canopy hay varias estrategias. Ralizar la poda tardía
(Duchene and Schneider 2005) para retrasar el comienzo de la brotación entre 8 y
11días con respecto a métodos tradicionales es una de ellas, lo que supondría un
retraso en la floración y en el envero de entre 4 a 5 días. Otras estrategias serían:
aumentar la altura del tronco de la vid, recorte de brotes y/o retirada de hojas para
intentar disminuir el ratio área foliar y peso de la fruta (<0,75m2/kg fruta) (Van Leeuwen
& Darriet, 2016), o bien aplicar sombreado para reducir la temperatura del fruto y así
aumentar la concentración del ácido málico y la acidez en la cosecha (Greer et al., 2011).

6) En la parte relacionada con el suelo, lo principal es controlar el aporte de agua, el

vigor de la planta y evitar la erosión del mismo. Para ello se pueden cambiar las técnicas
de laboreo. Por ejemplo, buscar alternativas al control químico de las malas hierbas a
través de gestión de suelos poco profundos (Pelegrino et al., 2004) o a través de la
cubierta de césped (limitamos la ETP del suelo especialmente en verano y en épocas
secas).

En conjunto, cambios en las prácticas de manejo, prácticas enológicas e introducción

de nuevas tecnologías en toda la cadena de valor (Lobell et al 2006, Olesen et al 2011,
Orduna 2010, Renée and Thach 2014).

Medidas de corto a medio plazo

Este tipo de medidas son aquellas que se implantan para momentos concretos que
tienen repercusión en el viñedo. Por ejemplo el manejo del riesgo de heladas a través
de métodos activos. Sería el caso de los métodos directos que se aplican en el momento
justo de la helada como máquinas de viento, calentadores, aspersores sobre la vid y/o
pasivos, métodos indirectos que sirven para preparar y reducir el impacto en viñedo.

Medidas a medio y largo plazo

Estas medidas están más pensadas para asegurar la viabilidad económica de la

viticultura tradicional y la reducción de gases de efecto invernadero asociados a su
actividad y enfocadas, principalmente, a la gestión del viñedo. Estas medidas requieren
de cambios de mayor envergadura. Algunas de las medidas que se pueden llevar a cabo
son:

1) Cambios en las localizaciones (no solo zonas climáticas sino en latitud y altura) (Fraga
et al 2014, White et al 2006, Hall and Jones 2009, Lobell et al 2006, Hannah et al 2013,
Moriondo et al 2013). Por otro lado, la proximidad a masas de agua que reducen la
amplitud térmica diaria así como la cercanía a los bosques donde la humedad es mayor
y por tanto más favorables para la proliferación de algunas enfermedades (Hancock
2005).

2) Cambios en las orientaciones, favoreciendo (Intrieri et al., 1998; Grifoni et al 2008;

Hancock 2005) o disminuyendo (Greer et al., 2011) la incidencia de radiación en función
de las necesidades.

205
3) Cambios en las variedades, adaptándose a las nuevas condiciones térmicas y
pluviométricas (Jones et al., 2006; Duchene et al., 2012) e intentando mantener las ya
existentes que garanticen la biodiversidad natural (Tello et al 2012; Karoglan et al 2018).
Realizar la elección de la variedad de uva en función del periodo de maduración
eligiendo maduración tardía e incluso sustitución de variedades de blanco a rojo como
consecuencia de la demanda de calor (Stock et al., 2004) son otras opciones, siempre
y cuando la legislación internacional y nacional lo permitan.

4) Selección clonal (Koundouras et al., 2008; Duchene et al., 2012). En la mayoría de

los casos hay una diferencia de entre 8 a 10 días en la fecha de maduración entre clones
de la misma variedad de uva y no supone una modificación en la tipicidad del vino.

5) Selección del portainjerto en función de las condiciones hídricas y físicas del suelo
(Ozden et al 2010; Pavlousek et al 2011; Harbertson and Keller 2012; Koundouras et al
2008) así como atendiendo a las interacciones entre ambas (Romero et al., 2006;
Pelegrino et al., 2004). Esta medida no solo no modifica la tipicidad del vino sino que es
respetuosa con el medioambiente, no aumenta los costes de producción y puede
suponer un retraso en la maduración de hasta 7-10 días.

6) Establecer un buen diseño del viñedo: densidad de plantación, orientación de las filas,
mallas de sombreo, gestión sostenible del agua, estrategias de venta etc. (Renée and
Thach 2014)

7) Prevención frente a pestes y enfermedades. Un sondeo realizado sobre las

principales adversidades a las que ha tenido que hacer frente el sector vitícola europeo
en las últimas décadas (Battaglini et al., 2008) destaca el aumento de las enfermedades
y pestes que han afectado a la vid. La intensificación de esta amenaza está ligada al
aumento de las temperaturas y de los patrones de humedad.

8) Agrodiversidad como herramienta para la pérdida del cultivo (Morales-Castilla 2020)

9) Introducir sistemas de riego. Esta opción supone un coste económico, social y

medioambiental.

La vid es un cultivo muy agradecido que puede crecer en suelos y condiciones

edafológicas por lo que gracias a una buena planificación es posible su desarrollo en
casi cualquier suelo (Bucelli & Constantini, 2006; Faila et al., 2008). Para poder
implementar cualquier medida de adaptación de forma satisfactoria es necesario
identificar para cada tarea de cada momento de la cadena vitícola que tipo de impacto
climático se produce a corto, medio y largo plazo y actuar en consecuencia (Graça et
al., 2016).

Las medidas de mitigación forman parte también de las estrategias del sector para
enfrentarse al Cambio Climático. Uno de los planteamientos que se están afrontando
desde el mundo vitícola es reducir la huella de carbono: considerando la vid como
sumidero de CO2 y reduciendo el balance de carbono en la viticultura (Jones & Webb
2010), así como analizando el potencial mitigador de algunas prácticas agrarias que
sirvan para controlar las emisiones, la erosión, la contaminación difusa, la pérdida de
nutrientes o el gasto energético (Iglesias y Moreno 2009).

206
5.6. Aspectos metodológicos a considerar

Consideraciones sobre la metodología de downscaling estadístico

Toda metodología de downscaling estadístico presenta ciertas limitaciones que han de

tenerse en cuenta a la hora de interpretar los resultados obtenidos (es decir, las
proyecciones climáticas):

1) En general, las técnicas de downscaling tienden a suavizar las series simuladas

respecto a las observadas, es decir, tienen dificultades recogiendo los extremos. Por
ejemplo, en el caso de las olas de calor/frío, al ser más larga la serie del reanálisis que
la observada, el número de días considerado para calcular el P95 y el P05 es mayor,
por lo que la intensidad y duración de dichos episodios es simulada de forma más suave.

Una simulación más suave puede llevar a que en el análisis de eventos extremos
categorizados, estos no se sitúen en las mismas categorías considerando evento
simulado y evento observado.

2) Un modelo climático no reproduce el clima día a día por lo que la forma de validarlos
ha de hacerse en base a estadísticos climáticos sobre periodos de tiempo largos. Estos
estudios, en ocasiones, suponen una pérdida de información sobre la variabilidad
climática. Además, para evaluar su comportamiento no se puede comparar el periodo
Historical con las observaciones (estás presentan lagunas en sus series) y ha de
hacerse frente a un reanálisis.

Estas dos limitaciones se contemplan en nuestro estudio a la hora de corregir el error

sistemático a las series simuladas “brutas” y por tanto suavizar el efecto de ambas.

3) Trabajar con 9 ESMs y 2 RCPs, permite disponer de un conjunto de proyecciones

climáticas con las que evaluar las incertidumbres asociadas al proceso de generación
de proyecciones climáticas, lo que favorece la toma de decisiones dentro de un rango
de actuación.

Consideraciones sobre la simulación de indicadores bioclimáticos

La metodología de downscaling empleada ha permitido tener en cuenta dos aspectos

importantes en estudios de impacto climático en el viñedo:

1) Hay estudios que refuerzan la idea de que los cambios diarios en las condiciones
atmosféricas juegan un papel importante en la fenología de las plantas (Jones y
Davis, 2000), y estos cambios pueden ser más o menos significativos
dependiendo de la región donde se produzcan. Estos aspectos se consideran
intrínsecamente en el tipo de metodología de downscaling utilizada en este
estudio basada en una estratificación analógica de campos atmosféricos.

2) Un factor a considerar son los beneficios que un aumento de CO2 en condiciones

climáticas futuras tiene en el desarrollo de la vid (Bindi et al., 2001; Goncalves et
al., 2009; Moutinho-Pereira et al., 2009). Aunque en este estudio no se analiza
directamente este punto, se tiene en cuenta a la hora de considerar los diferentes
RCPs (contemplando distintos forzamientos radiativos).

207
Así mismo, en la evaluación de indicadores bioclimáticos se presentan ciertas
limitaciones que se han solventado gracias a la manera en la que se ha abordado el
estudio:

3) Las DOs se caracterizan, precisamente, por ser las condiciones locales las que
proporcionan los matices que hacen que sus vinos sean únicos y reconocibles.
Al trabajar a escala local se están teniendo en cuenta las características micro
climáticas que proporcionan dicha especificidad.

4) Evaluar los valores absolutos, tal como se ha hecho en este estudio, permite
conocer las particularidades de cada indicador climático a escala de observatorio
y por tanto, no perder el valor añadido que proporciona el estudio al trabajar a
escala local. Los indicadores climáticos tienen una categorización muy amplia,
es decir, el rango de valores que engloba cada categoría es muy grande, y
considerar solo dichas categorías supone una pérdida de información de gran
valor para la adaptación al cambio climático.

Finalmente, hay ciertas limitaciones que no se han podido considerar dentro del estudio
y que son un punto importante a tener en cuenta en estudios futuros:

5) Posibles alteraciones en las concentraciones de ozono y precursores del mismo

(NOx, CO, CH4, NMCH), como consecuencia del aumento de las temperaturas
en un marco de cambio climático, pueden tener un efecto indirecto en la vid
(Isaksen and Wang, 2002). Algunos estudios señalan al ozono como el causante
de la pérdida de productividad y el descenso del contenido de azúcar en la vid
(Ascenso et al., 2021).

6) Los resultados de este estudio han de completarse con otros factores que
complementan a los climáticos como son longitud, latitud, altitud, topografía,
proximidad a masas de agua, orientación y exposición, propiedades del suelo,
prácticas de manejo e interacción entre todos los actores. Estos factores en
conjunto son lo que actúan además del clima para dar lugar a esos vinos únicos
que caracterizan cada DO. Dentro de una misma DO pueden diferenciarse
diferentes tipos de uvas, suelos y características de campo que juegan un papel
importante en la diferenciación vitícola. Aunque estos factores son cruciales no
suponen un reto tan importante como el que nos plantea el cambio climático.

208
6. CONCLUSIONES

209
CONCLUSIONES

1. Es la primera vez que se generan escenarios a escala local de temperatura y

precipitación para Aragón en base a las salidas proporcionadas por modelos
CMIP5 y usando para ello una metodología de downscaling estadístico.

2. Los escenarios de temperatura son más robustos que los que obtuvimos en
estudios previos, y publicados en Ribalaygua et al 2013b, en base a fases
previas del CMIP5 por lo que resultan más adecuados y fiables para su uso en
estudios de adaptación al cambio climático.

3. También los escenarios de precipitación resultado de este estudio proporcionan

mayor consenso entre modelos que los obtenidos en estudios previos, donde la
dispersión entre las proyecciones climáticas era mayor y por tanto presentaban
una mayor incertidumbre.

4. Nuestros resultados en relación a las temperaturas esperadas en Aragón para

el Siglo XXI, confirman el consenso científico global de un aumento gradual de
temperaturas a nivel planetario:
 los mayores incrementos de temperatura se esperan en Aragón para los
meses estivales: hasta 7 ºC y 5.6 ºC en el peor de los escenarios,
(RCP8.5) para temperatura máxima y mínima, respectivamente.
 y los más suaves en invierno: hasta 4-4.2 ºC también de acuerdo con el
escenario RCP8.5.
 las variaciones esperadas en base al CMIP5 se esperan más acusadas
que las obtenidas en base a escenarios anteriores, siendo incluso 2 ºC
más intensas en los meses estivales bajo el escenario más desfavorable.
 A nivel del territorio, las regiones noroeste y suroeste de la región son las
que se verán más afectadas por estas variaciones.

5. No se esperan grandes cambios en el régimen pluviométrico de Aragón. En

base a los resultados obtenidos se esperan variaciones medias de alrededor de
un ±10% (el equivalente a 1mm/día) en todas las estaciones del año a excepción
del verano, donde no se esperan cambios, de acuerdo al escenario más
desfavorable.

6. La ampliación del estudio a todo el territorio ibérico-balear se ha realizado con el

fin de poder evaluar el impacto del cambio climático sobre la vid para toda
España. Entre los resultados obtenidos de las proyecciones climáticas de
temperatura y precipitación destacan:
 Las variaciones de temperatura media en España se esperan graduales
a lo largo de todo el Siglo XXI, llegándose a alcanzar incrementos de
hasta 4 ºC en el escenario más desfavorable. Respecto a las
temperaturas máximas, los mayores incrementos se esperan en los
meses estivales (hasta 7 ºC en el caso más desfavorable a finales de
siglo) lo que supondrían temperaturas de hasta 38 ºC en gran parte de la

210
 meseta peninsular. Las temperaturas mínimas experimentarán ascensos
menos acusados (alrededor de 5 ºC a finales de siglo), siendo,
nuevamente, los meses estivales los más afectados.
 No se aprecian variaciones de precipitación destacables respecto a
valores actuales. La zona noroeste seguirá siendo la región de mayor
pluviometría.
 A nivel de territorio, la meseta sur peninsular, especialmente Andalucía,
Extremadura y Murcia, junto con el Valle del Ebro serán las regiones
dónde se alcancen lpos mayores incrementos de temperatura.

7. Es la primera vez que se generan escenarios a escala local de episodios de

olas de calor/frío para la zona de Aragón en base a proyecciones climáticas
locales de temperatura generadas con una metodología de dowsncaling
estadístico y en base a la información proporcionada por el CMIP5.

8. Los escenarios de olas de calor muestran episodios cada vez más largos e
intensos a lo largo de las próximas décadas en Aragón:
 En promedio para todo el territorio, se espera un aumento en la
intensidad media cercano a los 2 ºC y a los 3.6 ºC en el caso de la
intensidad máxima (alcanzándose temperaturas medias de casi 39 ºC y
máximas superiores a 41.5 ºC) a finales de siglo bajo el escenario
RCP8.5.
 Respecto a la duración de la ola de calor hemos estimado que aumentará
en 7 días, lo que supone una duración media de las olas de calor de 12
días.
 Los episodios más intensos se esperan en la zona del Valle del Ebro, con
episodios de olas de calor que pueden alcanzar valores máximos de 46
ºC, con una intensidad media de 41 ºC y una duración media de 11 días.
Esto provocará los impactos socioeconómicos y sobre la salud más
importantes al concentrarse en esta zona la mayor parte de la población.
 Por otra parte, los cambios extremos en relación a la duración de las olas
de calor, con respecto al clima actual, se observarán en los Pirineos, con
un aumento en la duración de las olas de calor de 10 días, haciendo que
su duración se extienda hasta 17 días y con temperaturas medias
superiores a los 30 ºC. Esto sin duda impactará en la biodiversidad de
esta zona sensible, así como en el régimen hídrico de toda la comunidad.

9. Los escenarios de olas de frío se esperan que se mantengan en intensidad y

duración similares a los actuales en todo el territorio. Estos resultados van en
línea con la teoría, ampliamente contrastada, de que las variaciones en la
distribución de la temperatura afectan tanto a la media como a la varianza, por
lo que las variaciones en las temperaturas mínimas extremas serán muy poco
acusados.

10. Es la primera vez que se generan escenarios a escala local de episodios de

sequía calculados a partir de los indicadores SPI/SPEI para la zona de Aragón
en base a proyecciones climáticas locales de temperatura generadas con una

211
 metodología de dowsncaling estadístico y en base a la información
proporcionada por el CMIP5.

11. Estos escenarios de SPI/SPEI suponen una manera novedosa de representar la

magnitud, duración y la intensidad de los episodios de sequía meteorológica
en base a distintas escalas temporales y su representación gráfica.

12. Nuestros resultados muestran una clara tendencia hacía episodios de sequía
más intensos, especialmente a finales de siglo y cuando se consideran periodos
acumulativos más largos. Esta tendencia se aprecia sólo en escenarios de SPEI
(sequía severa a extrema) mientras que en los escenarios SPI esta tendencia es
muy suave (sequía normal).

13. A nivel de impactos en territorio, el área más afectada por episodios de sequía y
descenso de precipitación es la zona del Valle de Ebro. Estos resultados no solo
coinciden con estudios previos realizados sobre la zona sino que además van
en línea con los resultados históricos, que ponen de manifiesto cómo la zona del
Valle del Ebro se ha visto afectada por episodios de sequía cada vez más
recurrentes e intensos en las últimas décadas.

14. Los escenarios futuros de eventos extremos (olas de calor/frío y sequía) en el

territorio español ibérico-peninsular muestran las siguientes conclusiones
generales:
 Intensificación de los episodios de olas de calor en todo el territorio,
especialmente en toda la costa Mediterránea, con episodios de más de
20 días de duración (bajo el RCP8.5 a finales de siglo).
 Aunque se esperan cambios en las temperaturas mínimas medias, no se
esperan grandes cambios en los extremos, por lo que no se aprecian
alteraciones importantes de los episodios de olas de frío futuros respecto
a los actuales.
 Al no apreciarse cambios significativos en el régimen pluviométrico, los
escenarios de SPI no muestran alteraciones significativas con respecto
al periodo actual, por lo que según estos escenarios toda la región se
encontraría bajo un estado de sequía considerado normal (debido a la
propia variabilidad climática).
 Los escenarios de SPEI, bajo el escenario RCP8.5 y a finales de siglo,
ponen de manifiesto como toda la región se mueve hacía episodios de
sequía cada vez más extrema y extensa en el tiempo.

15. Se obtienen por primera vez resultados rigurosos y a escala local de los
impactos que el cambio climático supone para el sector vitícola español
(peninsular y balear) en base a proyecciones climáticas regionalizadas del
CMIP5.

16. Por primera vez, se analizan los indicadores climáticos en base a valores
absolutos junto con valores categorizados. Esta manera de evaluar los
resultados permite tomar decisiones mucho más precisas considerando el valor
añadido que supone tener en cuenta las condiciones micro climáticas locales.

212
17. Los resultados obtenidos a través de la simulación de diferentes indicadores
bioclimáticos reflejan que el territorio ibérico-balear será uno de los más
afectados a nivel, tanto térmico como hídrico. Estos datos concuerdan con los
que ponen de manifiesto estudios similares pero aplicados en áreas más
extensas, como Europa o el Mediterráneo, sin tener en cuenta las condiciones
locales muestran tendencias similares.

18. Se aprecia un aumento en los valores termales (como indican el HI y el CI),

mientras que se espera un descenso las reservas de almacenamiento hídrico
(como pone de manifiesto el DI). La combinación de ambas situaciones
compromete el cultivo de la vid que verá alteradas sus condiciones hidrotermales
actuales.

19. Estos cambios supondrán alteraciones en las características organolépticas de

la uva y posiciona a gran parte del territorio (especialmente la meseta central y
el sur peninsular) en los límites térmicos óptimos para el cultivo de la vid.

20. A nivel del territorio el impacto es heterogéneo:

 En el sur de la península, con valores de HI superiores a 3500°C y CI por
encima de 20°C y DI inferior a -200 mm, la continuidad del sector vitícola
en su estado actual se verá seriamente amenazada, con una disminución
de los años climáticamente óptimos como muestran los valores de
CompI.
 El norte peninsular y las zonas montañosas, a pesar de los aumentos
previstos, con HI inferiores a 2500°C, noches frescas (IC inferior a 15°C)
y suministro de agua suficiente (DI superior a 150 mm) mejorarán
considerablemente su aptitud climática (CompI), aunque el riesgo de
enfermedad de mildiu se mantiene debido al aumento de la temperatura
y la humedad.

21. Para poder mantener los niveles de calidad de las vides y los vinos, lo más
probable es que los productores vitícolas se vean obligados a adoptar medidas
de adaptación al cambio climático y replantearse la manera actual en la que
funcionan sus viñedos y buscar como potenciar los efectos positivos y minimizar
los negativos asociados al cambio climático.

22. Estos resultados proporcionan información de gran valor añadido para la toma
de decisiones, nos solo en el sector vitícola, sino también en otros sectores
sensibles a los eventos extremos (olas de calor/frío y sequía meteorológica).

23. Así mismo, estos resultados, no solo sirven para el territorio objeto de estudio,
sino que debido a la gran variedad climática y orográfica tanto de Aragón como
de toda España, pueden ser de utilidad para áreas similares como la zona del
Mediterráneo (Italia, Grecia), la zona atlántica (Portugal) e incluso las regiones
vitícolas de Napa en California o las grandes regiones vitícolas de Chile y
Argentina (ubicadas en regiones climáticas de latitudes medias).

213
BIBLIOGRAFÍA

Abaurrea J., Asín J. and Cebrián A.C., Modelling the occurrence of heat waves in
maximum and minimum temperatures over Spain and projections for the period
2031-60, Global and Planetary Change, Volume 161, 2018, Pages 244-260,
ISSN 0921-8181, https://doi.org/10.1016/j.gloplacha.2017.11.015.
Adger WN, Dessai S, Goulden M et al. Are there social limits to adaptation to climate
change? Climatic Change, 2008 93, 335–354.
Alary V, Messad S, Aboul-Naga A, Osman MA, Daoud I, Bonnet P, et al. Livelihood
strategies and the role of livestock in the processes of adaptation to drought in
the Coastal Zone of Western Desert (Egypt). Agricultural Systems 2014; 128: 44-
54.
Aleixandre JL, Giner JF, Aleixandre–Tudó JL. Evaluación del efecto terroir sobre la
calidad de la uva y el vino. Enoviticultura N20. 2013.
Alig RJ, Adams DM, McCarl BA. Projecting impacts of global climate change on the US
forest and agriculture sectors and carbon budgets. Forest Ecology and
Management 2002; 169: 3-14.
Alonso AD, O'Neill MA. Climate change from the perspective of Spanish wine growers:
a three-region study. British Food Journal 2011; 113: 205-221.
Anderson BG, Bell ML. Weather-Related Mortality How Heat, Cold, and Heat Waves
Affect Mortality in the United States. Epidemiology 2009; 20: 205-213.
Andrés, S., Pascual, N, Diago, M.P., Miranda, J., Ramalle-Gomara, E., Bengoechea, C.,
Arpón, L. and Zúñiga Crespo, J. Impacto, adaptación y percepción del cambio
climático en la DOCa Rioja, 2020.
Ascenso A, Gama C, Blanco-Ward D, Monteiro A, Silveira C, Viceto C, et al. Assessing
Douro Vineyards Exposure to Tropospheric Ozone. Atmosphere 2021; 12.
Ashenfelter, O., Ashmore, D., Lalonde, R., Bordeaux wine vintage quality and the
weather. In: Chance, 1995 Vol. 8, No. 4, 7 pp.
Asong ZE, Khaliq MN, Wheater HS. Projected changes in precipitation and temperature
over the Canadian Prairie Provinces using the Generalized Linear Model
statistical downscaling approach. Journal of Hydrology 2016; 539: 429-446.
Barbeau G.,. Climat et vigne en moyenne vallée de la Loire, France. In : Proceedings of
the Conference on Climate and Viticulture, Zaragoza (Spain),2007, pp. 106- 111.
Barrera-Escoda A, Goncalves M, Guerreiro D, Cunillera J, Baldasano JM. Projections of
temperature and precipitation extremes in the North Western Mediterranean
Basin by dynamical downscaling of climate scenarios at high resolution (1971-
2050). Climatic Change 2014; 122: 567-582.
Bates B, Kundzewicz Z, Wu S, Palutikof J. Climate Change and water. 2008. Eds. IPCC
Secretariat, Geneva, 210 pp
Battaglini A, Barbeau G, Bindi M, Badeck F-W. European winegrowers' perceptions of
climate change impact and options for adaptation. Regional Environmental
Change 2009; 9: 61-73.
Begueria S, Vicente-Serrano SM, Reig F, Latorre B. Standardized precipitation
evapotranspiration index (SPEI) revisited: parameter fitting, evapotranspiration
models, tools, datasets and drought monitoring. International Journal of
Climatology 2014; 34: 3001-3023.
Bellia S., Douguedroit A. and Seguin B.,. Impacts du réchauffement sur les étapes
phénologiques du Grenache et de la Syrah dans les Côtes du Rhône
méridionales et les Côtes de Provence (1976-2000). 2008 In : Journées de
Climatologie, Nantes (France), Climat et société : climat et végétation, pp. 45-55.

214
Bender MA, Knutson TR, Tuleya RE, Sirutis JJ, Vecchi GA, Garner ST, et al. Modeled
Impact of Anthropogenic Warming on the Frequency of Intense Atlantic
Hurricanes. Science 2010; 327: 454-458.
Benestad RE, Hanssen-Bauer I, Forland EJ. An evaluation of statistical models for
downscaling precipitation and their ability to capture long-term trends.
International Journal of Climatology 2007; 27.
Beniston M, Stephenson DB, Christensen OB, Ferro CAT, Frei C, Goyette S, et al. Future
extreme events in European climate: an exploration of regional climate model
projections. Climatic Change 2007; 81: 71-95.
Bentsen M, Bethke I, Debernard JB, Iversen T, Kirkevag A, Seland O, et al. The
Norwegian Earth System Model, NorESM1-M - Part 1: Description and basic
evaluation of the physical climate. Geoscientific Model Development 2013; 6:
687-720.
Berg A, Sheffield J. Climate Change and Drought: the Soil Moisture Perspective. Current
Climate Change Reports 2018; 4: 180-191.
Berg A, Sheffield J, Milly PCD. Divergent surface and total soil moisture projections
under global warming. Geophysical Research Letters 2017; 44: 236-244.
Bernetti, I., Menghini, S., Marinelli, N., Sacchelli, S., Alampi Sottini, V., 2012.
Assessment of climate change impact on viticulture: economic evaluations and
adaptation strategies analysis for the Tuscan wine sector. Wine Economics and
Policy 2012,1, 73-86.
Bindi M, Fibbi L, Gozzini B, Orlandini S, Miglietta F. Modelling the impact of future climate
scenarios on yield and yield variability of grapevine. Climate Research 1996; 7:
213-224.
Bindi M, Fibbi L, Lanini M, Miglietta F. Free Air CO2 Enrichment (FACE) of grapevine
(Vitis vinifera L.): I. Development and testing of the system for CO2 enrichment.
European Journal of Agronomy 2001; 14: 135-143.
Blanco-Ward D, Garcia Queijeiro JM, Jones GV. Spatial climate variability and viticulture
in the Mino River Valley of Spain. Vitis 2007; 46: 63-70.
Blenkinsop S, Fowler HJ. Changes in European drought characteristics projected by the
PRUDENCE regional climate models. International Journal of Climatology 2007;
27: 1595-1610.
Bock A, Sparks T, Estrella N, Menzel A. Changes in the phenology and composition of
wine from Franconia, Germany. Climate Research 2011; 50: 69-81.
Bois, B., M. Moriondo, and G. V. Jones,: Thermal risk assessment for viticulture using
monthly temperature data. In : Proceedings of the 10th International Terroir
Congress, 2014, Vol. 2. Corvinus University of Budapest
Bois B., Cartographie agroclimatique à méso-échelle : méthodologie et application à la
variabilité spatiale en Gironde viticole. Conséquences pour le développement de
la vigne et la maturation du raisin. Thèse de Doctorat, Université Bordeaux1
(France). 2007
Bonfante, A.; Alfieri, S.M.; Albrizio, R.; Basile, A.; De Mascellis, R.; Gambuti, A.; Giorio,
P.; Langella, G.; Manna, P.; Monaco, E.; et al. Evaluation of the effects of future
climate change on grape quality through a physically based model application: A
case study for the Aglianico grapevine in Campania region, Italy. Agric. Syst.
2017, 152, 100–109
Bonfante, A.; Monaco, E.; Langella, G.; Mercogliano, P.; Bucchignani, E.; Manna, P.;
Terribile, F. A dynamic viticultural zoning to explore the resilience of terroir
concept under climate change. Sci. Total Environ. 2018, 624, 294–308
Bonnefoy C., Quénol H., Bonnardot V., Barbeau G., Madelin M., Planchon O. and
Neethling E.,. Temporal and spatial analyses of temperature in a French wine-
producing area: the Loire Valley. Int. J. Climatol., 2013. 33, 1849-1862 (DOI :
10.1002/joc.3552).
Branas J. Viticulture générale. Dehan, Montpellier. 1974; 990 p
Branas J, Bernon G, Levadoux L. léments de viticulture générale 1946.

215
Brands S, Herrera S, Fernandez J, Gutierrez JM. How well do CMIP5 Earth System
Models simulate present climate conditions in Europe and Africa? Climate
Dynamics 2013; 41: 803-817.
Brands S, Herrera S, San-Martin D, Gutierrez JM. Validation of the ENSEMBLES global
climate models over southwestern Europe using probability density functions,
from a downscaling perspective. Climate Research 2011a; 48: 145-161.
Brands S, Taboada JJ, Cofino AS, Sauter T, Schneider C. Statistical downscaling of daily
temperatures in the NW Iberian Peninsula from global climate models: validation
and future scenarios. Climate Research 2011b; 48.
Breshears D. An ecologist’s perspective of ecohydrology, Bull. Ecol. Soc. Am., 2005; 86:
296 – 300.
Brixner, G.F., Schöffel, E.R. and Tonietto, J. Determinacao da evapotranspiracaao por
diferentes métodos e sua aplicacao no índice de seca na campnha Gaúcha,
Brasil. Rev. Bras.Fructi., Jaboticabal-SP, 2014, V.36, n4, p780-793
Bryant. EA. Natural hazards. International Journal of Climatology, E. Cambridge
University Press, Cambridge. ISBN 0 521 37295 X. 1993; 13:344-346.
https://doi.org/10.1002/joc.3370130310
Bryant. EA. Natural hazards. International Journal of Climatology, E. Cambridge
University Press, Cambridge. ISBN 0 521 37295 X. 1993; 13:344-346.
https://doi.org/10.1002/joc.3370130310
Bucelli, P., & Costantini, E. Vite da vino e zonazioni vitivinicole. In Metodi di valutazione
dei suoli e delle terre (p. 922). 2006. Siena: Edizioni Cantagalli.
Buerger CM, Kolditz O, Fowler HJ, Blenkinsop S. Future climate scenarios and rainfall-
runoff modelling in the Upper Gallego catchment (Spain). Environmental Pollution
2007; 148.
Bureau, S. M., A. J. Razungles, and R. L. Baumes. The aroma of Muscat of Frontignan
grapes: effect of the light environment of vine or bunch on volatiles and
glycoconjugates. J. Sci. Food Agric. 2000, 80:2012–2020
Burke EJ, Brown SJ. Evaluating uncertainties in the projection of future drought. Journal
of Hydrometeorology 2008; 9: 292-299.
Buttrose, M. S., C. R. Hale, and W. M. Kliewer. Effect of temperature on composition of
cabernet-sauvignon berries. Am. J. Enol. Vitic. 1971,22:71–75.
Calbo J. Possible Climate Change Scenarios with Specific Reference to Mediterranean
Regions. In: Sabater S, Barcelo D, editors. Water Scarcity in the Mediterranean:
Perspectives under Global Change. 8, 2010, pp. 1-13.
Camps, J. O., and M. C. Ramos. Grape harvest and yield responses to inter-annual
changes in temperature and precipitation in an area of north-east Spain with a
Mediterranean climate. Int. J. Biometeorol. 2012, 56:853–864.
Carbonneau, A. Ecophysiologie de la vigne et terroir. Terroir, zonazione, viticoltura.
Trattato internazionale, 2003. Phytoline 1:61–102.
Carbonneau A, Tonietto J. La géoviticulture: de la géographie viticole aux évolutions
climatiques et technologiques à l’échelle mondiale. Rev. Oenol. Tech.Vitivin.
Oenol.1998; 87, 16-18.
Cardil, A., Molina Terrén, D., Ramirez, J. and Vega-Garcia, C. Trends in adverse weather
patterns and large wildland fires in Aragón (NE Spain) from 1978 to 2010. Natural
Hazards and Earth System Sciences, 2013; 13. 1393-1399. 10.5194/nhess-13-
1393-2013.
Carmona, R., Díaz, J., Miron, I., Ortiz, C., Luna, M. Y., Linares, C. Mortality attributable
to extreme temperatures in Spain: A comparative analysis by city. Environment
International, 2016; 91. 22-28. 10.1016/j.envint.2016.02.018.
Carvalho D, Rocha A, Gomez-Gesteira M, Santos CS. Potential impacts of climate
change on European wind energy resource under the CMIP5 future climate
projections. Renewable Energy 2017; 101: 29-40.

216
Carvalho, D., Pereira, S. and Rocha, A. Future surface temperature changes for the
Iberian Peninsula according to EURO-CORDEX climate projections. Climate
Dynamics. 56. 2021 1-16. 10.1007/s00382-020-05472-3.
Cheeseman, John. Food Security in the Face of Salinity, Drought, Climate Change, and
Population Growth, 2016. 10.1016/B978-0-12-801854-5.00007-8.

Chen YD, Li JF, Zhang Q. Changes in site-scale temperature extremes over China
during 2071-2100 in CMIP5 simulations. Journal of Geophysical Research-
Atmospheres 2016; 121: 2732-2749.
Cherlet M, Hutchinson C, Reynolds J, Hill J, Sommer S, von Maltitz G. (Eds.), World
Atlas of Desertification, Publication Office of the European Union, Luxembourg,
2018. United Nations Environment Program
Christensen JH, Kjellstrom E, Giorgi F, Lenderink G, Rummukainen M. Weight
assignment in regional climate models. Climate Research 2010; 44: 179-194.
Chuine, I., Yiou, P., Viovy, N., Seguin, B., Daux, V., and Ladurie, E. L. Historical
phenology: Grape ripening as a past climate indicator. Nature, 2014. 432, 289–
290.
Chylek P, Li J, Dubey MK, Wang M, Lesins G. Observed and model simulated 20th
century Arctic temperature variability: Canadian Earth System Model CanESM2.
Atmos. Chem. Phys. Discuss. 2011; 11: 22893–22907, doi: 10.5194/acpd-11-
22893-2011
Clark, J.S., Iverson, L., Woodall, C.W., Allen, C.D., Bell, D.M., Bragg, D.C., D'Amato,
A.W., Davis, F.W., Hersh, M.H., Ibanez, I., Jackson, S.T., Matthews, S.,
Pederson, N., Peters, M., Schwartz, M.W., Waring, K.M. and Zimmermann, N.E.
(2016), The impacts of increasing drought on forest dynamics, structure, and
biodiversity in the United States. Glob Change Biol, 22: 2329-
2352. https://doi.org/10.1111/gcb.13160
Collados-Lara AJ, Pulido-Velazquez D, Pardo-Iguzquiza E. An Integrated Statistical
Method to Generate Potential Future Climate Scenarios to Analyse Droughts.
Water 2018; 10.
Collins W, Bellouin N, Doutriaux-Boucher M, Gedney N, Hinton T, Jones CD, Liddicoat
S, Martin G, O’Connor F, Rae J, Senior C, Totterdell I, Woodward S, Reichler T,
Kim J, Halloran P. 2008.Evaluation of the HadGEM2 model. Hadley Centre
Technical Note. HCTN 74, Met Office Hadley Centre, Exeter, UK.
Combris P, Lecocq S, Visser, M. 1997. Estimation of a hedonic price equation for
Bordeaux wine: does quality matter? Econom. J. 1997; 107:3-390
Cook J, Nuccitelli D, Green SA, Richardson M, Winkler B, Painting R, Way R, Jacobs P,
Skuce A. Quantifying the consensus on anthropogenic global warming in the
scientific literature. Environ Res Lett 8:024024. 2013. doi:10.1088/1748-
9326/8/2/024024
Cos, J., Doblas-Reyes, F., Jury, M., Marcos, R., Bretonnière, P.-A., and Samsó, M.: The
Mediterranean climate change hotspot in the CMIP5 and CMIP6 projections,
Earth Syst. Dynam., 13, 321–340, https://doi.org/10.5194/esd-13-321-2022,
2022.
Costa JM, Ortuño MF, Lopes CM, Chaves MM. Grapevine varieties exhibiting differences
in stomatalresponse to water deficit. Funct Plant Biol 39: 179–189, 2012
Dai, A. G., T. B. Zhao, and J. Chen. 2018. "Climate Change and Drought: A Precipitation
and Evaporation Perspective." Current Climate Change Reports 4, no. 3 (Sep):
301-312. http://dx.doi.org/10.1007/s40641-018-0101-6.
Dai AG. Drought under global warming: a review. Wiley Interdisciplinary Reviews-
Climate Change 2011; 2: 45-65.
Dai AG. Increasing drought under global warming in observations and models (vol 3, pg
52, 2013). Nature Climate Change 2013; 3: 171-171.

217
Damiano N, Arena C, Bonfante A, Caputo R, Erbaggio A, Cirillo C, et al. How Leaf Vein
and Stomata Traits Are Related with Photosynthetic Efficiency in Falanghina
Grapevine in Different Pedoclimatic Conditions. Plants-Basel 2022; 11.
De Martino G, De Paola F, Fontana N, Marini G, Ranucci A. Experimental assessment
of level pool routing in preliminary design of floodplain storage. Science of the
Total Environment 2012; 416: 142-147.

De Salvo, M., Raffaelli, R., Moser, R. The impact of climate change on permanent crops
in an Alpine region: A Ricardian analysis. Agricultural Systems, 2013. 118, 23-32
Déqué M, Somot S, Sanchez-Gomez E, Goodess CM, Jacob D, Lenderink G,
Christensen OB. The spread amongst ENSEMBLES regional scenarios: regional
climate models, driving general circulation models and inter-annual variability.
Clim Dyn, 2012. 38:951–964. doi:10.1007/s00382-011-1053-x
Diaz, Henry & Murnane, Richard. Preface: The significance of weather and climate
extremes to society: An introduction. xiii-xvi. 2008
10.1017/CBO9780511535840.002.
D'Ippoliti, D., Michelozzi, P., Marino, C. et al. The impact of heat waves on mortality in 9
European cities: results from the EuroHEAT project. Environ Health, 2010; 9, 37.
https://doi.org/10.1186/1476-069X-9-37
Doll P. Impact of climate change and variability on irrigation requirements: A global
perspective. Climatic Change 2002; 54: 269-293.
Dosio A. Projection of temperature and heat waves for Africa with an ensemble of
CORDEX Regional Climate Models. Climate Dynamics 2017; 49: 493-519.
Downey, M. O., N. K. Dokoozlian, and M. P. Krstic. Cultural practice and environmental
impacts on the flavonoid composition of grapes and wine: a review of recent
research. Am. J. Enol. Vitic., 2006. 57:257–268.
Dracup JA, Lee KS, Paulson EG. ON THE STATISTICAL CHARACTERISTICS OF
DROUGHT EVENTS. Water Resources Research 1980; 16: 289-296.
Duchene E, Huard F, Dumas V, Schneider C, Merdinoglu D. The challenge of adapting
grapevine varieties to climate change. Climate Research 2010; 41: 193-204.
Duchene E, Schneider C. Grapevine and climatic changes: a glance at the situation in
Alsace. Agronomy for Sustainable Development 2005; 25: 93-99.
Duchene, E. How can grapevine genetics contribute to the adaptation to climate change?
OENO One 2016, 50
Duchene E, Butterlin G, Dumas V, Merdinoglu D. Towards the adaptation of grapevine
varieties to climate change: QTLs and candidate genes for developmental
stages. Theor. Appl. Genet. 2012. 124(4): 623–635, doi: 10.1007/s00122-011-
1734-1
Dunne JP, John JG, Adcroft AJ, Griffies SM, Hallberg RW, Shevliakova E, et al. GFDL's
ESM2 Global Coupled Climate-Carbon Earth System Models. Part I: Physical
Formulation and Baseline Simulation Characteristics. Journal of Climate 2012;
25: 6646-6665.
Easterling, D.R., Evans, J.L., Groisman, P. Ya., Karl, T.R., Kunkel, K.E. and Ambenje,
P. Observed variability and trends in extreme climate events: A brief review. Bull.
Amer. Meteorol. Soc. 2000. 81:417-425.
Escariz A, Blanco J, Miranda D, Crecente R. Zonificación y metodología para la
delimitación de comarcas vitícolas. Aplicación a la ampliación de vino de la tierra
de Betanzos. XI congreso internacional de Ingenieria de proyectos Lugo 2007.
Estrela T, Perez-Martin MA, Vargas E. Impacts of climate change on water resources in
Spain. Hydrological Sciences Journal-Journal Des Sciences Hydrologiques
2012; 57: 1154-1167.
Estrella N, Menzel A. Recent and future climate extremes arising from changes to the
bivariate distribution of temperature and precipitation in Bavaria, Germany.
International Journal of Climatology 2013; 33: 1687-1695.

218
EXPANSIÓN, 2016.
https://www.expansion.com/economia/2019/12/16/5df6afff468aebed0c8b467f.ht
ml
EXPANSIÓN 2019
https://www.researchgate.net/publication/345973796_Impacto_adaptacion_y_percepci
on_del_cambio_climatico_en_la_DOCa_Rioja
Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., Stouffer, R. J., and Taylor,
K. E.: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6)
experimental design and organization, Geosci. Model Dev.,2016, 9, 1937-1958
Failla, O., Mariani, L., & Toninato, L. Viticoltura di territorio, 2008. In La vite e il vino (p.
624). Milano: Bayer CropScience.
Fernandez-Montes S, Rodrigo FS. Trends in seasonal indices of daily temperature
extremes in the Iberian Peninsula, 1929-2005. International Journal of
Climatology 2012; 32: 2320-2332.
Feyen L, Dankers R. Impact of global warming on streamflow drought in Europe. Journal
of Geophysical Research-Atmospheres 2009; 114.
FEV, 2021.
https://www.expansion.com/economia/2019/12/16/5df6afff468aebed0c8b467f.ht
ml
Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken, D.J., Ebi, K.L., Mastrandrea, M.D.,
Mach, K.J., Plattner, G.K., Allen, S.K., Tignor, M., Midgley, P.M., IPCC, 2012.
Managing the Risks of Extreme Events and Disasters to Advance Climate
Change Adaptation. Cambridge University Press, Cambridge, United Kingdom
and New York, NY, USA, p. 582 pp.
Fischer EM, Schar C. Consistent geographical patterns of changes in high-impact
European heatwaves. Nature Geoscience 2010; 3: 398-403.
Flato G, Marotzke J, Abiodun B, Braconnot P, Chou SC, Collins W, et al. Evaluation of
Climate Models, 2014.
Flexas J, Galmes J, Galle A, Gulias J, Pou A, Ribas-Carbo M, Tomas M, Medrano H.
Improving water use efficiency in grapevines: potential physiological targets for
biotechnological improvement. Aust. J. Grape Wine Res., 2010. 16: 106–121,
doi: 10.1111/j.1755-0238.2009.00057.x.
Fonseca Conceicao MA, Tonietto J, Fialho FB. Using temperature to calculate the
dryness index of grape production regions. Revista Brasileira De Fruticultura
2012; 34: 175-182.
Fonseca D, Carvalho MJ, Marta-Almeida M, Melo-Goncalves P, Rocha A. Recent trends
of extreme temperature indices for the Iberian Peninsula. Physics and Chemistry
of the Earth 2016; 94: 66-76.
Forzieri G, Feyen L, Rojas R, Florke M, Wimmer F, Bianchi A. Ensemble projections of
future streamflow droughts in Europe. Hydrology and Earth System Sciences
2014; 18: 85-108.
Fragoso M, Carraca MD, Alcoforado MJ. Droughts in Portugal in the 18th century: A
study based on newly found documentary data. International Journal of
Climatology 2018; 38: 5522-5541.
Fraga, H.; García de Cortázar Atauri, I.; Santos, J.A. Viticultural irrigation demands under
climate change scenarios in Portugal. Agric. Water Manag. 2018, 196, 66–74
Fraga H, Malheiro AC, Moutinho-Pereira J, Jones GV, Alves F, Pinto JG, et al. Very high
resolution bioclimatic zoning of Portuguese wine regions: present and future
scenarios. Regional Environmental Change 2014; 14: 295-306.
Fraga H, Malheiro AC, Moutinho-Pereira J, Santos JA. Future scenarios for viticultural
zoning in Europe: ensemble projections and uncertainties. International Journal
of Biometeorology 2013; 57: 909-925.
Fraga H, Santos JA, Malheiro AC, Moutinho-Pereira J. Climate change projections for
the portuguese viticulture using a multi-model ensemble. Ciencia E Tecnica
Vitivinicola 2012; 27: 39-48.

219
Fraga, H.; Santos, J.A. Daily prediction of seasonal grapevine production in the Douro
wine region based on favourable meteorological conditions. Aust. J. Grape Wine
Res. 2017
Fraga, H.; Pinto, J.G.; Santos, J.A. Climate change projections for chilling and heat
forcing conditions in European vineyards and olive orchards: A multi-model
assessment. Clim. Chang. 2019, 152, 179–193
France I, Dubourdieu AD. Climate Change: Field Reports from Leading Winemakers.
Journal of Wine Economics 2016; 11: 5-47.
Fregoni, M. L’indice bioclimatico di qualitá Fregoni. Terroir, Zonazione Viticoltura 2003;
115-127.
Frias MD, Fernandez J, Saenz J, Rodriguez-Puebla C. Operational predictability of
monthly average maximum temperature over the Iberian Peninsula using
DEMETER simulations and downscaling. Tellus Series a-Dynamic Meteorology
and Oceanography 2005; 57.
Frias MD, Herrera S, Cofino AS, Gutierrez JM. Assessing the Skill of Precipitation and
Temperature Seasonal Forecasts in Spain: Windows of Opportunity Related to
ENSO Events. Journal of Climate 2010; 23.
Caffarra, A., Rinaldi, M., Eccel, E., Rossi, V., Pertot, I. Modelling the impact of climate
change on the interaction between grapevine and its pests and pathogens:
European grapevine moth and powdery mildew. Agriculture, Ecosystems and
Environment, 2012. 148, 89-101.
Gaitan E, Monjo R, Portoles J, Pino-Otin MR. Projection of temperatures and heat and
cold waves for Aragon (Spain) using a two-step statistical downscaling of CMIP5
model outputs. Science of the Total Environment 2019; 650: 2778-2795.
Gaitan E, Monjo R, Portoles J, Rosa Pino-Otin M. Impact of climate change on drought
in Aragon (NE Spain). Science of the Total Environment 2020; 740.
Gallego MC, Trigo RM, Vaquero JM, Brunet M, Garcia JA, Sigro J, et al. Trends in
frequency indices of daily precipitation over the Iberian Peninsula during the last
century. Journal of Geophysical Research-Atmospheres 2011; 116.
Gallo AE, Perez Pena JE, Prieto JA. Mechanisms underlying photosynthetic acclimation
to high temperature are different between Vitis vinifera cv. Syrah and Grenache.
Functional Plant Biology 2021; 48: 342-357
Galbreath J. Climate Change Response: Evidence from the Margaret River Wine Region
of Australia. Business Strategy and the Environment, 2014. 23, 89–104.
Ganichot, B. Evolution de la date des vendanges dans les Côtes du Rhône méridionales.
In 6emes Recontres Rhodaniennes, 2002 (pp. 38–41). Orange, France: Institut
Rhodanien.
García de Cortázar-Atauri, I.; Duchêne, E.; Destrac-Irvine, A.; Barbeau, G.; de
Rességuier, L.; Lacombe, T.; Parker, A.K.; Saurin, N.; van Leeuwen, C.
Grapevine phenology in France: From past observations to future evolutions in
the context of climate change. OENO One 2017, 51
Garcia-Barron L, Aguilar M, Sousa A. Evolution of annual rainfall irregularity in the
southwest of the Iberian Peninsula. Theoretical and Applied Climatology 2011;
103: 13-26.
Garcia-Herrera R, Diaz J, Trigo RM, Hernandez E. Extreme summer temperatures in
Iberia: health impacts and associated synoptic conditions. Annales Geophysicae
2005; 23: 239-251.
Carroll, N., Frijters, P., and Shields, M. A. Quantifying the Costs of Drought: New
Evidence from Life Satisfaction Data. Journal of Population Economics,
2009, 22(2), 445–461. http://www.jstor.org/stable/40344739
Choat, Brendan & Jansen, Steven & Brodribb, Tim & Cochard, Hervé & Delzon, Sylvain
& Bhaskar, Radika & Bucci, Sandra & Feild, Taylor & Gleason, Sean & Hacke,
Uwe & Jacobsen, Anna & Lens, Frederic & Maherali, Hafiz & Martinez Vilalta,
Jordi & Mayr, Stefan & Mencuccini, Maurizio & Mitchell, Patrick & Nardini, Andrea

220
 & Pittermann, Jarmila & Zanne, Amy. (2012). Global convergence in the
vulnerability of forests to drought. Nature. 491. 10.1038/nature11688.
Gilaberte-Burdalo M, Lopez-Martin F, Pino-Otin MR, Lopez-Moreno JI. Impacts of
climate change on ski industry. Environmental Science & Policy 2014; 44: 51-61.
Gilaberte-Burdalo M, Lopez-Moren JI, Moran-Tejeda E, Jerez S, Alonso-Gonzalez E,
Lopez-Martin F, et al. Assessment of ski condition reliability in the Spanish and
Andorran Pyrenees for the second half of the 20th century. Applied Geography
2017; 79: 127-142.
Giorgi F, Jones C, Asrar GR. Addressing climate information needs at the regional level:
the CORDEX framework. WMO Bull. 2009; 58(3):175–183
Giorgi F, Gutowski WJ. Regional Dynamical Downscaling and the CORDEX Initiative.
Annual Review of Environment and Resources, Vol 40 2015; 40: 467-490.
Giorgi F, Mearns LO. Approaches to the simulation of regional climate change - a review.
Reviews of Geophysics 1991; 29: 191-216.
Gladstones, J. Viticulture and environment. Winetitles ed. Winetitles; Underdale, South
Australia, 1992; Australia.
Gobiet A, Kotlarski S, Beniston M, Heinrich G, Rajczak J, Stoffel M. 21st century climate
change in the European Alps-A review. Science of the Total Environment 2014;
493: 1138-1151.
Gomez-Gesteira M, Gimeno L, deCastro M, Lorenzo MN, Alvarez I, Nieto R, et al. The
state of climate in NW Iberia. Climate Research 2011; 48: 109-144.
Gomez-Martinez G, Galiano L, Rubio T, Prado-Lopez C, Redolat D, Blazquez CP, et al.
Effects of Climate Change on Water Quality in the Jucar River Basin (Spain).
Water 2021; 13
Gommes, R., and Petrassi, F. Rainfall variability and drought in sub-Saharan Africa since
1960. Agrometeorology Series Working Paper 9, Food and Agriculture
Organization, 1994, Rome, Italy, 100 pp.
Goncalves B, Falco V, Moutinho-Pereira J, Bacelar E, Peixoto F, Correia C. Effects of
Elevated CO2 on Grapevine (Vitis vinifera L.): Volatile Composition, Phenolic
Content, and in Vitro Antioxidant Activity of Red Wine. Journal of Agricultural and
Food Chemistry 2009; 57: 265-273.
Goncalves M, Barrera-Escoda A, Guerreiro D, Baldasano JM, Cunillera J. Seasonal to
yearly assessment of temperature and precipitation trends in the North Western
Mediterranean Basin by dynamical downscaling of climate scenarios at high
resolution (1971-2050). Climatic Change 2014; 122: 243-256.
González, J., and Valdés, J. 2006. New drought frequency index: Definition and
comparative performance analysis. Water Resour. Res. 42 (11): W11421.
doi:10.1029/2005WR004308.
Gordo, O. and Sanz, J.J. Climate change and bird phenology: a long-term study in the
Iberian Peninsula. Global Change Biology, 2006; 12: 1993-2004.
https://doi.org/10.1111/j.1365-2486.2006.01178.x
Graça, A., Gishen, M. and Jones, G. Proposal for the development of a framework for a
globally relevant wine sector climate change adaptation strategy 2016.
10.13140/RG.2.2.11875.14883.
Greenwood O, Mossman HL, Suggitt AJ, Curtis RJ, Maclean IMD (2016) Using in situ
management to conserve biodiversity under climate change. Journal of Applied
Ecology, doi: 10.1111/1365-2664.12602.
Greer, D. H., M. M. Weedon, and C. Weston. Reductions in biomass accumulation,
photosynthesis in situ and net carbon balance are the costs of protecting Vitis
vinifera “Semillon” grapevines from heat stress with shade covering, 2011. AoB
plants. 2011: plr023.
Grifoni, D., G. Carreras, G. Zipoli, F. Sabatini, A. Dalla Marta, and S. Orlandini. Row
orientation effect on UV-B, UV-A and PAR solar irradiation components in
vineyards at Tuscany, 2008. Italy. Int. J. Biometeorol. 52:755–763.

221
Gross MH, Alexander LV, Macadam I, Green D, Evans JP. The representation of health-
relevant heatwave characteristics in a Regional Climate Model ensemble for New
South Wales and the Australian Capital Territory, Australia. International Journal
of Climatology 2017; 37: 1195-1210.
Gu, S., Ding, P. and Howard, S. Effect of temperature and exposure time on cold
hardiness of primary buds during the dormant season in 'Concord', 'Norton',
'Vignoles' and 'St. Vincent' grapevines. Journal of Horticultural Science and
Biotechnology 2002.77. 635-639. 10.1080/14620316.2002.11511550.
Gudmundsson L, Rego F C, Rocha M and Seneviratne S I. Environ. Res. Lett., 2014. 9
Guerreiro SB, Dawson RJ, Kilsby C, Lewis E, Ford A. Future heat-waves, droughts and
floods in 571 European cities. Environmental Research Letters 2018; 13.
Guido V, Impact of Climate Change on Vitis Vinifera L. over Mediterranean Area. Tesi di
Dottorato in Agrometeorologia ed Ecofisiologia dei Sistemi Agrari e Forestali,
Università degli Studi di Sassari. 2015.
Gutierrez JM, Maraun D, Widmann M, Huth R, Hertig E, Benestad R, et al. An
intercomparison of a large ensemble of statistical downscaling methods over
Europe: Results from the VALUE perfect predictor cross-validation experiment.
International Journal of Climatology 2019; 39: 3750-3785.
Guttman NB. Comparing the Palmer Drought Index and the standardized precipitation
index. Journal of the American Water Resources Association 1998; 34: 113-121.
Hagman G. Prevention better than cure: report on human and natural disasters in the
Third World, 1984. Swedish Red Cross, Stockholm, Sweden.
Hall, A., and G. V. Jones. Effect of potential atmospheric warming on temperature-based
indices describing Australian winegrape growing conditions. Aust. J. Grape Wine
R, 2009, 15:97– 119
Hannah L, Roehrdanz PR, Ikegami M, Shepard AV, Shaw MR, Tabor G, et al. Climate
change, wine, and conservation. Proceedings of the National Academy of
Sciences of the United States of America 2013; 110: 6907-6912.
Hansen J, Sato M, Ruedy R. Perception of climate change. Proceedings of the National
Academy of Sciences of the United States of America 2012; 109: E2415-E2423.
Hancock J.M. Feature Review – `Terrior. The Role of Geology, Climate, and Culture in
Making of French Wines’ by Wilson J.E. Journal of Wine Research 10, 1999. (1),
43–49.
Hancock J.M. Geology of wine. In Selley R.C. et al., (Eds.). The Encyclopaedia of
Geology Volume III, Elsevier, Amsterdam, 2005. 85–90.
Hao ZC, Hao FH, Singh VP, Ouyang W. Quantitative risk assessment of the effects of
drought on extreme temperature in eastern China. Journal of Geophysical
Research-Atmospheres 2017; 122: 9050-9059.
Harbertson, J. F., and M. Keller. Rootstock effects on deficit-irrigated winegrapes in a
dry climate: grape and wine composition. Am. J. Enol. Vitic., 2012. 63:40–48.
Hargreaves GL, Samani ZA. Reference crop evapotranspiration from temperature. Appl.
Eng. Agric. 1985;1: 96–99.
Hartmann, D.L., Klein Tank, A.M.G., Rusticucci, M., Alexander, L.V., Brönnimann, S.,
Kaplan,A., Soden, B.J., Thorne, P.W., Wild, M., Zhai, P.M. Observations:
Atmosphere and Surface. In T.F. Stocker, D. Qin, G.K. Plattner, M. Tignor, S.K.
Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, P.M. Midgley (eds.), Climate
Change 2013: The Physical Science Basis. Contribution of Working Group I to
the Fifth Assessment Report of the Intergovernmental Panel on Climate Change,
Cambridge University Press, Cambridge, United Kingdom and New York, NY,
USA. 2013.
Hausfather Z, Peters GP. Emissions - the 'business as usual' story is misleading. Nature
2020; 577: 618-620.
Hidalgo, T. and Togores, J. EL concepto de "Terroir" en el viñedo. Ponencia en el XII
curso de verano, Innovación vitivinícola en la ribera del Duero, sostenibilidad

222
 II.2012. Editado por el consejo regulador de la denominación de origen Ribera
del Duero, pp 9-45
Hayes M, Svoboda M, Wall N, Widhalm M. The lincoln declaration on drought indices.
Bulletin of the American Meteorological Society 2011; 92: 485-488.
Hayes MJ, Svoboda MD, Wilhite DA, Vanyarkho OV. Monitoring the 1996 drought using
the standardized precipitation index. Bulletin of the American Meteorological
Society 1999; 80: 429-438.
Heavens N, Ward D and Natalie M. Studying and Projecting Climate Change with Earth
System Models. Nature Education Knowledge.2013; 4(5):4
Hertig E, Seubert S, Jacobeit J. Temperature extremes in the Mediterranean area: trends
in the past and assessments for the future. Natural Hazards and Earth System
Sciences 2010; 10: 2039-2050.
Hervada-Sala C, Pawlowsky-Glahn V, Jarauta-Bragulat E. A statistical method to
downscale temperature forecasts. A case study in Catalonia. Meteorological
Applications 2000; 7.
Hoerling M, Eischeid J, Perlwitz J, Quan XW, Zhang T, Pegion P. On the Increased
Frequency of Mediterranean Drought. Journal of Climate 2012; 25: 2146-2161.
Holland, T., Smit, B., 2010. Climate change and the wine industry: current research
themes and new directions. Journal of Wine Research 21, 125-136.
Holland T, Smit B. Climate change and the wine industry: current research themes and
new directions. J. Wine Res. 2010; 21:125-136.
Howden SM, Soussana J-F, Tubiello FN, Chhetri N, Dunlop M, Meinke H. Adapting
agriculture to climate change. Proceedings of the National Academy of Sciences,
2007, 104, 19691-19696.
Hu Q, Willson GD. Effects of temperature anomalies on the Palmer Drought Severity
Index in the central United States. International Journal of Climatology 2000; 20:
1899-1911.
Huglin P. Nouveau Mode d’Évaluation des Possibilités Héliothermiques d’un Milieu
Viticole. C. R. Acad. Agr. France, 1978; 1117-1126.
Hunter J.J., Volschenk C.G. and Bonnardot V., 2010. Linking grapevine row orientation
to a changing climate in South Africa. In : Proceedings of the Inter vitis Interfructa
Conference, Stuttgart (Germany), pp. 60-70
Hussein M, Cholette S, Castaldi RM. An Analysis of Globalization Forces in the Wine
Industry. Journal of Global Marketing, 2008, 21, 33-47.
Ibacache, A., Martínez, L. and Sturla, C. Zonificación del territorio de la denominación
de origen Pisco. Instituto Nacional de Investigaciones Agropecuarias INIA: La
Serena (Chile). 2010
Iglesias A, Garrote L, Flores F, Moneo M. Challenges to manage the risk of water scarcity
and climate change in the Mediterranean. Water Resources Management 2007;
21: 775-788.
Iglesias A. y Medina F. (2009). Consecuencias del cambio climático para la agricultura:
¿un problema de hoy o del futuro?. Revista española de Estudios Agrosociales
y Pesqueros, nº 221, 2009.
Iglesias A, Quiroga S, Moneo M, Garrote L. From climate change impacts to the
development of adaptation strategies: Challenges for agriculture in Europe.
Climatic Change 2012; 112: 143-168
Imbert A, Benestad RE. An improvement of analog model strategy for more reliable local
climate change scenarios. Theoretical and Applied Climatology 2005; 82: 245-
255.
Intrieri, C., S. Poni, B. Rebucci, and E. Magnanini. 1998. Row orientation effects on
whole-canopy gas exchange of potted and field-grown grapevines. JKI,
Siebeldingen, Germany, 37
IPCC, 2021a: Climate Change 2021: The Physical Science Basis. Contribution of
Working Group I to the Sixth Assessment Report of the Intergovernmental Panel
on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C.

223
 Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K.
Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R.
Yu, and B. Zhou (eds.)]. Cambridge University Press. In Press
IPCC, 2021b: Summary for Policymakers. In: Climate Change 2021: The Physical
Science Basis. Contribution of Working Group I to the Sixth Assessment Report
of the Intergovernmental Panel on Climate Change [MassonDelmotte, V., P. Zhai,
A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I.
Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T.
Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press.
In Press.
IPCC, 2014a Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A:
Global and Sectoral Aspects. Contribution of Working Group II to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, Cambridge, United Kingdom and New York, NY,
USA, p. 1132
IPCC, 2014b Climate Change 2014: Synthesis Report. Contribution of Working Groups
I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)].
IPCC, Geneva, Switzerland, 151 pp.
IPCC, 2013 In: Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels
A, Xia Y, Bex V, Midgley PM (eds) Climate Change 2013: the Physical Science
Basis. Contribution of Working Group I to the Fifth Assessment Report of the
Intergovernmental PanelonClimateChange. Cambridge UniversityPress,
Cambridge. doi:10.1017/CBO9781107415324
IPCC, 2013: Glosario [Planton, S. (ed.)]. En: Cambio Climático 2013. Bases físicas.
Contribución del Grupo de trabajo I al Quinto Informe de Evaluación del Grupo
Intergubernamental de Expertos sobre el Cambio Climático [Stocker, T.F., D.
Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex
y P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, Reino Unido y
Nueva York, NY, Estados Unidos de América.
IPCC, 2007: Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, 2007. Solomon, S., D.Qin,
M.Manning, Z.Chen, M.Marquis, K.B. Averyt, M. Tiggnor and H.L. Miller (eds)
Cambridge University Press, Cambridge, United Kingdom and New York, NY,
USA.
IPCC, 2001: climate change 2001: the scientific basis. Contribution of Working Group 1
to the Third Assessment Report of the Intergovernmental Panel on Climate
Change, edited by J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. van der
Linden, X. Dai, K. Maskell and C. A. Johnson (eds). Cambridge University Press,
Cambridge, UK, and New York, USA, 2001
IPCC, 2000. Special report on emissions scenarios‚ Cambridge University Press‚
Cambridge, UK. ISBN 0 521 80493 0
Irimia, L., Patriche, C., and Quénol, H. Viticultural zoning: A comparative study regarding
the accuracy of different approaches in vineyards climate suitability assessment.
Cercetari agronomice in Moldova, 2013, 46(3), 95-106.
Irmak S, Kabenge I, Skaggs KE, Mutiibwa D. Trend and magnitude of changes in climate
variables and reference evapotranspiration over 116-yr period in the Platte River
Basin, central Nebraska-USA. Journal of Hydrology 2012; 420: 228-244.
Isaksen ISA, Wang WC. Atmospheric ozone and climate change. 3rd International
Symposium on Non-CO2 Greenhouse Gases, Maastricht, Netherlands, 2002, pp.
319-330.
Iversen T, Bentsen M, Bethke I, Debernard JB, Kirkevag A, Seland O, et al. The
Norwegian Earth System Model, NorESM1-M - Part 2: Climate response and
scenario projections. Geoscientific Model Development 2013; 6: 389-415.
Jensen M and Haise H. Estimating Evapotranspiration from Solar Radiation. Journal

224
Jeong DI, Sushama L, Diro GT, Khaliq MN, Beltrami H, Caya D. Projected changes to
high temperature events for Canada based on a regional climate model
ensemble. Climate Dynamics 2016; 46: 3163-3180.
Jiménez-Donaire, M.d.P.; Giráldez, J.V.; Vanwalleghem, T. Impact of Climate Change
on Agricultural Droughts in Spain. Water 2020, 12, 3214.
https://doi.org/10.3390/w12113214
Jones, Gregory V. and Webb, Leanne B. 'Climate Change, Viticulture, and Wine:
Challenges and Opportunities', Journal of Wine Research, 2010, 21: 2, 103 —
106
Jones, G.V. and Davis, R.E. Climate influences on grapevine phenology, grape
composition, and wine production and quality for Bordeaux, France. American
Journal of Enology and Viticulture, 2000, 51, 249–261
Jones CD, Hughes JK, Bellouin N, Hardiman SC, Jones GS, Knight J, et al. The
HadGEM2-ES implementation of CMIP5 centennial simulations. Geoscientific
Model Development 2011; 4: 543-570.
Jones GV, Alves F. Impact of climate change on wine production: a global overview and
regional assessment in the Douro Valley of Portugal. International Journal of
Global Warming 2012; 4: 383-406.
Jones GV, Davis RE. Climate influences on grapevine phenology, grape composition,
and wine production and quality for Bordeaux, France. American Journal of
Enology and Viticulture 2000; 51: 249-261.
Jones GV, White MA, Cooper OR, Storchmann K. Climate change and global wine
quality. Climatic Change 2005; 73: 319-343.
Jones, G.V. Climate and Terroir: Impacts of Climate Variability and Change on Win, in
Fine Wine and Terroir - The Geoscience Perspective. Macqueen, R.W., and
Meinert, L.D., (eds.), Geoscience Canada Reprint Series Number 9, Geological
Association of Canada, St. John's, Newfoundland, 2006. 247 p.
Kattenberg A, Amer Meteorol SOC. Climate models: Projections of future climate.
Seventh Symposium on Global Change Studies 1996.
Kenny GH, Harrison PA. The effects of climatic variability and change on grape suitability
in Europe. J Wine Res 1993; 4:163–183.
Kent ST, McClure LA, Zaitchik BF, Smith TT, Gohlke JM. Heat Waves and Health
Outcomes in Alabama (USA): The Importance of Heat Wave Definition.
Environmental Health Perspectives 2014; 122: 151-158.
Khalili M, Nguyen VTV. An efficient statistical approach to multi-site downscaling of daily
precipitation series in the context of climate change. Climate Dynamics 2017; 49:
2261-2278.
Kim S, Sinclair VA, Raisanen J, Ruuhela R. Heat waves in Finland: present and projected
summertime extreme temperatures and their associated circulation patterns.
International Journal of Climatology 2018; 38: 1393-1408.
Kizildeniz T, Pascual I, Irigoyen JJ, Morales F. Future CO2, warming and water deficit
impact white and red Tempranillo grapevine: Photosynthetic acclimation to
elevated CO2 and biomass allocation. Physiologia Plantarum 2021; 172: 1779-
1794.
Knutti R, Sedlacek J. Robustness and uncertainties in the new CMIP5 climate model
projections. Nature Climate Change 2013; 3: 369-373.
Koufos G, Mavromatis T, Koundouras S, Fyllas MN, Jones GV. 2014. Viticulture–climate
relationships in Greece: the impact of recent climate trends on harvest dates
variation. Int. J. Climatol. 34(5): 1445–1459
Koufos, G., Mavromatis, T., Koundouras, S. and Jones, G. Response of viticulture-
related climatic indices and zoning to historical and future climate conditions in
Greece: response of viticulture-related climatic indices and zoning in greece.
International Journal of Climatology, 2017. 38. 10.1002/joc.5320.
Koundouras, S., C. Van Leeuwen, G. Seguin, and Y. Glories. Influence of water status
on vine vegetative growth, berry ripening and wine characteristics in

225
 mediterranean zone (example of Nemea, Greece, variety Saint-George, 1997).
J. Int. Sci. Vigne. Vin. 1999. 33:149–160.
Koundouras, S., I. T. Tsialtas, E. Zioziou, and N. Nikolaou. Rootstock effects on the
adaptive strategies of grapevine (Vitis vinifera L. cv. Cabernet-Sauvignon) under
contrasting water status: leaf physiological and structural responses. Agric.
Ecosyst. Environ., 2008, 128:86–96

Kovaleski AP, Londo JP. Tempo of gene regulation in wild and cultivated Vitis species
shows coordination between cold deacclimation and budbreak. Plant Science
2019; 287
Köppen W, Geiger R. Das geographische System der Klimate, 1936.
Kuglitsch FG, Toreti A, Xoplaki E, Della-Marta PM, Zerefos CS, Turkes M, et al. Heat
wave changes in the eastern Mediterranean since 1960. Geophysical Research
Letters 2010; 37.
Kulakowski D, Seidl R, Holeksa J, Kuuluvainen T, Nagel TA, Panayotov M, et al. A walk
on the wild side: Disturbance dynamics and the conservation and management
of European mountain forest ecosystems. Forest Ecology and Management
2017; 388: 120-131.
Lastrada, E.; Cobos, G.; Torrijo, F.J. Analysis of Climate Change’s Effect on Flood Risk.
Case Study of Reinosa in the Ebro River Basin. Water 2020, 12, 1114.
https://doi.org/10.3390/w12041114
Lavaysse C, Vrac M, Drobinski P, Lengaigne M, Vischel T. Statistical downscaling of the
French Mediterranean climate: assessment for present and projection in an
anthropogenic scenario. Natural Hazards and Earth System Sciences 2012; 12:
651-670.
Lazoglou G, Anagnostopoulou C, Koundouras S. Climate change projections for Greek
viticulture as simulated by a regional climate model. Theoretical and Applied
Climatology 2018; 133: 551-567.
Lee SH, Yoo SH, Choi JY, Bae S. Assessment of the Impact of Climate Change on
Drought Characteristics in the Hwanghae Plain, North Korea Using Time Series
SPI and SPEI: 1981-2100. Water 2017; 9.
Lemos MC, Kirchhoff CJ, Ramprasad V. Narrowing the climate information usability gap.
Nature Climate Change, 2012, 2, 789–794.
Leolini, L.; Moriondo, M.; Romboli, Y.; Gardiman, M.; Costafreda-Aumedes, S.; de
Cortazar-Atauri, I.G.; Bindi, M.; Granchi, L.; Brilli, L. Modelling sugar and acid
content in Sangiovese grapes under future climates: An Italian case study. Clim.
Res. 2019, 78, 211–224
Lereboullet A-L, Beltrando G, Bardsley DK. Socio-ecological adaptation to climate
change: A comparative case study from the Mediterranean wine industry in
France and Australia. Agriculture, Ecosystems & Environment, 2013a, 164, 273–
285.
Lesk C, Rowhani P, Ramankutty N. Influence of extreme weather disasters on global
crop production. Nature 2016; 529: 84-+
Lewis, S.L., Brando, P.M., Phillips, O.L., van der Heijden, G.M.F., Nepstad, D., The 2010
Amazon drought. Science, 2011. 331 (6017), 554.
http://dx.doi.org/10.1126/science.1200807.
Linares, C., Diaz, J., Tobías, A. et al. Impact of heat and cold waves on circulatory-cause
and respiratory-cause mortality in Spain: 1975–2008. Stoch Environ Res Risk
Assess, 2015; 29, 2037–2046. https://doi.org/10.1007/s00477-014-0976-2
Lin S, Luo M, Walker RJ, Liu X, Hwang SA, Chinery R. Extreme High Temperatures and
Hospital Admissions for Respiratory and Cardiovascular Diseases. Epidemiology
2009; 20: 738-746.
Livneh B, Hoerling MP. The Physics of Drought in the US Central Great Plains. Journal
of Climate 2016; 29: 6783-6804.

226
Lloyd-Hughes B, Saunders MA. A drought climatology for Europe. International Journal of
Climatology 2002; 22: 1571-1592.
Lobell, D.; Field, C.; Cahill, K. y Bonfils, C. “Impacts of future climate change on California
perennial crop yields: Model projections with climate and crop uncertainties“.
Agric For Meteorol., 2006, 141 (2-4), pgs.208-218
Lopes, J., J. E. Eiras-Dias, F. Abreu, P. Climaco, J. P. Cunha, and J. Silvestre. Thermal
requirements, duration and precocity of phenological stages of grapevine
cultivars of the Portuguese collection. Ci^encia Tec. Vitiv., 2008, 23:61–71.

Lopez-Bustins JA, Pascual D, Pla E, Retana J. Future variability of droughts in three

Mediterranean catchments. Natural Hazards 2013; 69: 1405-1421.
López F, Cabrera M, Cuadrat JM. Atlas Climático de Aragón. First ed.Spain: J. Factory;
2007.
Lopez-Moreno JI, Goyette S, Vicente-Serrano SM, Beniston M. Effects of climate change
on the intensity and frequency of heavy snowfall events in the Pyrenees. Climatic
López-Moreno, J. I., Revuelto, J., Rico, I., Chueca-Cía, J., Julián, A., Serreta, A.,
Serrano, E., Vicente-Serrano, S. M., Azorin-Molina, C., Alonso-González, E., and
García-Ruiz, J. M.: Thinning of the Monte Perdido Glacier in the Spanish
Pyrenees since 1981, The Cryosphere, 2016; 10, 681–694,
https://doi.org/10.5194/tc-10-681-2016,
López-Moreno, J. I., Vicente-Serrano, S. M., Moran-Tejeda, E., Zabalza, J., Lorenzo-
Lacruz, J., and García-Ruiz, J. M.: Impact of climate evolution and land use
changes on water yield in the Ebro basin, Hydrol. Earth Syst. Sci.,2011; 15, 311–
322, https://doi.org/10.5194/hess-15-311-2011
Lorenzo, M. N.; Taboada, J. J.; Lorenzo, J. F.; Ramos, A. M. Influence of climate on
grape production and wine quality in the Rías Baixas, north-western Spain.
Regional Environmental Change, 2013, 13(4), 887–896. doi:10.1007/s10113-
012-0387-1
Lorenzo, M.N. and Alvarez, I., Climate change patterns in precipitation over Spain using
CORDEX projections for 2021–2050, Science of The Total Environment, Volume
723,2020,138024,ISSN 0048-9697,
https://doi.org/10.1016/j.scitotenv.2020.138024.
Lorenzo M.N., Díaz-Poso, A. and Dominic R. Heatwave intensity on the Iberian
Peninsula: Future climate projections. Atmospheric Research,
2021. doi:10.1016/j.atmosres.2021.105655
Lorenzo, M.N. and Alvarez, I. Future changes of hot extremes in Spain: towards warmer
conditions. Natural Hazards. 2022, 113. 10.1007/s11069-022-05306-x.
Lorenzo MN, Taboada JJ, Lorenzo JF, Ramos AM (2013) Influence of climate on grape
production and wine quality in the Rı´as Baixas, north-western Spain. Reg
Environ Change 13:887–896. doi:10.1007/s10113-012-0387-1
Lorenzo, Nieves & Ramos, Alexandre & Brands, Swen. (2016). Present and future
climate conditions for winegrowing in Spain. Regional Environmental Change.
2016. 617. 10.1007/s10113-015-0883-1.
Machado MJ, Benito G, Barriendos M, Rodrigo FS. 500 Years of rainfall variability and
extreme hydrological events in southeastern Spain drylands. Journal of Arid
Environments 2011; 75: 1244-1253.
Maclean IMD, Hopkins JJ, Bennie J, Lawson CR and Wilson RJ. Microclimates buffer
the responses of plant communities to climate change. Global Ecology and
Biogeography, 2015, 24, 1340–1350

Magalhaes, N. Tratado de viticultura: a videira, a vinha e o terroir. Chaves Ferreira

Publicacoes, Lisboa, Portugal 2008.
MAGRAMA, 2016 : Análisis de los procesos de desertificación en España en función de
los distintos escenarios climáticos.

227
Makra, L., B. Vitanyi, A. Gal, J. Mika, I. Matyasovszky, and T. Hirsch. Wine quantity and
quality variations in relation to climatic factors in the Tokaj (Hungary), 2009.
Malheiro AC, Santos JA, Fraga H, Pinto JG. Climate change scenarios applied to
viticultural zoning in Europe. Climate Research 2010; 43: 163-177.
Marcos-Garcia P, Lopez-Nicolas A, Pulido-Velazquez M. Combined use of relative
drought indices to analyze climate change impact on meteorological and
hydrological droughts in a Mediterranean basin. Journal of Hydrology 2017; 554:
292-305.
Marko, K., Prtenjak, M., Šimon, S., Osrečak, M., Anić, M.,Karoglan Kontic, J., Andabaka,
Ž., Tomaz, I., Grisogono, B., Belušić Vozila, A., Marki, A., Prša, Ž., Omazić, B.,
Jelić, D., Večenaj, Ž., Vučetić, V., Počakal, D., Petric, I., Leder, R. and Prša, Ivan.
Classification of Croatian winegrowing regions based on bioclimatic indices. E3S
Web of Conferences, 2018, 50. 01032. 10.1051/e3sconf/20185001032.

Marsland SJ, Haak H, Jungclaus JH, Latif M, Roske F. The Max-Planck-Institute global
ocean/sea ice model with orthogonal curvilinear coordinates. Ocean Modelling
2003; 5: 91-127.
Martínez-Casasnovas JA, Ramos MC, Ribes-Dasi M. Soil erosion caused by extreme
rainfall events: mapping and quantification in agricultural plots from very detailed
digital elevation models. Geoderma, 2002, 105:125–140
Maxwell, J. T., Darren L. F., Grant Logan H. and Gregory V. J.. “Projecting future
winegrape yields using a combination of Vitis vinifera L. growth rings and soil
moisture simulations, northern California, USA.” Australian Journal of Grape and
Wine Research 22, 2016, 73-80.
McVicar TR, Roderick ML, Donohue RJ, Li LT, Van Niel TG, Thomas A, et al. Global
review and synthesis of trends in observed terrestrial near-surface wind speeds:
Implications for evaporation. Journal of Hydrology 2012a; 416: 182-205.
McVicar TR, Roderick ML, Donohue RJ, Van Niel TG. Less bluster ahead?
Ecohydrological implications of global trends of terrestrial near-surface wind
speeds. Ecohydrology 2012b; 5: 381-388.
Mckee T, Doesken N and Kleist J. The Relationship of Drought Frequency and Duration
Times Scales. American Meteorological Society. 8th Conference on Applied
Climatology. 1993: January 17–22 Anaheim, California, pp. 179–184.
Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye AT, Gregory JM et al. Global
climate projections 2007: the physical science basis 2007; 747–845
Menzel A, Seifert H, Estrella N. Effects of recent warm and cold spells on European plant
phenology. Int. J. Biometeorol., 2011, 55(6): 921–932, doi: 10.1007/s00484-011-
0466-x
Michael A, S.J, Enke W, Deutschländer Th, M. G. Impact of expected increase in
precipitation intensities on soil loss — results of comparative model simulations.
Catena, 2005, 61:155–164
Ministerio de Medio Ambiente, 2005. Assessment report of the preliminary impacts in
Spain due to Climate Change. Centro de publicaciones del Ministerio de Medio
Ambiente, Madrid.
Miró JJ, Estrela MJ, Caselles V, Olcina-Cantos J. Fine-scale estimations of bioclimatic
change in the Valencia region, Spain. Atmospheric Research 2016; 180: 150-
164.
Miró J.J.; Estrela, M.J.; Olcina-Cantos, J.; Martin-Vide, J. Future Projection of
Precipitation Changes in the Júcar and Segura River Basins (Iberian Peninsula)
by CMIP5 GCMs Local Downscaling. Atmosphere 2021, 12, 879. https://
doi.org/10.3390/atmos12070879
Mishra AK, Singh VP. A review of drought concepts. Journal of Hydrology 2010; 391:
204-216.
Mishra AK, Singh VP. Drought modeling - A review. Journal of Hydrology 2011; 403:
157-175.

228
Molina, M.O., Sánchez, E. & Gutiérrez, C. Future heat waves over the Mediterranean
from an Euro-CORDEX regional climate model ensemble. Sci Rep 10, 8801
2020. https://doi.org/10.1038/s41598-020-65663-0
Molitor, D.; Junk, J. Climate change is implicating a two-fold impact on air temperature
increase in the ripening period under the conditions of the Luxembourgish
grapegrowing region. OENO One 2019, 53, 409–422.
Monjo R, Gaitan E, Portoles J, Ribalaygua J, Torres L. Changes in extreme precipitation
over Spain using statistical downscaling of CMIP5 projections. International
Journal of Climatology 2016; 36: 757-769.
Monjo R., Paradinas C., Gaitán E., Redolat D., Paradinas C., Prado C., Portolés J.,
Torres L., Ribalaygua J. Russo B., Velasco M., Pouget L., Vela S., David M. L.,
Morais M., Ribalaygua J. Report on extreme events predictions. Deliverable 1.3
RESCCUE Project. Grant Agreement No. 700174, 2018
Morales-Castilla, I.; García de Cortázar-Atauri, I.; Cook, B.I.; Lacombe, T.; Parker, A.;
van Leeuwen, C.; Nicholas, K.A.; Wolkovich, E.M. Diversity buffers winegrowing
regions from climate change losses. Proc. Natl. Acad. Sci. USA 2020
Moral García, F. J., Rebollo, F., Paniagua, L., Garcia, A. and Salazar, Enrique.
Application of climatic indices to analyse viticultural suitability in Extremadura,
south-western Spain. Theoretical and Applied Climatology, 2015, 123.
10.1007/s00704-014-1363-0.
Moriondo M, Maselli F, Bindi M. A simple model of regional wheat yield based on NDVI
data. European Journal of Agronomy 2007; 26: 266-274.
Mosedale JR, Abernethy KE, Smart RE, Wilson RJ, Maclean IMD. Climate change
impacts and adaptive strategies: lessons from the grapevine. Global Change
Biology 2016; 22: 3814-3828.
Moss RH, Edmonds JA, Hibbard KA, Manning MR, Rose SK, van Vuuren DP, et al. The
next generation of scenarios for climate change research and assessment.
Nature 2010; 463: 747-756.
Moutahir H, Bellot P, Monjo R, Bellot J, Garcia M, Touhami I. Likely effects of climate
change on groundwater availability in a Mediterranean region of Southeastern
Spain. Hydrological Processes 2017; 31: 161-176.
Moutinho-Pereira, J., N. Magalh~aes, B. Goncalves, E. Bacelar, M. Brito, and C. Correia.
Gas exchange and water relations of three Vitis vinifera L. cultivars growing under
Mediterranean climate. Photosynthetica, 2007, 45:202–207
Moutinho-Pereira J, Goncalves B, Bacelar E, Cunha JB, Coutinho J, Correia CM. Effects
of elevated CO2 on grapevine (Vitis vinifera L.): Physiological and yield attributes.
Vitis 2009; 48: 159-165.
Mukherjee S, Mishra A, Trenberth KE. Climate Change and Drought: a Perspective on
Drought Indices. Current Climate Change Reports 2018; 4: 145-163.
Naulleau, A,. Gary, C., Prevot, L. and Hossard, L. Evaluating Strategies for Adaptation
to Climate Change in Grapevine Production–A Systematic Review. Frontiers in
Plant Science 2021. 11. 10.3389/fpls.2020.607859.
Neethling E., Barbeau G., Bonnefoy C. and Quénol H., 2012. Change in climate and
berry composition for grapevine varieties cultivated in the Loire Valley. Clim. Res.,
53, 89-101.
Neumann PA, Matzarakis A. Viticulture in southwest Germany under climate change
conditions. Climate Research 2011; 47: 161-169
Nychka D, Furrer R, Paige J and Sain S (2015). Fields: Tools for spatial data. R package
version 9.0 doi: 10.5065/D6W957CT (URL: http://doi.org/10.5065/D6W957CT)
<URL: www.image.ucar.edu/fields>.
Odo Camps J, Ramos MC. Grape harvest and yield responses to inter-annual changes
in temperature and precipitation in an area of north-east Spain with a
Mediterranean climate. International Journal of Biometeorology 2012; 56: 853-
864.

229
Ollat, N.; Touzard, J.M.; van Leeuwen, C. Climate Change Impacts and Adaptations:
New Challenges for the Wine Industry. J. Wine Econ. 2016, 11, 139–149
Olesen JE, Trnka M, Kersebaum KC, Skjelvag AO, Seguin B, Peltonen-Sainio P, et al.
Impacts and adaptation of European crop production systems to climate change.
European Journal of Agronomy 2011; 34: 96-112.
Orduna, R. M. Climate change associated effects on grape and wine quality and
production. Food Res. Int., 2010, 43:1844–1855
OPCC2, 2018: Climate change in the Pyrenees: impacts, vulnerability and adaptation.
Executive summary of the OPCC-CTP 2018 report.
Ojeda MGV, Gamiz-Fortis SR, Castro-Diez Y, Esteban-Parra MJ. Evaluation of WRF
capability to detect dry and wet periods in Spain using drought indices. Journal
of Geophysical Research-Atmospheres 2017; 122: 1569-1594.
Ojeda MGV, Gámiz-Fortis SR, Romero-Jiménez, E., Rosa-Cánovas, J.J., Yeste, P.,
Castro-Díez, Y., Esteban-Parra, M.J. Projected changes in the Iberian Peninsula
drought characteristics, Science of The Total Environment (2021), ,Volume
757,2021,143702,ISSN 0048-9697,
https://doi.org/10.1016/j.scitotenv.2020.143702.
Ozden, M., Vardin, H., Simsek, M., and Karaaslan, M. Effects of rootstocks and irrigation
levels on grape quality of Vitis vinifera L. cv. Shiraz. Afr. J. Biotechnol., 2010.
9:3801–380
Palmer W. «Meteorological Drought». Research paper no.45, U.S. Department of
Commerce Weather Bureau, febrero de 1965 (58 páginas). Available in National
Climatic Data Center de NOAA: http://www.ncdc.noaa.gov/temp-and-
precip/drought/docs/palmer.pdf
Paparrizos S, Maris F, Weiler M, Matzarakis A. Analysis and mapping of present and
future drought conditions over Greek areas with different climate conditions.
Theoretical and Applied Climatology 2018; 131: 259-270.
Parry ML, Canziani OF, Palutikof JP, Van der Linden PJ,Hanson CE, editors. IPCC
impacts, adaptation and vulnerability contribution of Working Group II to the
Fourth Assessment Report of the IPCC. Cambridge University Press; 2007.Patz
JA, Campbell-Lendrum D, Holloway T, Foley JA. Impact of regional climate
change on human health. Nature 2005; 438: 310-317.
Pavlousek, P. Evaluation of drought tolerance of new grapevine rootstock hybrids. J.
Environ. Biol., 2011. 32:543–549.
Pedro-Monzonís, M., Solera, A., Ferrer, J., Estrela, T., and Paredes-Arquiola, J. A review
of water scarcity and drought indexes in water resources planning and
management. Journal of Hydrology, 2015. 527, 482–493.
https://doi.org/10.1016/j.jhydrol.2015.05.003
Pellegrino, A., E. Lebon, M. Voltz, and J. Wery. Relationships between plant and soil
water status in vine (Vitis vinifera L.). Plant Soil, 2004. 266:129–142.
Pereira, S.C.; Carvalho, D.; Rocha, A. Temperature and Precipitation Extremes over the
Iberian Peninsula under Climate Change Scenarios: A Review. Climate 2021, 9,
139. https://doi.org/ 10.3390/cli9090139
Pereira, Susana & Marta-Almeida, Martinho & Carvalho, Ana cristina. (2017). Heat
Waves and Cold Spells changes in Iberia for a future climate scenario.
International Journal of Climatology. 37. 10.1002/joc.5158.
Perez J, Menendez M, Mendez FJ, Losada IJ. Evaluating the performance of CMIP3 and
CMIP5 global climate models over the north-east Atlantic region. Climate
Dynamics 2014; 43: 2663-2680.
Perez, L. and Barreiro-Hurlé, J. Assessing the socio-economic impacts of drought in the
Ebro River Basin. Spanish Journal of Agricultural Research, 2009. 7. 269.
10.5424/sjar/2009072-418.
Petgen, M. Möglichkeiten und Grenzen der Reifesteuerung: Wie flexibel reagiert die
Rebe? Das Deutsche Weinmagazin, 2007. 7/8, 42–47

230
Pieri P. Changement climatique et culture de la vigne: l’essentiel des impacts. In:
Changement Climatique, Agriculture et Forêt en France : Simulations d’Impacts
sur les Principales Espèces. Le Livre Vert du projet CLIMATOR (2007-2010),
ANR and INRA (eds), 2010. Ademe Editions, pp. 213-223.
Planton S, Deque M, Chauvin F, Terray L. Expected impacts of climate change on
extreme climate events. Comptes Rendus Geoscience 2008; 340: 564-574.
Pons M, Lopez-Moreno JI, Rosas-Casals M, Jover E. The vulnerability of Pyrenean ski
resorts to climate-induced changes in the snowpack. Climatic Change 2015; 131:
591-605.
Porter JR, Semenov MA. Crop responses to climatic variation. Philosophical
Transactions of the Royal Society B-Biological Sciences 2005; 360: 2021-2035.
PricewaterhouseCoopers 2009. Efectos del cambio climático sobre la industria
vitivinícola de la Argentina y Chile
R Development Core Team (2010): R: a language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria.
http://www.R-project.org. ISBN: 3-900051-07-0. Accessed 6 July 2011
Raddatz TJ, Reick CH, Knorr W, Kattge J, Roeckner E, Schnur R, et al. Will the tropical
land biosphere dominate the climate-carbon cycle feedback during the twenty-
first century? Climate Dynamics 2007; 29: 565-574.
Ramos MC, Jones GV, Yuste J. Variability of Tempranillo grape quality within the Ribera
del Duero do (Spain) and relationships with climatic characteristics. Journal of
Viticulture and Enology 2017; N 12/1
Ramos, M.C. Projection of phenology response to climate change in rainfed vineyards
in north-east Spain. Agric. For. Meteorol. 2017, 247, 104–115
Ramos MC, Jones GV, Martinez-Casasnovas JA. Structure and trends in climate
parameters affecting winegrape production in northeast Spain. Climate Research
2008; 38: 1-15.
Ramos MC, Martinez de Toda F. Variability in the potential effects of climate change on
phenology and on grape composition of Tempranillo in three zones of the Rioja
DOCa (Spain). European Journal of Agronomy 2020; 115
Ramos MC, Martinez de Toda F. Interannual and spatial variability of grape composition
in the Rioja DOCa show better resilience of cv. Graciano than cv. Tempranillo
under a warming scenario. Oeno One 2021; 55: 85-100.
Rebetez M, Mayer H, Dupont O, Schindler D, Gartner K, Kropp JP, et al. Heat and
drought 2003 in Europe: a climate synthesis. Annals of Forest Science 2006; 63:
569-577.
Renée Mozell, M. and Thach, E. The impact of climate change on the global wine
industry: Challenges & Solutions. Wine Economics and Policy.2014. 3.
10.1016/j.wep.2014.08.001
Resco P, Iglesias A, Bardaji I, Sotes V. Exploring adaptation choices for grapevine
regions in Spain. Regional Environmental Change 2016; 16: 979-993.
Resco de Dios, J.H., Cunill Camprubí, A., Thapa, P., Martínez del Castillo, E., Martínez
de Aragón, J., Bonet, J.A., Balaguer-Romano, R., Díaz-Sierra, R., Yebra, M.,
Boer, M.M., Climate change induced declines in fuel moisture may turn currently
fire-free Pyrenean mountain forests into fire-prone ecosystems, Science of The
Total Environment, Volume 797, 2021, 149104, ISSN 0048-9697,
https://doi.org/10.1016/j.scitotenv.2021.149104.
Revuelto, J., Gómez, D., Alonso-González, E. et al. Intermediate snowpack melt-out
dates guarantee the highest seasonal grasslands greening in the Pyrenees. Sci
Rep, 2022, 12, 18328. https://doi.org/10.1038/s41598-022-22391-x
https://www.sciencedirect.com/science/article/pii/S0048969721041760
Ribalaygua J, Gaitan E, Portoles J, Monjo R. Climatic change on the Gulf of Fonseca
(Central America) using two-step statistical downscaling of CMIP5 model outputs.
Theoretical and Applied Climatology 2018; 132: 867-883.

231
Ribalaygua J, Rosa Pino M, Portoles J, Roldan E, Gaitan E, Chinarro D, et al. Climate
change scenarios for temperature and precipitation in Aragon (Spain). Science
of the Total Environment 2013a; 463: 1015-1030.
Ribalaygua J, Torres L, Portoles J, Monjo R, Gaitan E, Pino MR. Description and
validation of a two-step analogue/regression downscaling method. Theoretical
and Applied Climatology 2013b; 114: 253-269.
Riou CH, Becker N, Sotes Ruiz V, Gomez-Miguel V, Carbonneau A, Panagiotou M, Calo
A, Costacurta A, Castro de R, Pinto A, Lopes C, Carneiro L, Climaco P. Le
déterminisme climatique de la maturation du raisin: application au zonage de la
teneur em sucre dans la communauté européenne. Office des Publications
Officielles des Communautés Européennes, Luxembourg, 1994; 322 pp
Robine JM, Cheung SLK, Le Roy S, Van Oyen H, Griffiths C, Michel JP, et al. Death toll
exceeded 70,000 in Europe during the summer of 2003. Comptes Rendus
Biologies 2008; 331: 171-U5.
Roderick, M. L., P. Greve, and G. D. Farquhar. 2015. "On the Assessment of Aridity with
Changes in Atmospheric Co2." Water Resources Research 51, no. 7 (Jul): 5450-
5463. http://dx.doi.org/10.1002/2015wr017031.
Rodriguez R, Navarro X, Casas MC, Ribalaygua J, Russo B, Pouget L, et al. Influence
of climate change on IDF curves for the metropolitan area of Barcelona (Spain).
International Journal of Climatology 2014; 34: 643-654.
Roldan E, Gomez M, Pino MR, Diaz J. The impact of extremely high temperatures on
mortality and mortality cost. International Journal of Environmental Health
Research 2015; 25: 277-287.
Roldan E, Gomez M, Pino MR, Portoles J, Linares C, Diaz J. The effect of climate-
change-related heat waves on mortality in Spain: uncertainties in health on a local
scale. Stochastic Environmental Research and Risk Assessment 2016; 30: 831-
839.
Romero, P., J. M. Navarro, J. Perez-Perez, F. Garcia-Sanchez, A. Gomez-Gomez, I.
Porras, et al. 2006. Deficit irrigation and rootstock: their effects on water relations,
vegetative development, yield, fruit quality and mineral nutrition of Clemenules
mandarin. Tree Physiol. 26:1537–1548.
Royé, D., Codesido, R., Tobías, A. and Taracido, M. Heat wave intensity and daily
mortality in four of the largest cities of Spain, Environmental Researc, 2020;
Volume 182, 109027, ISSN 0013-9351
https://doi.org/10.1016/j.envres.2019.109027.
Royé, D., Sera, F. Tobías, A., Lowe, R., Gasparrini, A., Pascal, M., Donato, F. Nunes,
B., Teixeira, J.P. Effects of Hot Nights on Mortality in Southern Europe.
Epidemiology: July 2021 - Volume 32 - Issue 4 - p 487-498 doi:
10.1097/EDE.0000000000001359
Ruml, M., A. Vukovic, M. Vujadinovic, V. Djurdjevic, Z. Rankovic-Vasic, Z. Atanackovic,
et al. 2012. On the use of regional climate models: implications of climate change
for viticulture in Serbia. Agric. For. Meteorol. 158:53–62.
Russo S, Sillmann J, Sterl A. Humid heat waves at different warming levels. Scientific
Reports 2017; 7.
Russo, S., Dosio, A., Graversen, R. G., Sillmann, J., Carrao, H., Dunbar, M. B.,
Singleton, A., Montagna, P., Barbola, P., and Vogt, J. V. Magnitude of extreme
heat waves in present climate and their projection in a warming world, J.
Geophys. Res. Atmos.,2014, 119, 12,500– 12,512, doi:10.1002/2014JD022098.
Saeed F, Almazroui M, Islam N, Khan MS. Intensification of future heat waves in
Pakistan: a study using CORDEX regional climate models ensemble. Natural
Hazards 2017; 87: 1635-1647.
Santiago JM, Munoz-Mas R, Solana-Gutierrez J, de Jalon DG, Alonso C, Martinez-Capel
F, et al. Waning habitats due to climate change: the effects of changes in
streamflow and temperature at the rear edge of the distribution of a cold-water
fish. Hydrology and Earth System Sciences 2017; 21.

232
Santos JA, Malheiro AC, Karremann MK, Pinto JG. Statistical modelling of grapevine
yield in the Port Wine region under present and future climate conditions.
International Journal of Biometeorology 2011; 55: 119-131.
Santos JA, Malheiro AC, Pinto JG, Jones GV. Macroclimate and viticultural zoning in
Europe: observed trends and atmospheric forcing. Climate Research 2012; 51:
89-103.
Santos, M., Fonseca, A., Helder, F., Jones, G..; Santos, João A. Bioclimatic conditions
of the Portuguese wine denominations of origin under changing climates.
International Journal of Climatology, 2019, joc.6248–. doi:10.1002/joc.6248
Santos M, Fonseca A, Fraga H, Jones GV, Santos JA. Bioclimatic conditions of the
Portuguese wine denominations of origin under changing climates. International
Journal of Climatology 2020a; 40: 927-941.
Santos, J., Fraga, H., Malheiro, A., Moutinho Pereira, J., L-T, Dinis, Correia, C.,
Moriondo, M., Leolini, L., Dibari, C., Costafreda-Aumedes, S., Kartschall, T.,
Menz, C., Molitor, D., Junk, J., Beyer, M., Schultz, H. A Review of the Potential
Climate Change Impacts and Adaptation Options for European Viticulture.
Applied Sciences.2020b 10. 10.3390/app10093092. INTRO
Sanz, J.J., Potti, J., Moreno, J., Merino, S. and Frías, O. Climate change and fitness
components of a migratory bird breeding in the Mediterranean region. Global
Change Biology, 2003, 9: 461-472.
https://doi.org/10.1046/j.1365-2486.2003.00575.x
Schoetter R, Cattiaux J, Douville H. Changes of western European heat wave
characteristics projected by the CMIP5 ensemble. Climate Dynamics 2015; 45:
1601-1616.
Schroter D, Cramer W, Leemans R, Prentice IC, Araujo MB, Arnell NW, et al. Ecosystem
service supply and vulnerability to global change in Europe. Science 2005; 310.
Schultz HR, Stoll M. Some critical issues in environmental physiology of grapevines:
future challenges and current limitations. Australian Journal of Grape and Wine
Research 2010; 16: 4-24.
Schultz HR. Climate change and viticulture: a European perspective on climatology,
carbon dioxide and UV-B effects. Aust J Grape Wine Res 2000; 6:2–12
Seager, R., Hoerling, M., Schubert, S., Wang, H., Lyon, B., Kumar, A., Henderson, N.
(2015). Causes of the 2011–14 California drought. Journalof Climate, 28(18),
6997–7024. https://doi.org/10.1175/JCLI-D-14-00860.1
Semenov MA, Stratonovitch P. Use of multi-model ensembles from global climate
models for assessment of climate change impacts. Climate Research 2010; 41:
1-14.
Sepulcre-Canto, G., Horion, S., Singleton, A., Carrao, H., & Vogt, J. Development of a
Combined Drought Indicator to detect agricultural drought in Europe. Natural
Hazards and Earth System Sciences, () 12(11), 3519–3531.
https://doi.org/10.5194/nhess-12-3519-2012
Seubert S, Fernandez-Montes S, Philipp A, Hertig E, Jacobeit J, Vogt G, et al.
Mediterranean climate extremes in synoptic downscaling assessments.
Theoretical and Applied Climatology 2014; 117: 257-275.
Sgubin, G.; Swingedouw, D.; Dayon, G.; Garcia de Cortazar-Atauri, I.; Ollat, N.; Pagé,
C.; Van Leeuwen, C. The risk of tardive frost damage in French vineyards in a
changing climate. Agric. For. Meteorol. 2018, 250–251, 226–242.
Sheffield J, Wood EF, Roderick ML. Little change in global drought over the past 60
years. Nature 2012; 491: 435-+.
Sheffield J, Wood EF. Projected changes in drought occurrence under future global
warming from multi-model, multi-scenario, IPCC AR4 simulations. Climate
Dynamics 2008; 31: 79-105.
Sigler, J. In den Zeiten des Klimawandels: Von der Süßreserve zur Sauerreserve? Der
Badische Winzer, 2008. 33, 21–25.

233
Sillmann J, Donat MG, Fyfe JC, Zwiers FW. Observed and simulated temperature
extremes during the recent warming hiatus. Environmental Research Letters
2014; 9.
Skaugen T, Astrup M, Roald LA, Forland E. Scenarios of extreme daily precipitation for
Norway under climate change. Nordic Hydrology 2004; 35: 1-13.
Smith M, Allen R, Pereira L. Revised FAO methodology for crop-water requirements.
Management of nutrients and water in rainfed arid and semi-arid areas.
Proceedings of a consultants meeting. International Atomic Energy Agency
(IAEA). 1998; IAEA-TECDOC--1026. Ref. number: 29062763.
Smith TT, Zaitchik BF, Gohlke JM. Heat waves in the United States: definitions, patterns
and trends. Climatic Change 2013; 118: 811-825.
Spain Government. Mapa de las Denominaciones de Origen Protegidas de Vinos de
España (Map of Protected Designations of Origin of Wines of Spain). 2014.
(https://www.mapa.gob.es/es/cartografia-y-
sig/publicaciones/alimentacion/mapa_dop_vinos.aspx)

Soares PMM, Cardoso RM, Miranda PMA, de Medeiros J, Belo-Pereira M, Espirito-

Santo F. WRF high resolution dynamical downscaling of ERA-Interim for
Portugal. Climate Dynamics 2012; 39: 2497-2522.
Sonmez FK, Komuscu AU, Erkan A, Turgu E. An analysis of spatial and temporal
dimension of drought vulnerability in Turkey using the standardized precipitation
index. Natural Hazards 2005; 35: 243-264.
Sotés Ruiz, V.; Gómez-Miguel, V.; Almorox, J.; Vidal Ragout, J. y Vida Navarro, L.
“Clima, zonificación; tipicidad del vino en España“. In J. Tonietto, V. Sotés Ruiz,
y V. GómezMiguel (Eds.), Clima, zonificación; tipicidad del vino en regiones
vitivinícolas Iberoamericanas (Vols. ISBN 978-84-15413-10-3), 2012. Madrid:
CYTED
Spellman G. Wine, weather and climate, 1999. Weather 54:230–239
Spinoni J, Naumann G, Vogt JV, Barbosa P. The biggest drought events in Europe from
1950 to 2012. Journal of Hydrology-Regional Studies 2015; 3: 509-524.
Stagge JH, Kohn I, Tallaksen LM, Stahl K. Modeling drought impact occurrence based
on meteorological drought indices in Europe. Journal of Hydrology 2015; 530: 37-
50.
Stahl, Kerstin & Blauhut, Veit & Kohn, Irene & Acácio, V. & Assimacopoulos, D. & Bifulco,
Carlo & De Stefano, Lucia & Dias, Susana & Eilertz, D. & Frielingsdorf, B. &
Hegdahl, Trine & Kampragou, E. & Kourentzis, V. & Melsen, L.A. & Van Lanen,
Henny & Van Loon, Anne & Massarutto, Antonio & Musolino, Dario & Paoli, L. &
Urquijo Reguera, Julia. (2012). A European Drought Impact Report Inventory
(EDII): Design and Test for Selected Recent Droughts in Europe.
Stanke C, Kerac M, Prudhomme C, Medlock J, Murray V. Health effects of drought: a
systematic review of the evidence. PLoS Curr. 2013 Jun 5;5. pii:
ecurrents.dis.7a2cee9e980f91ad7697b570bcc4b004.
Stegehuis AI, Vautard R, Ciais P, Teuling AJ, Miralles DG, Wild M. An observation-
constrained multi-physics WRF ensemble for simulating European mega heat
waves. Geoscientific Model Development 2015; 8: 2285-2298.
Stephenson, D. B. (2008). Definition, diagnosis, and origin of extreme weather and
climate events. In H. F. Diaz & R. J. Murnane (Eds.), Climate Extremes and
Society (pp. 11–23). Cambridge, England: Cambridge University Press.
Retrieved from http://ebooks.cambridge .org/ref/id/CBO9780511535840A011
Steul KS, Latasch L, Jung HG, Heudorf U. Health Impact of the Heatwave of 2015:
Hospital Admissions in Frankfurt/Main, Germany. Gesundheitswesen 2018; 80:
353-359.
Stock M, Gerstengarbe FW, Kartschall T, Werner PC. Reliability of climate change
impact assessments for viticulture. 7th International Symposium on Grapevine
Physiology and Biotechnology, Davis, CA, 2004, pp. 29-39.

234
Subrahmanyam, V.P., and Subramaniam, A.R.. Application of water balance concepts
for a climatic study of droughts in South India.1964 I.J.M.G., 15, 393-402.
Tate AB. Global warming’s impact on wine. J Wine Res 2001; 12:95–109
Taylor KE, Stouffer RJ, Meehl GA. AN OVERVIEW OF CMIP5 AND THE EXPERIMENT
DESIGN. Bulletin of the American Meteorological Society 2012; 93: 485-498.
Teixeira, A.H. de C.; Tonietto, J.; Pereira, G.A.; Angelloti, F. Delimitação da aptidão
agroclimática para videira sob irrigação no Nordeste brasileiro. Revista Brasileira
de Engenharia Agrícola e Ambiental, v.16, p.399-407, 2012. DOI:
10.1590/S1415-43662012000400010.
Thornthwaite C. An approach toward a rational classification of climate. Geogr.
Rev.1948; 38: 55–94.
Tobías, A., Armstrong, B., Gasparrini, A. et al. Effects of high summer temperatures on
mortality in 50 Spanish cities. Environ Health 2014; 13, 48
https://doi.org/10.1186/1476-069X-13-48
Tomozeiu R, Cacciamani C, Pavan V, Morgillo A, Busuioc A. Climate change scenarios
for surface temperature in Emilia-Romagna (Italy) obtained using statistical
downscaling models. Theoretical and Applied Climatology 2007; 90: 25-47.
Tonietto J, Carbonneau A. A multicriteria climatic classification system for grape-growing
regions worldwide. Agricultural and Forest Meteorology 2004; 124: 81-97.
Tonietto, J. Les Macroclimats Viticoles Mondiaux et l’Influence du Mésoclimat sur la
Typicité de la Syrah et du Muscat de Hambourg dans le Sud de la France:
Méthodologie de Caráctérisation. Thèse Doctorat. Ecole Nationale Supérieure
Agronomique, Montpellier 1999.
Tonietto, J., Sotés, V., Zanus, M., Montes, C., Uliarte, E., Antelo, L., Carbonneau, A.
L'effet du climat viticole sur la typicité des vins rouges- Caractérisation au niveau
des régions viticoles Ibéro-Americaines. In VIII International Terroirs Congress
(Vol. 3, p. 17-22), 2010. Soave, Italy: CRA-VIT Centro di Ricerca per la Viticoltura.
Tonietto J, Carbonneau A. Análise mundial do clima das regiões vit´ıcolase de sua
influencia sobre a tipicidade dos vinhos: a posição da viticultura brasileira
comparada a 100 regiões em 30 paıses. Anais do Congresso Brasileiro de
Viticultura e Enologia, Embrapa Uva e Vinho. Bento Gonçalves, 1999; 75–90.
Torres C, Jordà G, de Vílchez P, Vaquer-Sunyer R, Rita J, Canals V, Cladera A,
Escalona JM, Miranda MÁ. Climate change and their impacts in the Balearic
Islands: a guide for policy design in Mediterranean regions. Reg Environ Change.
2021;21(4):107. doi: 10.1007/s10113-021-01810-1. Epub 2021 Oct 23. PMID:
34720740; PMCID: PMC8536903.
Trenberth, K.E., Dai, A., Van der Schrier, G., Jones, P.D., Barichivich, J., Briffa, K.R.,
Sheffield, J. Global warming and changes in drought. Nature Climate Change,
2014. 4, 17-22. Doi:10.1038/nclimate2067
Tripathi S, Srinivas VV, Nanjundiah RS. Dowinscaling of precipitation for climate change
scenarios: A support vector machine approach. Journal of Hydrology 2006; 330:
621-640.
Tsakiris, G., and Vangelis, H. 2005. Establishing a drought index incorporating
evapotranspiration. European Water 2005; 9 (10): 3–11.
Tselepidaki I, Zarifis B, Asimakopoulos DN. LOW PRECIPITATION OVER GREECE
DURING 1989-1990. Theoretical and Applied Climatology 1992; 46: 115-121.
Turco, M., Quintana Seguí, P., Llasat, M., Herrera García, S. and Gutiérrez, J. Testing
MOS precipitation downscaling for ENSEMBLES regional climate models over
Spain. Journal of Geophysical Research, 2011. 116. 10.1029/2011JD016166.
Turco M, Marcos R, Quintana-Segui P, Llasat MC. Testing instrumental and downscaled
reanalysis time series for temperature trends in NE of Spain in the last century.
Regional Environmental Change 2014; 14: 1811-1823.
Uppala SM, Kallberg PW, Simmons AJ, Andrae U, Bechtold VD, Fiorino M, et al. The
ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society 2005;
131: 2961-3012.

235
Van Leeuwen, C., & Darriet, P. The Impact of Climate Change on Viticulture and Wine
Quality. Journal of Wine Economics, 2016, 11(1), 150-167.
doi:10.1017/jwe.2015.21
Van Leeuwen C, Friant P, Chone X, Tregoat O, Koundouras S, Dubourdieu D. Influence
of climate, soil, and cultivar on terroir. American Journal of Enology and Viticulture
2004; 55: 207-217.
Vanderlinden K, Giraldez JV, Van Meirvenne M. Assessing reference evapotranspiration
by the Hargreaves method in southern Spain. Journal of Irrigation and Drainage
Engineering 2004; 130: 184-191.
Van der Linden P, Mitchell JFB (eds) 2009: ENSEMBLES: Climate change and its
impacts: summary of research and results from the ENSEMBLES project. Met
Office Hadley Centre, UK, 160 pp.http://ensembles-
eu.metoffice.com/docs/Ensembles_final_report_Nov09.pdf. (see p 68).
Accessed 10 Feb 2012
Van-Rooy, M.P. 1965. A rainfall anomaly index (RAI) independent of time and space.
Notos, 14: 43–48.

Vautard R, Gobiet A, Jacob D, Belda M, Colette A, Deque M, et al. The simulation of

European heat waves from an ensemble of regional climate models within the
EURO-CORDEX project. Climate Dynamics 2013; 41: 2555-2575.
Velasco, M, Russo, B.; Monjo, R et al. Increased Urban Resilience to Climate Change.
Key Outputs from the RESCCUE Project. Sustainability, 2020, 12(23), 9881
doi:10.3390/su12239881
Vicente-Serrano SM, McVicar TR, Miralles DG, Yang YT, Tomas-Burguera M.
Unraveling the influence of atmospheric evaporative demand on drought and its
response to climate change. Wiley Interdisciplinary Reviews-Climate Change
2020; 11.
Vicente-Serrano, M., Beguería, S., Reig, F., Latorre, B., Peña-Gallardo, M., Luna, M.Y.,
Morata, A. and González-Hidalgo, J. C. A High Resolution Dataset of Drought
Indices for Spain. Data 2017, 2, 22; doi:10.3390/data2030022
Vicente-Serrano SM, Dominguez-Castro F, McVicar TR, Tomas-Burguera M, Pena-
Gallardo M, Noguera I, et al. Global characterization of hydrological and
meteorological droughts under future climate change: The importance of
timescales, vegetation-CO2 feedbacks and changes to distribution functions.
International Journal of Climatology. 2019. https://doi.org/10.1002/joc.6350
Vicente-Serrano SM, Begueria S. Comment on 'Candidate distributions for climatological
drought indices (SPI and SPEI)' by James H. Stagge et al. International Journal
of Climatology 2016; 36: 2120-2131.
Vicente-Serrano SM, Van der Schrier G, Begueria S, Azorin-Molina C, Lopez-Moreno JI.
Contribution of precipitation and reference evapotranspiration to drought indices
under different climates. Journal of Hydrology 2015; 526: 42-54.
Vicente-Serrano, S.M. Spatial and temporal evolution of precipitation droughts in Spain
in the last century in Adverse Weather in Spain; Martínez, C.C.-L.; Rodríguez,
F.V., Eds.; WCRP Spanish Committee: Madrid, Spain, 2013; 283–296.
Vicente-Serrano SM, Begueria S, Lopez-Moreno JI. A Multiscalar Drought Index
Sensitive to Global Warming: The Standardized Precipitation Evapotranspiration
Index. Journal of Climate 2010a; 23: 1696-1718.
Vicente-Serrano SM, Begueria S, Lopez-Moreno JI, Angulo M, El Kenawy A. A New
Global 0.5 degrees Gridded Dataset (1901-2006) of a Multiscalar Drought Index:
Comparison with Current Drought Index Datasets Based on the Palmer Drought
Severity Index. Journal of Hydrometeorology 2010b; 11: 1033-1043.
Vicente-Serrano SM, Gonzalez-Hidalgo JC, de Luis M, Raventos J. Drought patterns in
the Mediterranean area: the Valencia region (eastern Spain). Climate Research
2004; 26: 5-15.

236
Vicente-Serrano SM, Lasanta T, Gracia C. Aridification determines changes in forest
growth in Pinus halepensis forests under semiarid Mediterranean climate
conditions. Agricultural and Forest Meteorology 2010c; 150: 614-628.
Vicente-Serrano SM, Lopez-Moreno JI. Hydrological response to different time scales of
climatological drought: an evaluation of the Standardized Precipitation Index in a
mountainous Mediterranean basin. Hydrology and Earth System Sciences 2005;
9: 523-533.
Vicente-Serrano, S.M.; Beguería, S. Comment on “Candidate Distributions for
Climatological Drought Indices (SPI and SPEI)” by James H. Stagge et al. Int. J.
Clim. 2016, 36, 2120–2131.
Vidaller, I., Revuelto, J., Izagirre, E., Rojas-Heredia, F., Alonso-González, E., Gascoin,
S., et al. Toward an ice-free mountain range: Demise of Pyrenean glaciers during
2011–2020. Geophysical Research Letters, 2021, 48, e2021GL094339.
https://doi.org/10.1029/2021GL094339
Voldoire A, Sanchez-Gomez E, Melia DSY, Decharme B, Cassou C, Senesi S, et al. The
CNRM-CM5.1 global climate model: description and basic evaluation. Climate
Dynamics 2013; 40: 2091-2121.
Vuković, A., Vujadinović, M., Djurdjević, V., Ranković-Vasić, Z., Marković, N.,
Atanacković, Z., . . . Petrović, N. Appliance of climate projections for climate
change study in serbian vineyard regions. In VIII International Terroirs Congress,
2010. (Vol. 3, p. 36-41). Soave, Italy: CRAVIT Centro di Ricerca per la Viticoltura.
Wang B, Zhou TJ, Yu YQ. A View of Earth System Model Development. Acta
Meteorologica Sinica 2009; 23: 1-17.
Wang, Xueqiu & Wang, Hua & Li, Hua. (2020). The influence of recent climate variability
on viticultural zoning and variety regionalization of Vitis vinifera in China. OENO
One. 54. 523-541. 10.20870/oeno-one.2020.54.3.2971.
Watanabe S, Hajima T, SudNagashima T, Takemura T, Okajima H, et al. MIROC-ESM
2010: model description and basic results of CMIP5-20c3m experiments.
Geoscientific Model Development 2011; 4: 845-872.
Webb LB, Whetton PH, Barlow EWR. Modelling the relationship between climate,
winegrape price and winegrape quality in Australia. Climate Research 2008; 36:
89-98.
Weigel AP, Knutti R, Liniger MA, Appenzeller C. Risks of Model Weighting in Multimodel
Climate Projections. Journal of Climate 2010; 23: 4175-4191.
Westerling AL, Hidalgo HG, Swetnam TW . Warming and earlier spring increase Western
U.S. forest wildfire activity. Science, 2006. 313, 940–943.
Westphalen, S.L. and Maluf, J.R.T. Caracterização das áreas bioclimáticas para o cultivo
de Vitis vinifera L.: regiões da Serra do Nordeste e Planalto do Estado do Rio
Grande do Sul. Brasília: Embrapa Comunicação para Transferência de
Tecnologia, 2000. 99p.
White MA, Diffenbaugh NS, Jones GV, Pal JS, Giorgi F. Extreme heat reduces and shifts
United States premium wine production in the 21st century. Proceedings of the
National Academy of Sciences of the United States of America 2006; 103: 11217-
11222.
Wild, M., Folini, D., Schaer, C., Loeb, N., Dutton, E.G., Koning-Langlo, G. 2013. The
global energy balance from a surface perspective. Climate Dynamics 40 (11),
3107-3134. Doi:10.1007/s00382-012-1569-8.
Wilhite D. Drought as a natural hazard: concepts and definitions. In: Wilhite DA, ed.
Droughts: Global Assessment. London: Routledge; 2000; 3–18.
Wilhite, D.A., Glantz, M.H. Understanding the drought phenomenon: The role of
definitions. Water International 10(3): 111-120, 1985
Wilhite, D. Drought, Encyclopaedia of Earth System Science. San Diego,. CA: Academic
Press.1992; 2: 81–92.

237
Willett KM, Dunn RJH, Thorne PW, Bell S, de Podesta M, Parker DE, et al. HadISDH
land surface multi-variable humidity and temperature record for climate
monitoring. Climate of the Past 2014; 10: 1983-2006.
Winkler AJ. General Viticulture. University of California Press, CA 1974
WMO (World Meteorological Organization), 2010. No. 1055. ISBN 978-92-63-11055-8.
Switzerland www.wmo.int
WMO (World Meteorological Organization), 2012. In: Svoboda, M., Hayes, M. and Wood,
D.(Eds.) Standardized Precipitation Index User Guide. Geneva: WMO
WMO (World Meteorological Organization), 2015. Guidelines on the Definition and
Monitoring of Extreme Weather and Climate Events.
WMO (World Meteorological Organization), 2017. Statement on the State of the Global
Climate in 2016.Nº 1189.2017. ISBN 978-92-63-11189-0
Xiao-Ge X, Tong-Wen W, Jie Z. Introduction of CMIP5 experiments carried out with the
climate system models of Beijing Climate Center. Adv. Clim. Chang. Res. 2013;
4:41–49. https://doi.org/10.3724/SP.J.1248.2013.041
Yang YT, Roderick ML, Zhang SL, McVicar TR, Donohue RJ. Hydrologic implications of
vegetation response to elevated CO2 in climate projections. Nature Climate
Change 2019; 9: 44-+.
Yang YT, Zhang SL, McVicar TR, Beck HE, Zhang YQ, Liu B. Disconnection Between
Trends of Atmospheric Drying and Continental Runoff. Water Resources
Research 2018; 54: 4700-4713.
Yu S, Eder B, Dennis R, Chu S-H, Schwartz SE. New unbiased symmetric metrics for
evaluation of air quality models. Atmospheric Science Letters 2006; 7.
Yukimoto S, Yoshimura H, Hosaka M, Sakami T, Tsujino H, Hirabara M, Tanaka T,
Deushi M, Obata A, Nakano H, Adachi Y, Shindo E, Yabu S, Ose T, Kitoh A.
2011. Meteorological research institute- earth system model v1 (MRI-ESM1)—
model description. Technical Report of MRI. vol 64.
Zargar A., Rehan Sadiq, Bahman Naser, and Faisal I. Khan (2011) A review of drought
indices. Environ. Rev. 19: 333–349. doi:10.1139/A11-013
Zambrano-Bigiarini, M., Majone, B., Bellin, A., Bovolo, C., Blenkinsop, S. and Fowler, H.
Hydrological Impacts of Climate Change on the Ebro River Basin, 2010.
10.1007/698_2010_85.
Zittis G, Hadjinicolaou P, Fnais M, Lelieveld J. Projected changes in heat wave
characteristics in the eastern Mediterranean and the Middle East. Regional
Environmental Change 2016; 16: 1863-1876.
Zha Q, Xi X, He Y, Jiang A. Comprehensive evaluation of heat resistance in 68 Vitis
germplasm resources. Vitis 2018; 57: 75-81.
Zorita, Eduardo & Von Storch, Hans. (1999). The Analog Method as a Simple Statistical
Downscaling Technique: Comparison With More Complicated Methods. J.
Climate. 12. 2474-2489. 10.1175/1520-
0442(1999)012<2474:TAMAAS>2.0.CO;2.

238
ANEXO DE PUBLICACIONES

239
 Science of the Total Environment 650 (2019) 2778–2795

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Projection of temperatures and heat and cold waves for Aragón (Spain)
using a two-step statistical downscaling of CMIP5 model outputs
Emma Gaitán a,b, Robert Monjo b, Javier Pórtoles b, Mª. Rosa Pino-Otín a,⁎
a
Universidad San Jorge, Villanueva de Gállego, 50830, Zaragoza, Spain
b
Fundación para la Investigación del Clima, C/Tremps 11, 28040 Madrid, Spain

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Future temperature and heat and cold

wave scenarios are obtained for Aragón,
Spain.
• All scenarios are obtained using Earth
System Models from CMIP5 for the first
time.
• Intensity and duration of the cold waves
will remain stable during this century.
• Intensity and duration of the heat waves
will increase (3.6 °C, 7 days).
• The maximum temperature may
increase up to 7 °C in summer at the
end of the century.

a r t i c l e i n f o a b s t r a c t

Article history: Heat- and cold-wave scenarios and temperature scenarios during the 21st century were obtained for Aragón
Received 6 July 2018 (Spain), using, for the first time, nine Earth System Models (ESM) and two Representative Concentration Pathway
Received in revised form 18 September 2018 (RCP) scenarios – RCP4.5 and RCP8.5 - belonging to the 5th Coupled Model Intercomparison Project (CMIP5).
Accepted 28 September 2018
Local climate heat-wave scenarios show an increase of its mean intensity close to 2 °C (reaching temperatures of
Available online 4 October 2018
up to 38.8 °C) and an average increase of the maximum intensity of 3.6 °C (temperature of up to 41.5 °C) with
Editor: Pavlos Kassomenos respect to a historic period (1971–2000) for the RCP8.5 scenario at the end of the century. The duration of
heat waves will increase by 7 days at the end of the century (total average duration of 12 days). The future
Keywords: intensity and duration of cold-wave episodes will remain stable.
Climate change Local climate change scenarios for daily maximum temperatures show a gradual increase throughout the 21st
Heat wave century. The greatest increases will occur during the summer at the end of the century, reaching values of up
Cold wave to 7 °C for the RCP 8.5 scenario. The minimum temperature increases show similar behaviours to the maximum
Extreme temperatures temperatures, but with less marked increases (3 °C and 5.6 °C for the RCP4.5 and RCP8.5 scenarios respectively in
Downscaling
summer at the end of the century).
Spain
The highest temperatures and the intensity of the heat waves will be especially intense in the Ebro Valley, the
most populated area. In addition, the Pyrenees will suffer the longest heat waves, especially at the end of the
century, and the greatest increases in maximum temperatures.
The downscaling of the CMIP5 models, offers accurate scenarios -both spatially and temporally- of extreme
temperatures and heat and cold waves, useful for decision-making for local adaptation to climate change but
also as a reference for other European regions.
© 2018 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Universidad San Jorge, Campus Universitario Villanueva de Gállego Autovía A-23 Zaragoza-Huesca, km. 510, 50830 Villanueva de Gállego, Zaragoza, Spain
E-mail address: [email protected] (M.ªR. Pino-Otín).

https://doi.org/10.1016/j.scitotenv.2018.09.352
0048-9697/© 2018 Elsevier B.V. All rights reserved.
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2779

1. Introduction regional climate model (RCM) of atmospheric boundary conditions is

nested within the GCM (Giorgi and Gutowski, 2015; Giorgi and
There is widespread agreement in the scientific literature that the Mearns, 1991;). Both methods have been widely used and both have ad-
most severe effects of global warming will be related not only to a vantages and disadvantages. The main advantages of statistical down-
change in the mean climate, but especially to an increase in the fre- scaling techniques are that they are computationally inexpensive, they
quency and intensity of extreme events, such as floods, droughts and provide local information and they allow quantifying the uncertainty
heat waves as is reflected in the conclusions provided by the Intergov- associated to the downscaling process and to the climatic models.
ernmental Panel on Climate Change (IPCC) and in independent studies Several studies have been carried out to explore present tempera-
(Cook et al., 2013). Heat waves (i.e, a period of consecutive days with ture extremes in Europe (Soares et al., 2012) and in the Mediterranean
extreme hot temperatures) are one form of extreme weather that is region (Kuglitsch et al., 2010; Seubert et al., 2014), and other studies
likely to increase in frequency, intensity and duration under the influ- have developed scenarios of temperature extremes in the same areas
ence of a changing climate during upcoming years (IPCC, 2013; Field (Estrella and Menzel, 2013; Hertig et al., 2010; Lavaysse et al., 2012).
et al., 2012). Extreme temperature events can impact many aspects of However, none of them have applied the downscaling CMIP5 models.
human life, such as mortality, health, comfort, agriculture and hydrol- Although many climate models have difficulties in properly repro-
ogy (Garcia-Herrera et al., 2005; Patz et al., 2005; Roldan et al., 2016). ducing climate extremes, such as heat-wave conditions (Stegehuis
For example, in 2003, an extraordinary “mega heat wave” occurred in et al., 2015), a few studies have reported on heat-wave scenarios in
Central Europe, which had important consequences on society. Europe (Fischer and Schar, 2010; Schoetter et al., 2015), or located in
One of the main problems when evaluating extreme episodes of in the Mediterranian region (Zittis et al., 2016), in France (Planton
temperature (that is, cold or heat waves) is that there are no universal et al., 2008), Italy (Tomozeiu et al., 2007) or Finland (Kim et al., 2018).
definitions of these phenomena. Some definitions use a fixed term Schoetter et al. (2015) and Kim et al. (2018) are one of the few studies
(5 days N30 °C or 3 days N35 °C in the case of a heat wave); some include that already use the models from the IPPC5. As a part of the European
humidity, spatial extent or accumulated heat ((Anderson and Bell, Coordinated Regional Downscaling Experiment (EUROCORDEX)
2009; Russo et al., 2017; Suparta and Yatim, 2017); some set their (Giorgi et al., 2009), the EUROCORDEX initiative provides regional cli-
threshold according to a specific percentile (Anderson and Bell, 2009; mate projections for Europe. (Vautard et al., 2013) used the
Kent et al., 2014; Smith et al., 2013); and some are based on combined EUROCORDEX project to simulate heat waves at the European regional
indexes of different meteorological variables, such as World Meteoro- scale, providing a downscaling of CMIP5 simulations. These European-
logical Organization member guidelines (WMO, 2010). scale studies only featured a few locations in Spain. Several downscaling
Although large-scale and small-scale processes involved in the de- studies have been specifically conducted on temperature in Spain
velopment of European heat waves, such as interactions between land (Brands et al., 2011b; Frias et al., 2005; Frias et al., 2010; Hervada-Sala
surface and atmosphere and the influence of large-scale atmospheric et al., 2000; Miro et al., 2016; Turco et al., 2014), but very few studies
circulation, are not fully understood (Vautard et al., 2013), climate have explored temperature extremes in Spain (Fernandez-Montes and
models could be a suitable tool for providing accurate heat-wave assess- Rodrigo, 2012; Fonseca et al., 2016) and they do it in past periods
ments for future adaptation strategies. (1925–2006 and 1986–2005, respectively).
Global climate models (GCMs) are fundamental tools for the study of Only a few studies have developed temperature scenarios in the Ar-
future climate, and they are able to provide data to estimate large-scale agón region (northeastern Spain) (Buerger et al., 2007; Goncalves et al.,
aspects of climate change, to drive regional climate models or to be used 2014; Ribalaygua et al., 2013a), but none of them developed
directly by impact models. GCM outputs are also used for projections of temperature-extreme or heat-wave scenarios. Only (Barrera-Escoda
temperature extremes around the world (Dosio, 2017; Gross et al., et al., 2014) developed scenarios of extreme temperatures in the Ebro
2017; Jeong et al., 2016; Saeed et al., 2017; Sillmann et al., 2014). basin. That study showed that the projected increase in the number of
Recently, efforts to reduce model uncertainty have led to a new gener- tropical nights and extreme temperatures could have a negative effect
ation of global climate models called Earth System Models (ESM) that in- on human health and comfort conditions. Other studies have also
tegrate the individual parts of the climate system (atmosphere, ocean, analysed the possible impacts of climate change in this region on
land and sea ice) and the exchange of energy and mass between them human health (Roldan et al., 2016). Both studies used the climate
(Knutti and Sedlacek, 2013). They also include chemical processes, land models associated to the 4th assessment report of the IPCC.
use, plant and ocean ecology and an interactive carbon cycle, which en- Aragón is characterised by a highly complex topography that causes
ables integration of biochemical processes into the models (Heavens large climate gradients. Downscaling techniques were required to cap-
et al., 2013). These models are the basis of the fifth phase of the Coupled ture these orographic features, allowing for the areas that were most
Model Intercomparison Project (CMIP5) (Taylor et al., 2012), constituting vulnerable to extreme changes to be identified. Statistical downscaling
the most current set of coordinated climate model experiments (Brands is particularly recommended for areas with complex topography
et al., 2013; Carvalho et al., 2017; Chen et al., 2016; Perez et al., 2014). (Kattenberg and Amer Meteorol SOC, 1996). Nevertheless, downscaling
The CMIP5 models and their predecessors have become an important of CMIP5 simulations has not been used to date, and heat-wave scenar-
resource for climate scientists and others investigating possible future im- ios cannot be found in this sensible area of Spain.
pacts. However, the spatial resolution of these models remains too low for Therefore, the objectives of this study were:
many impact assessments at a local scale, on which policy and manage-
ment decisions need to be made. These models are run at coarse spatial • Generate local heat- and cold-wave scenarios of the 21st century to
resolutions, which is also a major limitation in projecting extreme events downscale GCMs from CMIP5 using a statistical methodology in the
because projections of climate extremes and simulations of heat-wave area of Aragón, Spain.
characteristics on regional scales are very sensitive to spatial resolution • In addition, generate new climate scenarios using CMIP5 models for
and the interpretation of gridded outputs (Chen et al., 2016). Aragón to simulate the future daily maximum and minimum temper-
The deficiency in scale is addressed through downscaling techniques atures in order to analyse the differences with respect to the scenarios
involving either statistical approaches or dynamic approaches. In the we generated in 2013 for the 21st century using the Special Report on
statistical approach, high-resolution predictands (surface variables) Emissions Scenarios, SRES (IPCC, 2000) previous to CMIP5.
are obtained by applying relationships that are identified from the ob-
served climate data to these predictands and to large-scale predictors
of GCM outputs (Asong et al., 2016; Benestad et al., 2007; Semenov Aragón, moreover, is a representative territory of different European
and Stratonovitch, 2010). In dynamical downscaling, a time-varying climates, from the low-land areas in the centre of Aragón (the Ebro
2780 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

River Basin) to the mountain regions in the north (the Pyrenees) which 2.1.2. Surface observation dataset (predictands)
makes it a good indicator of future European climate changes. Finally, an The observational dataset used in the present study consists of a
original two-step analogue/regression statistical downscaling method time series of the daily maximum and minimum temperatures distrib-
developed by the Climate Research Foundation was carried out. uted, quite homogeneously, throughout the Aragón territory (Fig. 1).
The information used in the present study was based on the most This dataset it is the same as the one used in a previous study
current data available. It was useful for identifying the areas that were (Ribalaygua et al., 2013a) in order to facilitate comparison of the results.
most vulnerable to extreme temperature changes in Aragón, and for Data were obtained from the extensive network of instrumental ob-
helping decision makers to design mitigation and adaptation strategies servatories owned by the Spanish Meteorological Agency (AEMET)
in response to territorial environmental impacts derived from climate (http://www.aemet.es). The stations are subject to strict data quality
change. control, which includes detecting and correcting multiple aberrant
points in the long-term climate time series, covering the gaps, assessing
2. Material and methods the homogeneity of the climate series and correcting any inhomogene-
ities found. This process was carried out by the Aragón Government
2.1. Study area and datasets (López et al., 2007) To guarantee the quality of the data, stations with
a large number of data gaps or without at least 15 years of daily records
2.1.1. Study area were discarded. Second, we analysed the dataset for the presence of ex-
The present study was carried out in the region of Aragón (Fig. 1), treme data points that could be considered suspect observations or
which is located in the northeast part of the Iberian Peninsula in Spain transcription errors (we considered data points to be extreme if they
and which has an area of approximately 47,720 km2 (340 km length were separated from the monthly average by more than four times
and 240 km width). Because of its location, Aragón falls within the West- the value of the standard deviation). Third, we verified that the data
ern Mediterranean climate area, with cool winters and hot, dry summers. did not contain inhomogeneities that could introduce bias into any re-
However, the extreme altitude differences of over 3000 m between the cords that were not related to the climate, (Ribalaygua et al., 2013a).
plains (the Ebro River Basin) and the mountains (the Pyrenees), together The first set of data (composed of 103 stations) was used to analyse
with the specific topography of the Ebro River Basin and the mountain and produce the regional climate scenarios of maximum and minimum
chains (the Pyrenees to the north of the basin and the Iberian Mountains temperatures (Fig. 1a). Of these 103 stations, those that had at least 75%
to the south), modify the local climate. As a result, the climate character- of maximum temperature data collected from June to September (i.e,
istics of the area are somewhat different from a standard Western Med- the reference period for the calculation of heat waves) between 1980
iterranean climate, and they typically consist of dryness of the land along and 2000 and stations that had at least 75% of minimum temperature
the banks of the Ebro River, random rain patterns, high thermal contrast data collected from November to April (that is, the reference period
between winter and summer as a consequence of the strong continental for the calculation of cold waves) between 1980 and 2000 perform a
characteristics of the region and the typical northeast “mistral” winds, second set of data selected for the study of extreme events. In both
which are frequent in the region. cases, the number of observatories was 71 (Fig. 1b).

Fig. 1. Location of the study Area: Aragon (Spain) in Europe. Points indicate the stations used in the study. a) Stations of temperature (103) used in the generation of climate regional sce-
narios of maximum and minimum temperature (verification, validation and scenarios). b) Stations used exclusively on the generation of heat waves (7, red), used exclusively on the gen-
eration of cold waves (7, blue) and those used in both extreme events (64, green). “(Map source: OpenStreetMap)”.
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2781

2.1.3. Atmospheric dataset (predictors) 2) It is a key for the achievement of climate simulations that are consis-
tent with observations which, in turn, are physically coherent
2.1.3.1. Reanalysis. As a reference dataset, we used the reanalysis from the (Ribalaygua et al., 2013b).
European Centre for Medium-Range Weather Forecasts (ECMWF ERA-40; 3) Provides local detail, which is useful information because nearby
http://www.ecmwf.int/research/era/do/get/era-40) (Uppala et al., 2005) data points in space can evolve under different future climate condi-
for the period 1958–2000. For the atmospheric window, we used geo- tions (Ribalaygua et al., 2013b).
graphical limits from 31.5°N to 55.1°N latitude and from 27.0°W to
14.6°E longitude, covering not only the geographic area under study,
A two-step analogue/regression statistical downscaling method that
but also the surrounding atmosphere areas, which exert a meteorological
was developed by the Climate Research Foundation (FIC) (Ribalaygua
influence all over the Iberian Peninsula (Ribalaygua et al., 2013a).
et al., 2013b) was applied to obtain future scenarios of the maximum
For verification of the methodology, it was necessary to reduce the
and minimum temperatures at a local scale in the region of Aragón.
temporal (six hourly) and spatial (125 km) scale of the reanalysis to
This methodology consisted of a two-step analogue/regression statisti-
that of the different climate models in order to compare both, ERA-40
cal method, which has been used in national and international projects
and the climate model simulations (Ribalaygua et al., 2013a, 2013b).
with good verification results. In the present paper, we present a sum-
mary of the two-step method (more details can be found in
2.1.3.2. Climate model data. In the present study, we worked with a set of
Ribalaygua et al., 2013b).
nine climate models that belong to the CMIP5 and that were provided by
The first step was analogous stratification. An analogue method was
the Program for Climate Model Diagnosis and Intercomparison (PCMDI)
applied based on the hypothesis that ‘analogue’ atmospheric patterns
archives (Table 1). The number of models chosen was limited to the avail-
(predictors) should cause analogue local effects (predictands), which
ability of data with daily time frequencies. All of the models were ESMs.
means that the number of days that were most similar to the day to
To generate the model predictor fields, we took the daily values of sev-
be downscaled were selected (Benestad et al., 2007; Zorita and von
eral large-scale fields of interest (that is, geopotential height, specific hu-
Storch, 1999). The similarity between any two days was measured
midity and wind) at different pressure levels for the nine ESMs showed in
using a pseudo-Euclidean distance between the large-scale fields used
the Table 1 associated with the CMIP5 (Taylor et al., 2012). Moreover, we
as predictors. For each predictor, the weighted Euclidean distance was
used an historical experiment (Taylor et al., 2012) which generated sim-
calculated and standardised by substituting it with the closest percen-
ulations of the past 20th century and which was useful for the evaluation
tile of a reference population of weighted Euclidean distances for that
of model performance against the present climate. Finally, the representa-
predictor. This method is a good method for reproducing nonlinear re-
tive concentration pathway (RCP) families were considered (Moss et al.,
lationships between predictors and the predictands, but it could not
2010), specifically the RCP8.5 ‘high’ scenario and the RCP4.5 ‘intermedi-
be used to simulate values outside of the range of observed values
ate’ scenario, both of which corresponded to different possible ranges of
((Imbert and Benestad, 2005). In order to overcome this problem and
radiative forcing reached in the year 2100 with respect to values of the
to obtain a better simulation, a second step was required.
pre-industrial era (4.5 and 8.5 W/m2, respectively).
The second step focuses on temperature. To determine the temper-
ature, a multiple linear regression analysis for the selected number of
2.2. Methodologies
most analogous days was performed for each station and for each prob-
lem day. From a group of potential predictors, the linear regression
2.2.1. Description of the downscaling methodology
selected those with the highest correlation, using a forward and back-
The methodology of statistical downscaling used in this study was
ward stepwise approach.
chosen according to three main advantages:
About the accuracy in simulating temperatures of our downscal-
1) Allows easier quantification of the main uncertainties associated ing technique, in our previous work about climate change in Aragón
with the generation of future climate scenarios (Van der Linden we achieved an average bias below 0.1 °C″ (Ribalaygua et al.,
and Mitchell, 2009). 2013a).

Table 1
Information about the nine climate models belonged to the 5 Coupled Model Intercomparison Project (CMIP5) corresponding to the fifth report of the IPCC. Models were supplied by the
Program for Climate Model Diagnosis and Intercomparison (PCMDI) archives.

Climatic model SPATIAL/TEMPORAL Research centre References

resolution

GFDL-ESM2M 2°×2,5° National Oceanic and Atmospheric Administration (NOAA), E.E.U.U. Dunne et al. (2012)
daily
CanESM2 2,8°×2,8° Canadian Centre for Climate Modeling and Analysis (CC-CMA), Chylek et al. (2011)
daily Canadá.
CNRM-CM5 1,4°×1,4° CNRM (Centre National de Recherches Meteorologiques), Meteo-France, Francia. Voldoire et al. (2013)
daily
BCC-CSM1-1 1,4°×1,4° Beijing Climate Center (BCC), China Meteorological Administration, China. Xiao-Ge et al. (2013)
daily
HADGEM2-CC 1,87°×1,25° Met Office Hadley Center, United Kingdom. Collins et al. (2008)
daily
MIROC-ESM-CHEM 2,8°×2,8° Japan Agency for marine-Earth Science and Technology (JAMSTEC), Watanabe et al. (2011)
daily Atmosphere and Ocean Research Institute (AORI), and National Institute
for Environmental Studies (NIES), Japan.
MPI-ESM-MR 1,8°×1,8° Max-Planck Institute for Meteorology (MPI-M), Germany. Raddatz et al. (2007);
daily Marsland et al. (2003)
MRI-CGCM3 1,2°×1,2° Meteorological Research Institute (MRI), Japan. Yukimoto et al. (2011)
daily
NorESM1-M 2,5°×1,9° Norwegian Climate Centre (NCC), Norway. Bentsen et al. (2013);
daily Iversen et al. (2013)
2782 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

2.2.2. Validation and climate simulation of temperature future climate projections (IPCC, 2013). For each scenario (i.e, RCP4.5
In order to determine the ability of each ESM to simulate the predic- and RCP8.5), an ensemble of the approved downscaled ESMs was used
tor fields, absolute and relative temperatures from the downscaled ESM to estimate the mean change (compared to 1976–2005) and to quantify
simulation of the historical experiment was compared with the down- the main contributions to the uncertainty.
scaled ERA-40 simulation (previously verified against the observations) Future local climate scenarios for maximum and minimum temper-
during a common historical period (1958–2000). ature, for nine GCMs and two RCPs have been produced at the daily
Due to the characteristics of the ESMs, the validation process pre- scale. To draw the temperature maps, we used Thin Plate Spline regres-
sents certain limitations. First, as climate models do not reproduce sion (TPS) from the R-Package “fields” (Nychka et al., 2015).
day-to-day meteorology, validation cannot be performed on a daily
scale and it must be done using climate statistics over long periods of 2.2.3. Definition of hot- and cold-wave episodes
time, resulting in a loss of information on climate variability. Second, There are many and varied terms used to define a heat wave. For ex-
we could not compare the climate characteristics obtained from the ample, the AEMET defines a heat wave as an “episode of at least three
ESM historical simulation with those obtained from the observations consecutive days, in which at least 10% of the stations considered record
because the latter had missing data and large gaps; therefore, we had maximums above the 95% percentile of their series of maxima daily
to compare ESMs simulations with simulations from the re-anlaysis of temperatures of the months of July and August of the period
a dataset (previously validated in the verification process). Both sources 1971–2000.” The WMO defines a heat wave as an extreme event with
of error should be considered in the final uncertainty analysis. marked warmed of the air or the invasion of very warm air over a
As error measures, bias and standard deviation were analysed on a large area, it usually last from a few days to a few weeks. The IPCC de-
seasonal scale for both maximum and minimum temperatures. An en- fines a heat wave as “a period [of] abnormally and uncomfortably hot
semble strategy was used to quantify the uncertainties inherent in weather”. Those definitions are not the only ones accepted in the

Fig. 2. Validation of the GCMs used for the simulation of temperatures. Absolute Bias for a) maximum and c) minimum temperature and absolute standard deviations for b) maximum and
d) minimum temperature, between the results obtained by downscaling the Historical scenario of each GCM used in the study with those obtained by downscaling the reanalysis ERA-40
for a common period (1958–2000). There are four boxplots, representing a season of the year (winter, spring, summer and autumn), for each GCM.
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2783

Fig. 3. For the four seasons (winter, spring, summer and autumn) simulated maximum temperature for the twenty-first century displayed as absolute increase against the value simulated
for the 1976–2005 Historical period. The vertical dotted line marks the end of the Historical data (2005). Data grouped for every RCP simulation of every global climate model used and for
the last 30 years of every station. The ensemble median (solid lines) and the 10th–90th percentile (shaded areas) values are displayed.

scientific literature. Some heat-wave definitions have been used to significant drop of air temperatures near the surface (max, min and
identify heat waves in a time series of temperature data (Smith et al., daily average) over a large area and persisting below certain thresholds
2013;), and the choice of the heat wave definition can influence both for at least two consecutive days during the cold season”.
projected heat-wave trends (Smith et al., 2013) and estimates of health Considering the climate characteristics of Aragón, we defined a cold
risks during events (Anderson and Bell, 2009; Chen et al., 2015; Kent wave as follows: at least three consecutive days with a minimum tem-
et al., 2014). perature below the fifth percentile of a minimum temperature series
According to the recommendations of the WMO, (2010), a practical and calculated between the months of November to April during the pe-
and qualitative definition of a heat wave must consider marked and un- riod 1980–2000.
usual hot weather over a region during at least two consecutive days in To better identify a heat or cold wave, we evaluated average dura-
the hot period of the year, based on local climate conditions, with ther- tion, maximum intensity and average intensity as it is recommended
mal conditions recorded above given thresholds. On the basis of these by the WMO in the Guidelines on the definition and monitoring of ex-
recommendations and the characteristics of the Aragón climatology, treme weather and Climate Events. The average duration refers to the
we defined a heat wave as follows: at least three consecutive days average number of days the heat wave lasted. The average intensity rep-
with a maximum temperature above the 95th percentile of the maxi- resents the average value of the temperature during the heat wave. The
mum temperature series and calculated between the months of June maximum intensity is equal to the maximum value that the tempera-
to September during the period 1980–2000. ture reached during the heat wave.
In broad terms, a cold wave can be defined as a meteorological event
characterised by a sharp drop of air temperature near the surface, lead- 2.2.4. Verification and simulation of heat and cold waves
ing to extremely low values. According to the IPCC 2007, (Parry et al., In order to assess the capacity of the downscaling methodology to
2007) a cold wave is an event that often causes problems and severe im- simulate heat and cold waves observed in the past, we evaluated how
pacts on the population, especially in northern altitudes. However, there the downscaling methodology simulated the 95th percentile of maxi-
is still a lack of a clear and consistent definition for cold-wave events in mum temperatures and the 5th percentile of minimum temperatures,
the world. which are the thresholds that indicate the existence of an episode of a
In the WMO's Meteoterm vocabulary, a cold wave is defined as heat and cold wave, respectively. In addition, we analysed the intensity
“marked cooling of the air, or the invasion of very cold air, over a large of the wave (average and maximum) and the average duration,
area.” For the AEMET, a cold wave is “an episode of at least three consec- comparing the heat and cold waves calculated from the simulated
utive days, in which at least 10% of the considered stations register min- ERA-40 temperature series with those obtained from the observed
imums below the 5% percentile of their series of minima daily series. Verification of the maximum and minimum temperatures can
temperatures for the months of January and February of the period be seen in a previous study by Ribalaygua et al. (2013a). The statistical
1971–2000”. The WMO's members guideline define a cold wave as measures used in the verification processes were the bias, the standard
“marked and unusual cold weather characterised by a sharp and deviation and the Pearson correlation at the daily scale. The statistical
2784 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

Fig. 4. For the four seasons (winter, spring, summer and autumn) simulated minimum temperature for the twenty-first century displayed as absolute increase against the value simulated
for the 1976–2005 Historical period. The vertical dotted line marks the end of the Historical data (2005). Data grouped for every RCP simulation of every global climate model used and for
the last 30 years of every station. The ensemble median (solid lines) and the 10th–90th percentile (shaded areas) values are displayed.

measures were calculated using R-cran package computing software and winter minimum temperature changes expected for the periods
(R Development Core Team, 2010). 2041–2070 and 2071–2100 as presentation of mid- and end-century
From the ESM simulated temperature series (nine ESMs and two expected changes, which were calculated according to the scenarios
RCPs), we determined the heat- and cold-wave episodes that were ex- RCP4.5 and RCP8.5.
pected in Aragón during the upcoming decades of the 21st century. Expected changes in the average temperature for the rest of the
The heat- and cold-wave scenarios were compared to a historical period seasons of the year can be found in the supplementary material for
(1976–2005) to analyse the future changes with respect to the actual maximum temperatures increase (Figs. S1–S3) and minimum tempera-
situation of these extreme events. tures increase (Fig. S4–S6). In addition, Fig. S7–S9 showed the maxi-
mum and Figs. S10–S12, minimum temperatures expected in this
3. Results century.
Fig. 3 shows a gradual increase in the maximum temperatures
3.1. Validation of the CMIP5 model statistical downscaling to predict throughout the 21st century. During the mid-century, the least signifi-
temperatures cant changes are expected during the winter and springs months
(around 1.5 and 1.7 °C for the RCP4.5 and RCP8.5, respectively), and
The seasonal bias that result from a comparison between the ERA-40 the most significant changes are expected for the summer months,
temperature simulations and the historical temperature simulations for with values of 2.1 °C for the RCP4.5 and 2.7 °C for the RCP8.5. At the
each ESM for a common period (1958–2000) are shown in Fig. 2 for ab- end of the century, the expected changes are more marked and differ
solute temperatures (that is, the difference between simulated Histori- notably between the two RCPs. During winter and spring months, the
cal ESM data and simulated ERA-40 data). For both the maximum and expected values are around 2.3 °C and 2.4 °C for the RPC4.5 and 4.3 °C
minimum temperatures, the obtained bias was around tenths of a de- and 4.7 °C for the RCP8.5, respectively. The greatest changes are ex-
gree in all months, so they were very close to zero. The error was not pected for summer months, reaching values of 7 °C for the RCP8.5 and
above half of a degree for any of the cases. Therefore, the results showed 3.6 °C for the RCP4.5.
that the ESMs were capable of adequately simulating both the maxi- The areas most affected by the maximum temperature increases are
mum and the minimum temperatures on annual and seasonal scales. expected to be the Pyrenees and the areas in the southwest and north of
Aragón during all the 21th century for the months of summer, spring
3.2. Local climate scenarios to predict future temperatures and autumn, especially in the RCP8.5 scenario (Fig. 5). In contrast, in
the winter months, the highest temperature increases at the end of
Figs. 3 and 4 show local climate-change scenarios for future daily the century are expected in the Pyrenees area and the Ebro Valley
maximum and minimum temperatures respectively, that have been (Fig. S1).
predicted on the basis of the nine models (see Table 1). Figs. 3 and 4 Scenarios of maximum temperatures of the 21st century (Figs. 5 and
allowed for the establishment of a general view of the changes expected S7–S9) point that expected maximum temperatures are not propor-
in Aragón. In addition, Figs. 5 and 6 represent the summer maximum tional to the expected increases shown before (Figs. 5 and S1–S3).
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2785

Fig. 5. Geographical representation of the expected changes of maximum temperature in summer for the periods 2041–2070 and 2071–2100 respect to the reference Historical Period
(1976–2005). Both emissions scenarios are represented: RCP4.5 (figures b and c) and RCP8.5 (figures d and e). Fig. 5a represents the Historical absolute temperature for the period
1976–2005.

The evolution of the minimum temperature increases throughout well simulated by applying the downscaling methodology to the ERA-
the 21st century demonstrate similar behaviours to the maximum tem- 40 series (Fig. 7). Data from the simulated ERA-40 series and the ob-
peratures, but with less marked increases, especially at the end of the served series were very close, and the error committed by simulating
century (Figs. 6 and S4–S6). By the middle of the century, all of the sea- the ERA-40 series respect to the observed one was very low, especially
sons of the year, except for the summer, are expected to show increases in the 95th percentile. For example, we found error values around
in the minimum temperature of b2.0 °C, and even in the spring, the in- 0.36 °C for summer months and around 0.69 °C for the winter months
creases are expected to be close to 1.0–1.2 °C. The summer months are in the 95th percentile, and 0.6 °C for summer months and around 0.68
expected to show the greatest increases in minimum temperatures, °C for the winter months in the 5th percentile.
varying from 1.8 °C for the RCP4.5 to 2.3 °C for the RCP8.5. At the end Regarding the verification of the identification parameters of a heat
of the century, increases in the minimum temperatures during the sum- wave, Fig. 8 shows the results obtained for the duration of a heat
mer months are expected to range from 3.0 °C (RCP4.5) to 5.6 °C wave, its temporary occurrence and its average intensity (maximum in-
(RCP8.5. Winter months are expected to show an increase between tensity is not shown here, but it was also taken into account in the pres-
2.8 °C and 4.1 °C, spring months between 2.1 °C and 4 °C and autumn ent study and can be seen in Support Information (S13)).
months between 2.6 °C and 5 °C for the RCP4.5 and RCP8.5 scenarios, As can be seen in Fig. 8a and b the average intensity of observed heat
respectively. waves and ERA-40 simulations are very similar in the 71 observatories.
The increases in the minimum temperature in are expected to be We found a high correlation between the average intensity of a heat
quite homogeneous throughout the region and the seasons according wave and the results from the observed and the simulated ERA-40 sce-
to the RCP4.5. In contrast, according to RCP8.5 (Fig. 6c,d), these changes narios (p = 0.9889). A similarly high correlation was obtained in terms
will be more pronounced in the Ebro Valley at the end of the 21st cen- of the maximum intensity of a heat wave (p = 0.9883, not shown here).
tury. The highest minimum temperatures are expected to be recorded Regarding to the average duration of heat waves episodes, Fig. 8c
in the central zone of Aragón during all seasons of the year (see and d showed a similar pattern of temporal and spatial distribution
supporting information, Figs. S10–S12). and a strong correlation (p = 0.8976). For example, both datasets
showed a clear signal of heat waves during the years 1982, 1987,
3.3. Verification of hot- and cold-wave episodes 1990, 1994 and 1998, and they both showed no heat wave during the
years 1986, 1987, 1996 and 1997.
To verify the simulation of both of heat waves and cold waves, we Similar results are present in Fig. 9 for cold waves. In this case, the
compared the wave episodes obtained from ERA-40 simulations against correlation obtained between the average duration of the observed
the episodes recorded by the observations. cold waves (Fig. 9c) and those simulated for ERA-40 (Fig. 9d) was p =
The first step was to verify that the methodology correctly simulated 0.8593, which was lower than the p-value obtained for heat waves.
both percentiles, the 95th percentile of the maximum temperature and The correlations of the intensities between both groups of data were
the 5th percentile of the minimum temperature. Both cases have been strong (p = 0.8849 and p = 0.792 for maximum intensity and average
2786 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

Fig. 6. Geographical representation of the expected changes of minimum temperature in winter for the periods 2041–2070 and 2071–2100 respect to the reference Historical Period
(1976–2005). Both emissions scenarios are represented: RCP4.5 (up, figures b and c and RCP8.5 (figures d and e). Panel a represents the Historical absolute temperature for the period
1976–2005.

intensity, respectively), but lower than those obtained for the heat In the case of the maximum intensity of the heat waves, the average
waves. The cold-wave pattern (that is, duration, average intensities increases with respect to the expected historical period were somewhat
and maximum intensities) were very similar in both datasets, marking more pronounced than those expected for the average intensity, espe-
episodes of cold waves during the same time periods. cially in the case of RCP8.5. According to RCP4.5, average increases of
1.4 °C and 1.7 °C (reaching 39.3 °C and 39.6 °C) are expected in the max-
imum intensity during the middle and end of the century, respectively.
3.4. Local future climate scenarios of hot- and cold-waves episodes In the case of RCP8.5, those average increases are expected to rise to 2 °C
and 3.6 °C (reaching 39.9 °C and 41.5 °C) during the middle and end of
Figs. 10 and 11 show the climate scenarios for heat waves in Aragón the century, respectively.
according to the RCP4.5 and the RCP8.5scenarios, respectively. The fig- Note that for the average (maximum) intensity, there was a differ-
ures show the temporal evolution (from historical data to 2100) for ence between observatories, such that in some of them the intensity
three main features of the heat waves: average intensity (first row), could reach 46 °C (48 °C), while in other parts of the region it would
maximum intensity (second row) and average duration (third row). not exceed 32 °C (34 °C) according to the RCP8.5.
The temporal periods analysed were 1976–2005 (historical; first col- The average and maximum intensity of the heat waves will be espe-
umn), 2041–2070 (second column) and 2071–2100 (third column). cially intense in the Ebro Valley in the middle of the century, and it will
These periods were chosen as representative of the present, mid- spread throughout the entire territory toward the end of the century,
century and end-century periods. Figs. 12 and 13 show the same infor- except for the Pyrenean area further north.
mation but for cold waves. According to RCP4.5, an average increase in the number of days that
In addition, in Support information heat and cold waves are repre- lasts a heat waves (duration) compared to the historical period of about
sented as increments in temperature for heat waves (S15 and S16 for 2 days is expected during mid-century (2041–2070), with a maximum
RCP4.5 and the RCP8.5 scenarios, respectively) and for cold waves increase of 11 days, which would mean that the average duration of
(S17 and S18 for RCP4.5 and the RCP8.5 scenarios, respectively). heat waves it would be around 7 days, with durations that vary between
During the coming decades, an increase in the duration of heat 4 and 18 days according to the station. At the end of the century, very
waves is expected. According to the RPC4.5, sharp increases in the aver- sharp variations are not expected with respect to half a century. Results
age intensity of heat waves are not expected with respect to the values from the RCP8.5 predict that the average duration of heat waves will in-
corresponding to the historical period (an average of approximately 0.8 crease by 3 days in the middle of the century and by 7 days at the end of
°C). This means that a mean intensity of 37.6 °C is expected at mid- the century, which means that heat waves would have an average dura-
century and 37.8 °C at the end of the century. According to the RCP8.5, tion of between 8 and 12 days, respectively. Note that according to the
the average intensity would increase an average of approximately RCP8.5 scenario, some observatories are expected to experience sharp
1–1.2 °C during the middle of the century (reaching 38 °C) and close increases during the historic period of up to 16 and 44 days for the pe-
to 2 °C at the end of the century (reaching 38.8 °C). riod 2041–2070 and 2071–2100 respectively, so there would be
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2787

Fig. 7. Results of the verification process obtained for a) the 95th percentile of maximum temperature and b) the 05th percentile of minimum temperature comparing the results obtained
by downscaling the reanalysis ERA-40 (black) along the observations (red) for a common (1980–2000). The solid lines represent the median and the shaded areas the 10-90th percentile
of the values.

observatories where the consecutive days that exceed the “hot-day” new ESMs belonging to the CMIP5. In addition, the climate scenarios for
threshold would be 23 and 52 days, for the same periods. temperatures in this region were reconstructed with the new ESMs in
In this case, the Pyrenees to the north and the Iberian to the south of order to assess differences with respect to the models used prior to
Aragón are the zones that would suffer longer heat waves; although, as CMIP5. These results offer one of the best snapshots of future climate
we have seen, they would be the least affected by the intensity of the ex- change based on currently available data on the risks that extreme tem-
pected heat waves. peratures can cause in the Aragón region, both spatially and temporally.
On the other hand, the episodes of cold waves would not change ei-
ther the average or maximum intensity values (see Figs. 12 and S17 for
the RCP4.5 scenario and Figs. 13 and S18 for the RCP8.5). The variations 4.1. Considerations of the temperature scenarios
of average intensity with respect to the historic period are, on average,
±0.5 °C, and in some observatories they could reach ±1.0–1.°5 °C, inde- The new models used in the present study to obtain climate scenar-
pendent of the RCP. ios in Aragón are ESMs, not climate models. The improvements incorpo-
A similar pattern is expected for the maximum intensity of the cold rated in the ESMs allow for greater precision in the simulation of climate
waves (Figs. 12b and 13b), with few variations with respect to the his- variables, which provides evidence that the results of the validation are
torical period and similar patterns to those expected for the mean inten- good.
sity. The north and the south of the territory are the regions that are The validation of the ESMs was good for both maximum and mini-
expected to be the most affected by the intensity of the cold waves. mum temperatures. The differences between seasonal means were
During the upcoming decades, an increase in the duration of cold- practically negligible (below a few tenths of a degree in temperature),
wave episodes is not expected. Both RCP4.5 (Figs. 12h and S17h) and and the differences between seasonal standard deviations were almost
RCP8.5 (Figs. 13h and S18h) predict that by the middle of the century, always below the reference values for every ESM (the standard devia-
the average duration of cold waves will remain around 4 to 5 days, with tion of the downscaled reanalysis data).
episodes of 3 to 7 days in the middle of the century and 11 days at the The validation results obtained in the present study show lower
end of the century. This would mean that the average duration of cold error rates than those obtained by the scenarios described in the fourth
waves would not be modified with respect to the historical period, unlike IPCC report and presented in previous studies (Ribalaygua et al., 2013a).
what is expected in the case of heat waves, and in the most severe cases, In that case, the error was 1 °C, and even in some models (for example,
the increase would be about three days at the end of the century. No dif- the CNCM3 model in the summer), it was around 1.5 °C; in contrast, the
ferences of this pattern are expected in the territory. results presented here did not reach such high values. Likewise, the
values of the standard deviation are lower than those obtained in the
4. Discussion previous study (5–10% in the present study vs 15–20% in a previous
study (Ribalaygua et al., 2013a).
In the present study, heat- and cold-wave scenarios for the 21st cen- Another key point that justifies the generation of new climate sce-
tury were obtained for the first time in the region of Aragón, Spain, using narios for Aragón is the current designation of climate scenarios
2788 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

Fig. 8. Results of the verification process for heat waves obtained comparing the simulated heat waves for the reanalysis Era40 along the observed heat waves for a common period
(1980–2000). a) Average intensity of the heat waves registered in the 71 stations used in the study against b) the average intensity obtained by downscaling of the reanalysis ERA-40
in these stations for the period 1980–2000. The colour of each pixel shows the average intensity that corresponds to each heat wave (°C), each column represents one year of the
considered period and each row one station The upper row shows the mean of the average intensity from the whole observatories. Down, c) and d) same as a) and b) but for the
duration of the heat waves. The colour of each pixel shows the duration corresponding to each heat wave (number of days).

provided by IPCC experts—that is, RCPs. Therefore, obtaining local cli- projections (using the greatest possible number of climate scenarios
mate scenarios that are based on the most current information is a and downscaling methodologies). Therefore, the scenarios generated
priority. by the AEMET are complementary to those presented here because dif-
The results obtained in the present study, both in the validation pro- ferent statistical downscaling methodologies, as well as different ESMs,
cess and in the verification process, allow for the maximum and mini- have been applied.
mum temperature scenarios to be possibly used in subsequent studies Similar processes were carried out in the generation of climate sce-
of extreme events and bioclimatic variables. Future climate scenarios narios for Spain published by the AEMET (www.Aemet.es) and in the
(maximum and minimum temperatures) show an evolution toward EsTcena project (Brands et al., 2013; Brands et al., 2011a). Results that
warmer climates throughout the Aragón region. These results agree were generated by different centres were presented on the basis of dif-
with those obtained from the CMIP5 models by the IPCC (IPCC, 2013) ferent methodologies, including the one used in the present study.
and by the AEMET (www.Aemet.es). Previous data collected by other We compared the new scenarios that were based on the models of
authors and assembled by the IPCC (IPCC, 2013) expect that the greatest the fifth report of the IPCC with those published by Ribalaygua
temperature increases will occur during the summer months. The re- (Ribalaygua et al., 2013a), which correspond to the fourth report of
sults of the present study showed that increases in temperature (both the IPCC. We observed that the most recent scenarios trended toward
maximum and minimum) were less marked than those published by higher temperatures than what was expected in (Ribalaygua et al.,
the AEMET (for example, our results showed changes of 3.6 and 7 °C 2013a), with greater increases occurring during summer months com-
at the end of the century, while the AEMET results predicted changes pared to winter months. For example, for the summer months, the
of 4.1 and 8.2 °C under RPC4.5 and RCP8.5 scenarios, respectively), al- new results expect that for the period 2071–2100, the maximum tem-
though both our results and the AEMET results suggest that the summer perature in the most extreme case will reach up to 7 °C, while those pre-
months will be the time during which the most abrupt changes in tem- sented by Ribalaygua (Ribalaygua et al., 2013a) for the same period
perature will be observed, while the winter months will be the time were estimated to reach 5 °C. However, both studies agreed that the
during which the least abrupt changes will be observed. northwest and southwest regions would be the most affected by these
Results presented by the IPCC came from the ESMs, so they do not variations.
have the added value of applying downscaling techniques to obtain re- In both cases, maximum and minimum temperatures, in view of the
sults at a local scale. On the other hand, one of the advantages of work- validation results and in the face of the same reanalysis of ERA-40, it is
ing with climate-change scenarios is to have the widest range of future justified that the scenarios generated for the fifth IPCC report are more
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2789

Fig. 9. Results of the verification process for cold waves obtained comparing the simulated cold waves for the reanalysis Era40 along the observed cold waves for a common period
(1980–2000). a) Average intensity of the heat waves registered in the 71 stations used in the study against b) the average intensity obtained by downscaling of the reanalysis ERA-40
in these stations for the period 1980–2000. The colour of each pixel shows the average intensity that corresponds to each cold wave (°C), each column represents one year of the
considered period and each row one station. The upper row shows the mean of the average intensity from the whole observatories. Down, c) and d) same as a) and b) but for the
duration of the cold wave. The colour of each pixel shows the duration corresponding to each cold wave (number of days).

precise than those that already exist for the Aragón region. Having a the 95th percentile (or 5th percentile) of the maximum (or mini-
range of climate-scenario projections in Aragón facilitates the evalua- mum) temperature was greater in the ERA-40 series compared to
tion of extreme phenomena of interest, such as heat and cold waves, the observed series.
which have strong impacts in the region. 2) In general, downscaling techniques tend to soften the simulated
temperature series with respect to the observed values, which im-
4.2. Consideration about the simulation of heat and cold waves plies that certain episodes of hot or cold waves are softened in
their duration or intensity compared to the observed data and
Our results allow for the simulation of episodes of heat and cold would therefore be classified differently. In the present study, it
waves throughout the 21st century and prove the validity of the simu- was appreciated that the average and maximum intensities of the
lations, as is evident by the good verification and validation results. As observed cold and heat waves was greater than that simulated for
the methodology is able to adequately simulate the 95th percentile of the ERA-40 reanalysis.
maximum temperatures and the 5th percentile of minimum tempera-
tures, it is assumed that the thresholds that define the heat or cold
In summary, the uncertainties associated with both verification and
waves are being simulated in an appropriate way.
validation must be considered when interpreting future scenarios.
The downscaling methodology that was used was able to success-
The results of this study about heat and cold waves are consistent
fully simulate heat and cold waves reported by the observations, as
with the results published by both the IPCC (IPCC, 2013) and the
shown by the results of the verification process. The ESMs were also
AEMET (www.Aemet.es). All models agree that the number of days con-
successful at simulating the maximum and minimum temperatures
sidered as “hot days” will increase in the coming decades and, therefore,
during the validation process. However, in the verification process of
the heat waves will be more frequent. On the other hand, the number of
heat and cold waves, were identified some limitations that must be con-
“cold days” will be maintained and, in some cases, will even decrease so,
sidered in the simulation of future scenarios:
on the contrary of what is expected for heat waves, the frequency of cold
1) There were gaps in the observed series. The lack of data in the ob- waves will remain at its current frequency.
served series reduced the length of observed data compared to the What is stated in the previous paragraph is plausible with one of the
reanalysed ERA-40 series (that is, the full series without gaps) for main theoretical conclusions about the trend of temperatures reported
the same period of time. Therefore, the number of days included in by the IPCC (IPCC, 2013), which is based on the change in the probability
2790 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

Fig. 10. Geographical representation of the expected evolution of the heat waves for Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period
(1971–2000) in terms of absolute values according to the RCP4.5. The rows show the three parameters analysed in the study (average intensity, maximum intensity and duration) and
the columns the three temporal periods (Historical, 2041–2070 and 2071–2100). The maps are generated by interpolating the available stations over the entire territory.

of reaching certain temperature values. The probability distribution of the end of the century and in the summer (around 40 °C), as well as
temperatures can either move to warmer climates without undergoing the greatest intensity of heat waves. This can have important impacts
changes, can be extended or a combination of both. The third situation is on the health of the population. The health risks associated with heat
the one that has been observed in recent decades in different studies waves are well known, both in terms of mortality (Robine et al., 2008;
(Hansen et al., 2012) IPCC, 2001) and is the most plausible situation in Roldan et al., 2016) and mobility (Lin et al., 2009; Steul et al., 2018), as
the context of the observed data. The new distribution of well as health costs (Roldan et al., 2015). Moreover, these extreme
temperatures implies a less pronounced change in the colder tempera- events will cause significant socio-economic impacts because a large
tures of the series, a much more marked change in the warmer temper- part of Aragonese industry and farming is situated in this zone
atures and a greater number of warm extreme events. These (Olesen et al., 2011).
conclusions support the results anticipated for the coming decades— In addition, the Pyrenees will suffer the longest heat waves (but not
i.e, the maintenance of cold waves (in number and intensity) and an the most intense ones), especially at the end of the century, and the
increase of heat waves. greatest increases in maximum temperatures with respect to the values
Regarding the territory of Aragón, the scenarios obtained in the pres- recorded during the historical period (1971–2000). The northern zone
ent study indicate that the Ebro valley, the most populated area in the is where the greatest intensity of cold waves will also be located, al-
region, will observe the highest maximum temperatures, especially at though their durations will not change much compared to the current
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2791

Fig. 11. Geographical representation of the expected evolution of the heat waves for Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period
(1971–2000) in terms of absolute values according to the RCP8.5. The rows show the three parameters analysed in the study (average intensity, maximum intensity and duration) and
the columns the three temporal periods (Historical, 2041–2070 and 2071–2100). The maps are generated by interpolating the available stations over the entire territory.

cold waves. Therefore, it is the heat that is the most outstanding future that could cause flooding in populated areas near the river, a risk that
risk for these ecosystems. has been well evaluated in the literature (Gobiet et al., 2014;
Changes in temperature (even if they are moderate) and warming Guerreiro et al., 2018) and that will demand new and innovative adap-
are potential risks to this rich ecological area, which is characterised tation strategies (De Martino et al., 2012).
by a diversity of species and a wealth of native plants and animals, giv- Finally, the disappearance of snow due to a rise in temperature in the
ing rise to national parks such as Ordesa and Monte Perdido (high area could lead to a negative impact on the winter tourism industry and
mountains). Therefore, these areas are particularly vulnerable to any the Pyrenean ski resorts, which are currently an important part of the
changes in climate that could lead to radical alterations in habitats or region's economy, especially in the Pyrenean valleys (Gilaberte-
cause losses in their rich biodiversity, as has been previously described Burdalo et al., 2014; Gilaberte-Burdalo et al., 2017; Lopez-Moreno
(Kulakowski et al., 2017). et al., 2011; Pons et al., 2015).
Furthermore, the Pyrenees area decides the amount of water avail-
able for urban and agricultural use, with a direct effect on hydrology 5. Conclusions
and agriculture in the Ebro Valley. Changes in temperature and snow ac-
cumulation could seriously threaten the sustainability and the equilib- Our study of data from Aragón, Spain, is the first to simulate episodes
rium between available resources and water demand. On the other of heat and cold waves throughout the 21st century in a valid way, based
hand, abrupt temperature changes can lead to accelerated thawing on the good verification and validation results. In addition, we used
2792 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

Fig. 12. Geographical representation of the expected evolution of the cold waves for Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period
(1971–2000) in terms of absolute values according to the RCP4.5. The rows show the three parameters analysed in the study (average intensity, maximum intensity and duration) and
the columns the three temporal periods (Historical, 2041–2070 and 20,712,100). The maps are generated interpolating the available observatories over the entire territory.

climate models to obtain more precise climate scenarios for the tempera- intensities will be especially high in the Ebro Valley, the most popu-
tures in this territory compared to previously available scenarios. lated area. However, the Pyrenees will suffer the longest heat waves,
The use of ESMs from the fifth report of the IPCC and a methodology especially at the end of the century, and the greatest increases in
of downscaling has proven to be effective, allowing us to obtain greater maximum temperatures.
precision in the simulation of the climate variables in this territory and The present study provides useful information coming from the
to predict a realistic picture of the risks that extreme temperatures can downscaling for the first time in Aragón of models from IPCC5 to sup-
cause in the Aragón region, both spatially and temporally. The climate port decision-making and in the development of specific measures to
scenarios predict higher maximum temperatures compared to previous prevent socio-economic, environmental and human health impacts
scenarios. The greatest increases will occur during summer at the end of due climate change in Aragón, a territory that can be a good indicator
the century, reaching values of up to 7 °C in the most unfavourable of the impacts of climate change in southern Europe.
scenario.
Climate heat- and cold-wave scenarios showed that heat waves Appendix A. Supplementary data
will be longer and more intense in upcoming decades. However, in-
tensity and duration of cold waves will not change much compared Supplementary data to this article can be found online at https://doi.
to the current cold waves. The temperatures and heat-wave org/10.1016/j.scitotenv.2018.09.352.
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2793

Fig. 13. Geographical representation of the expected evolution of the cold waves for Aragon in the periods 2041–2070 and 2071–2100 compared to the reference Historical Period
(1971–2000) in terms of absolute values according to the RCP8.5. The rows show the three parameters analysed in the study (average intensity, maximum intensity and duration) and
the columns the three temporal periods (Historical, 2041–2070 and 20,712,100). The maps are generated interpolating the available observatories over the entire territory.

References Brands, S., Herrera, S., Fernandez, J., Gutierrez, J.M., 2013. How well do CMIP5 Earth Sys-
tem Models simulate present climate conditions in Europe and Africa? Clim. Dyn. 41,
Anderson, B.G., Bell, M.L., 2009. Weather-related mortality how heat, cold, and heat 803–817.
waves affect mortality in the United States. Epidemiology 20, 205–213. Buerger, C.M., Kolditz, O., Fowler, H.J., Blenkinsop, S., 2007. Future climate scenarios and
Asong, Z.E., Khaliq, M.N., Wheater, H.S., 2016. Projected changes in precipitation and tem- rainfall-runoff modelling in the Upper Gallego catchment (Spain). Environ. Pollut.
perature over the Canadian prairie provinces using the generalized linear model sta- 148.
tistical downscaling approach. J. Hydrol. 539, 429–446. Carvalho, D., Rocha, A., Gomez-Gesteira, M., Santos, C.S., 2017. Potential impacts of climate
Barrera-Escoda, A., Goncalves, M., Guerreiro, D., Cunillera, J., Baldasano, J.M., 2014. Projec- change on European wind energy resource under the CMIP5 future climate projec-
tions of temperature and precipitation extremes in the North Western Mediterra- tions. Renew. Energy 101, 29–40.
nean Basin by dynamical downscaling of climate scenarios at high resolution Chen, K., Bi, J., Chen, J., Chen, X., Huang, L., Zhou, L., 2015. Influence of heat wave defini-
(1971–2050). Clim. Chang. 122, 567–582. tions to the added effect of heat waves on daily mortality in Nanjing, China. Sci.
Benestad, R.E., Hanssen-Bauer, I., Forland, E.J., 2007. An evaluation of statistical models for Total Environ. 506-507, 18–25. https://doi.org/10.1016/j.scitotenv.2014.10.092
downscaling precipitation and their ability to capture long-term trends. Int. PMID: 25460935.
J. Climatol. 27. Chen, Y.D., Li, J.F., Zhang, Q., 2016. Changes in site-scale temperature extremes over China
Bentsen, M., Bethke, I., Debernard, J.B., Iversen, T., Kirkevag, A., Seland, O., et al., 2013. The during 2071–2100 in CMIP5 simulations. J. Geophys. Res.-Atmos. 121, 2732–2749.
Norwegian Earth System Model, NorESM1-M - part 1: description and basic evalua- Chylek, P., Li, J., Dubey, M.K., Wang, M., Lesins, G., 2011. Observed and model simulated
tion of the physical climate. Geosci. Model Dev. 6, 687–720. 20th century Arctic temperature variability: Canadian Earth System Model CanESM2.
Brands, S., Herrera, S., San-Martin, D., Gutierrez, J.M., 2011a. Validation of the ENSEMBLES Atmos. Chem. Phys. Discuss. 11, 22893–22907. https://doi.org/10.5194/acpd-11-
global climate models over southwestern Europe using probability density functions, 22893-2011.
from a downscaling perspective. Clim. Res. 48, 145–161. Collins, W.J., Bellouin, N., Doutriaux-Boucher, M., Gedney, N., Hinton, T., Jones, C.D.,
Brands, S., Taboada, J.J., Cofino, A.S., Sauter, T., Schneider, C., 2011b. Statistical downscaling Liddicoat, S., Martin, G., O’Connor, F., Rae, J., Senior, C., Totterdell, I., Woodward, S.,
of daily temperatures in the NW Iberian Peninsula from global climate models: vali- Reichler, T., Kim, J., Halloran, P., 2008. Evaluation of the HadGEM2 model. Hadley Cen-
dation and future scenarios. Clim. Res. 48. tre Technical Note HCTN. vol 74. Met Office Hadley Centre, Exeter, UK.
2794 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795

Cook, J., Nuccitelli, D., Green, S.A., Richardson, M., Winkler, B., Painting, R., Way, R., Jacobs, Kattenberg, A., Amer Meteorol SOC, 1996. Climate models: projections of future climate.
P., Skuce, A., 2013. Quantifying the consensus on anthropogenic global warming in Seventh Symposium on Global Change Studies.
the scientific literature. Environ. Res. Lett. 8, 024024. https://doi.org/10.1088/1748- Kent, S.T., McClure, L.A., Zaitchik, B.F., Smith, T.T., Gohlke, J.M., 2014. Heat waves and
9326/8/2/024024. health outcomes in Alabama (USA): the importance of heat wave definition. Environ.
De Martino, G., De Paola, F., Fontana, N., Marini, G., Ranucci, A., 2012. Experimental assess- Health Perspect. 122, 151–158.
ment of level pool routing in preliminary design of floodplain storage. Sci. Total Envi- Kim, S., Sinclair, V.A., Raisanen, J., Ruuhela, R., 2018. Heat waves in Finland: present and
ron. 416, 142–147. projected summertime extreme temperatures and their associated circulation pat-
Dosio, A., 2017. Projection of temperature and heat waves for Africa with an ensemble of terns. Int. J. Climatol. 38, 1393–1408.
CORDEX regional climate models. Clim. Dyn. 49, 493–519. Knutti, R., Sedlacek, J., 2013. Robustness and uncertainties in the new CMIP5 climate
Dunne, J.P., John, J.G., Adcroft, A.J., Griffies, S.M., Hallberg, R.W., Shevliakova, E., et al., 2012. model projections. Nat. Clim. Chang. 3, 369–373.
GFDL's ESM2 global coupled climate-carbon Earth System Models. Part I: physical for- Kuglitsch, F.G., Toreti, A., Xoplaki, E., Della-Marta, P.M., Zerefos, C.S., Turkes, M., et al.,
mulation and baseline simulation characteristics. J. Clim. 25, 6646–6665. 2010. Heat wave changes in the eastern Mediterranean since 1960. Geophys. Res.
Estrella, N., Menzel, A., 2013. Recent and future climate extremes arising from changes to Lett. 37.
the bivariate distribution of temperature and precipitation in Bavaria, Germany. Int. Kulakowski, D., Seidl, R., Holeksa, J., Kuuluvainen, T., Nagel, T.A., Panayotov, M., et al.,
J. Climatol. 33, 1687–1695. 2017. A walk on the wild side: disturbance dynamics and the conservation and
Fernandez-Montes, S., Rodrigo, F.S., 2012. Trends in seasonal indices of daily temperature management of European mountain forest ecosystems. For. Ecol. Manag. 388,
extremes in the Iberian Peninsula, 1929–2005. Int. J. Climatol. 32, 2320–2332. 120–131.
Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken, D.J., Ebi, K.L., Mastrandrea, M.D., Mach, Lavaysse, C., Vrac, M., Drobinski, P., Lengaigne, M., Vischel, T., 2012. Statistical downscal-
K.J., Plattner, G.K., Allen, S.K., Tignor, M., Midgley, P.M., IPCC, 2012. Managing the Risks ing of the French Mediterranean climate: assessment for present and projection in an
of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge anthropogenic scenario. Nat. Hazards Earth Syst. Sci. 12, 651–670.
University Press, Cambridge, United Kingdom and New York, NY, USA, p. 582. Lin, S., Luo, M., Walker, R.J., Liu, X., Hwang, S.A., Chinery, R., 2009. Extreme high temper-
Fischer, E.M., Schar, C., 2010. Consistent geographical patterns of changes in high-impact atures and hospital admissions for respiratory and cardiovascular diseases. Epidemi-
European heatwaves. Nat. Geosci. 3, 398–403. ology 20, 738–746.
Fonseca, D., Carvalho, M.J., Marta-Almeida, M., Melo-Goncalves, P., Rocha, A., 2016. Recent López, F., Cabrera, M., Cuadrat, J.M., 2007. Atlas Climático de Aragón. First ed. J. Factory,
trends of extreme temperature indices for the Iberian Peninsula. Phys. Chem. Earth Spain.
94, 66–76. Lopez-Moreno, J.I., Goyette, S., Vicente-Serrano, S.M., Beniston, M., 2011. Effects of climate
Frias, M.D., Fernandez, J., Saenz, J., Rodriguez-Puebla, C., 2005. Operational predictability change on the intensity and frequency of heavy snowfall events in the Pyrenees. Clim.
of monthly average maximum temperature over the Iberian Peninsula using Chang. 105, 489–508.
DEMETER simulations and downscaling. Tellus Ser. A Dyn. Meteorol. Oceanogr. 57. Marsland, S.J., Haak, H., Jungclaus, J.H., Latif, M., Roske, F., 2003. The Max-Planck-Institute
Frias, M.D., Herrera, S., Cofino, A.S., Gutierrez, J.M., 2010. Assessing the skill of precipita- global ocean/sea ice model with orthogonal curvilinear coordinates. Ocean Model 5,
tion and temperature seasonal forecasts in Spain: windows of opportunity related 91–127.
to ENSO events. J. Clim. 23. Miro, J.J., Estrela, M.J., Caselles, V., Olcina-Cantos, J., 2016. Fine-scale estimations of biocli-
Garcia-Herrera, R., Diaz, J., Trigo, R.M., Hernandez, E., 2005. Extreme summer tempera- matic change in the Valencia region, Spain. Atmos. Res. 180, 150–164.
tures in Iberia: health impacts and associated synoptic conditions. Ann. Geophys. Moss, R.H., Edmonds, J.A., Hibbard, K.A., Manning, M.R., Rose, S.K., van Vuuren, D.P., et al.,
23, 239–251. 2010. The next generation of scenarios for climate change research and assessment.
Gilaberte-Burdalo, M., Lopez-Martin, F., Pino-Otin, M.R., Lopez-Moreno, J.I., 2014. Impacts Nature 463, 747–756.
of climate change on ski industry. Environ. Sci. Pol. 44, 51–61. Nychka, D., Furrer, R., Paige, J., Sain, S., 2015. Fields: tools for spatial data. R package ver-
Gilaberte-Burdalo, M., Lopez-Moren, J.I., Moran-Tejeda, E., Jerez, S., Alonso-Gonzalez, E., sion 9.0. (URL: http://doi.org/10.5065/D6W957CT). www.image.ucar.edu/fields.
Lopez-Martin, F., et al., 2017. Assessment of ski condition reliability in the Spanish and Olesen, J.E., Trnka, M., Kersebaum, K.C., Skjelvag, A.O., Seguin, B., Peltonen-Sainio, P., et al.,
Andorran Pyrenees for the second half of the 20th century. Appl. Geogr. 79, 127–142. 2011. Impacts and adaptation of European crop production systems to climate
Giorgi, F., Gutowski, W.J., 2015. Regional dynamical downscaling and the CORDEX initia- change. Eur. J. Agron. 34, 96–112.
tive. Annu. Rev. Environ. Resour. 40, 467–490. Parry, M.L., Canziani OF, Palutikof, J.P., Van der Linden, P.J., Hanson, C.E. (Eds.), 2007. IPCC
Giorgi, F., Mearns, L.O., 1991. Approaches to the simulation of regional climate change - a Impacts, Adaptation and Vulnerability Contribution of Working Group II to the Fourth
review. Rev. Geophys. 29, 191–216. Assessment Report of the IPCC. Cambridge University Press.
Giorgi, F., Jones, C., Asrar, G.R., 2009. Addressing climate information needs at the regional Patz, J.A., Campbell-Lendrum, D., Holloway, T., Foley, J.A., 2005. Impact of regional climate
level: the CORDEX framework. WMO Bull. 58 (3), 175–183. change on human health. Nature 438, 310–317.
Gobiet, A., Kotlarski, S., Beniston, M., Heinrich, G., Rajczak, J., Stoffel, M., 2014. 21st century Perez, J., Menendez, M., Mendez, F.J., Losada, I.J., 2014. Evaluating the performance of
climate change in the European Alps-a review. Sci. Total Environ. 493, 1138–1151. CMIP3 and CMIP5 global climate models over the north-east Atlantic region. Clim.
Goncalves, M., Barrera-Escoda, A., Guerreiro, D., Baldasano, J.M., Cunillera, J., 2014. Sea- Dyn. 43, 2663–2680.
sonal to yearly assessment of temperature and precipitation trends in the North Planton, S., Deque, M., Chauvin, F., Terray, L., 2008. Expected impacts of climate change on
Western Mediterranean Basin by dynamical downscaling of climate scenarios at extreme climate events. Compt. Rendus Geosci. 340, 564–574.
high resolution (1971–92050). Clim. Chang. 122, 243–256. Pons, M., Lopez-Moreno, J.I., Rosas-Casals, M., Jover, E., 2015. The vulnerability of Pyre-
Gross, M.H., Alexander, L.V., Macadam, I., Green, D., Evans, J.P., 2017. The representation of nean ski resorts to climate-induced changes in the snowpack. Clim. Chang. 131,
health-relevant heatwave characteristics in a regional climate model ensemble for 591–605.
New South Wales and the Australian Capital Territory, Australia. Int. J. Climatol. 37, R Development Core Team, 2010. R: A Language and Environment for Statistical Comput-
1195–1210. ing. R Foundation for Statistical Computing, Vienna, Austria 3-900051-07-0 http://
Guerreiro, S.B., Dawson, R.J., Kilsby, C., Lewis, E., Ford, A., 2018. Future heat-waves, www.R-project.org, Accessed date: 6 July 2011.
droughts and floods in 571 European cities. Environ. Res. Lett. 13. Raddatz, T.J., Reick, C.H., Knorr, W., Kattge, J., Roeckner, E., Schnur, R., et al., 2007. Will the
Hansen, J., Sato, M., Ruedy, R., 2012. Perception of climate change. Proc. Natl. Acad. Sci. U. tropical land biosphere dominate the climate-carbon cycle feedback during the
S. A. 109, E2415–E2423. twenty-first century? Clim. Dyn. 29, 565–574.
Heavens, N., Ward, D., Natalie, M., 2013. Studying and projecting climate change with Ribalaygua, J., Rosa Pino, M., Portoles, J., Roldan, E., Gaitan, E., Chinarro, D., et al., 2013a.
earth System Models. Nat. Educ. Knowl. 4 (5), 4. Climate change scenarios for temperature and precipitation in Aragon (Spain). Sci.
Hertig, E., Seubert, S., Jacobeit, J., 2010. Temperature extremes in the Mediterranean area: Total Environ. 463, 1015–1030.
trends in the past and assessments for the future. Nat. Hazards Earth Syst. Sci. 10, Ribalaygua, J., Torres, L., Portoles, J., Monjo, R., Gaitan, E., Pino, M.R., 2013b. Description
2039–2050. and validation of a two-step analogue/regression downscaling method. Theor. Appl.
Hervada-Sala, C., Pawlowsky-Glahn, V., Jarauta-Bragulat, E., 2000. A statistical method to Climatol. 114, 253–269.
downscale temperature forecasts. A case study in Catalonia. Meteorol. Appl. 7. Robine, J.M., SLK, Cheung, Le Roy, S., Van Oyen, H., Griffiths, C., Michel, J.P., et al., 2008.
Imbert, A., Benestad, R.E., 2005. An improvement of analog model strategy for more reli- Death toll exceeded 70,000 in Europe during the summer of 2003. C. R. Biol. 331,
able local climate change scenarios. Theor. Appl. Climatol. 82, 245–255. 171-U5.
IPCC, 2000. Special Report on Emissions Scenarios. Cambridge University Press, Cam- Roldan, E., Gomez, M., Pino, M.R., Diaz, J., 2015. The impact of extremely high tempera-
bridge, UK 0 521 80493 0. tures on mortality and mortality cost. Int. J. Environ. Health Res. 25, 277–287.
IPCC, 2001. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Roldan, E., Gomez, M., Pino, M.R., Portoles, J., Linares, C., Diaz, J., 2016. The effect of
Maskell, K., Johnson, C.A. (Eds.), Climate Change 2001: The Scientific Basis. Contribution climate-change-related heat waves on mortality in Spain: uncertainties in health
of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on on a local scale. Stoch. Env. Res. Risk A. 30, 831–839.
Climate Change. Cambridge University Press, Cambridge, UK, and New York, USA. Russo, S., Sillmann, J., Sterl, A., 2017. Humid heat waves at different warming levels. Sci.
IPCC, 2013. In: Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Rep. 7.
Nauels, A., Xia, Y., Bex, V., Midgley, P.M. (Eds.), Climate Change 2013: The Physical Saeed, F., Almazroui, M., Islam, N., Khan, M.S., 2017. Intensification of future heat waves in
Science Basis. Contribution of Working Group I to the Fifth Assessment Report of Pakistan: a study using CORDEX regional climate models ensemble. Nat. Hazards 87,
the Intergovernmental Panelon Climate Change. Cambridge University Press, Cam- 1635–1647.
bridge https://doi.org/10.1017/CBO9781107415324. Schoetter, R., Cattiaux, J., Douville, H., 2015. Changes of western European heat wave
Iversen, T., Bentsen, M., Bethke, I., Debernard, J.B., Kirkevag, A., Seland, O., et al., 2013. The characteristics projected by the CMIP5 ensemble. Clim. Dyn. 45, 1601–1616.
Norwegian Earth System Model, NorESM1-M - part 2: climate response and scenario Semenov, M.A., Stratonovitch, P., 2010. Use of multi-model ensembles from global climate
projections. Geosci. Model Dev. 6, 389–415. models for assessment of climate change impacts. Clim. Res. 41, 1–14.
Jeong, D.I., Sushama, L., Diro, G.T., Khaliq, M.N., Beltrami, H., Caya, D., 2016. Projected Seubert, S., Fernandez-Montes, S., Philipp, A., Hertig, E., Jacobeit, J., Vogt, G., et al., 2014.
changes to high temperature events for Canada based on a regional climate model Mediterranean climate extremes in synoptic downscaling assessments. Theor. Appl.
ensemble. Clim. Dyn. 46, 3163–3180. Climatol. 117, 257–275.
 E. Gaitán et al. / Science of the Total Environment 650 (2019) 2778–2795 2795

Sillmann, J., Donat, M.G., Fyfe, J.C., Zwiers, F.W., 2014. Observed and simulated tempera- Hadley Centre, UK , p. 68 160 pp. http://ensembles-eu.metoffice.com/docs/Ensem-
ture extremes during the recent warming hiatus. Environ. Res. Lett. 9. bles_final_report_Nov09.pdf, Accessed date: 10 February 2012.
Smith, T.T., Zaitchik, B.F., Gohlke, J.M., 2013. Heat waves in the United States: definitions, Vautard, R., Gobiet, A., Jacob, D., Belda, M., Colette, A., Deque, M., et al., 2013. The simula-
patterns and trends. Clim. Chang. 118, 811–825. tion of European heat waves from an ensemble of regional climate models within the
Soares, P.M.M., Cardoso, R.M., Miranda, P.M.A., de Medeiros, J., Belo-Pereira, M., Espirito- EURO-CORDEX project. Clim. Dyn. 41, 2555–2575.
Santo, F., 2012. WRF high resolution dynamical downscaling of ERA-interim for Voldoire, A., Sanchez-Gomez, E., Melia, D.S.Y., Decharme, B., Cassou, C., Senesi, S., et al.,
Portugal. Clim. Dyn. 39, 2497–2522. 2013. The CNRM-CM5.1 global climate model: description and basic evaluation.
Stegehuis, A.I., Vautard, R., Ciais, P., Teuling, A.J., Miralles, D.G., Wild, M., 2015. An Clim. Dyn. 40, 2091–2121.
observation-constrained multi-physics WRF ensemble for simulating European Watanabe, S., Hajima, T., Sudo, K., Nagashima, T., Takemura, T., Okajima, H., et al., 2011.
mega heat waves. Geosci. Model Dev. 8, 2285–2298. MIROC-ESM 2010: model description and basic results of CMIP5-20c3m experiments.
Steul, K.S., Latasch, L., Jung, H.G., Heudorf, U., 2018. Health impact of the heatwave of 2015: Geosci. Model Dev. 4, 845–872.
hospital admissions in Frankfurt/Main, Germany. Gesundheitswesen 80, 353–359. World Meteorological Organization, 2010. No. 1055. 978-92-63-11055-8 (Switzerland).
Suparta, W., Yatim, A.N.M., 2017. An analysis of heat wave trends using heat index in East www.wmo.int.
Malaysia. In: Suparta, W., Ismail, M., Abdullah, M. (Eds.), 2017 International Confer- Xiao-Ge, X., Tong-Wen, W., Jie, Z., 2013. Introduction of CMIP5 experiments carried out
ence on Space Science and Communication. 852. with the climate system models of Beijing Climate Center. Adv. Clim. Chang. Res. 4,
Taylor, K.E., Stouffer, R.J., Meehl, G.A., 2012. An overview of CMIP5 and the experiment de- 41–49. https://doi.org/10.3724/SP.J.1248.2013.041.
sign. Bull. Am. Meteorol. Soc. 93, 485–498. Yukimoto, S., Yoshimura, H., Hosaka, M., Sakami, T., Tsujino, H., Hirabara, M., Tanaka, T.Y.,
Tomozeiu, R., Cacciamani, C., Pavan, V., Morgillo, A., Busuioc, A., 2007. Climate change sce- Deushi, M., Obata, A., Nakano, H., Adachi, Y., Shindo, E., Yabu, S., Ose, T., Kitoh, A.,
narios for surface temperature in Emilia-Romagna (Italy) obtained using statistical 2011. Meteorological research institute- earth system model v1 (MRI-ESM1)—
downscaling models. Theor. Appl. Climatol. 90, 25–47. model description. Technical Report of MRI. vol 64.
Turco, M., Marcos, R., Quintana-Segui, P., Llasat, M.C., 2014. Testing instrumental and Zittis, G., Hadjinicolaou, P., Fnais, M., Lelieveld, J., 2016. Projected changes in heat wave
downscaled reanalysis time series for temperature trends in NE of Spain in the last characteristics in the eastern Mediterranean and the Middle East. Reg. Environ.
century. Reg. Environ. Chang. 14, 1811–1823. Chang. 16, 1863–1876.
Uppala, S.M., Kallberg, P.W., Simmons, A.J., Andrae, U., Bechtold, V.D., Fiorino, M., et al., Zorita, E., von Storch, H., 1999. The analog method as a simple statistical downscaling
2005. The ERA-40 re-analysis. Q. J. R. Meteorol. Soc. 131, 2961–3012. technique: comparison with more complicated methods. J. Clim. 12, 2474–2489.
Van der Linden, P., JFB, Mitchell (Eds.), 2009. ENSEMBLES: Climate Change and Its Im-
pacts: Summary of Research and Results From the ENSEMBLES Project. Met Office
 Science of the Total Environment 740 (2020) 140094

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Impact of climate change on drought in Aragon (NE Spain)

Emma Gaitán a,b, Robert Monjo b, Javier Pórtoles b, María Rosa Pino-Otín a,⁎
a
Universidad San Jorge, Villanueva de Gállego, 50830, Zaragoza, Spain
b
Fundación para la Investigación del Clima, C/ Tremps 11, 28040 Madrid, Spain

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Future SPI and SPEI scenarios for Aragon

(Spain) were downscaled.
• Scenarios were based on the most re-
cent Earth System Models (CMPI5) at
first time.
• The SPI, considering only precipitation,
shows no changes in the water balance.
• The SPEI, considering the global
warming, shows an increase of the
drought episodes.
• Hydroclimatic conditions of Aragon will
change towards a drier climate.

a r t i c l e i n f o a b s t r a c t

Article history: Droughts are one of the extreme climatic phenomena with the greatest and most persistent impact on health,
Received 12 December 2019 economic activities and ecosystems and are poorly understood due to their complexity. The exacerbation of
Received in revised form 4 June 2020 global warming throughout this century probably will cause an increase in droughts, so accurate studies of future
Accepted 7 June 2020
projections at a local level, not done so far, are essential.
Available online 10 June 2020
Climate change scenarios of drought indexes for the region of Aragon (Spain) based on nine Earth System Models
Editor: Martin Drews (ESMs) and two Representative Concentration Pathways (RCPs) corresponding to the fifth phase of the Coupled
Model Intercomparison Project (CMIP5) have been generated for the first time. Meteorological Drought episodes
were analysed from three main aspects: magnitude (index values), duration and spatial extent. The evolution of
drought is also represented in a novel way, allowing identification, simultaneously, of the intensity of the epi-
sodes as well as their duration in different periods of accumulation and, for the first time, at the observatory level.
Future meteorological drought scenarios based on the Standardized Precipitation Index (SPI) hardly show vari-
ations in water balance with respect to normal values. However, the Standardized Precipitation Evapotranspira-
tion Index (SPEI) which, in addition to precipitation, considers evapotranspiration, shows a clear trend towards
increasingly intense periods of drought, especially when considering cumulative periods and those at the end of
the century.
Representation of the territory of the drought indexes reflects that the most populated areas (Ebro Valley and SW
of the region), will suffer the longest and most intense drought episodes. These results are key in the develop-
ment of specific measures for adapting to climate change.
© 2020 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Universidad San Jorge, Campus Universitario Villanueva de Gállego Autovía A-23 Zaragoza-Huesca, km. 510, 50830 Villanueva de Gállego, Zaragoza, Spain.
E-mail addresses: emma@ficlima.org (E. Gaitán), robert@ficlima.org (R. Monjo), javier@ficlima.org (J. Pórtoles), [email protected], URL: https://www.usj.es (M.R. Pino-Otín).

https://doi.org/10.1016/j.scitotenv.2020.140094
0048-9697/© 2020 Elsevier B.V. All rights reserved.
2 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

1. Introduction To avoid this problem, (Vicente-Serrano et al., 2010a) proposed a

new climatic drought index, SPEI, which considers the difference be-
Drought is probably one of the extreme climatic phenomena with tween monthly precipitation and AED. Thus, SPEI best reflects climate
the greatest impact on the world's population and that can affect mil- change as it makes a more realistic measurement of water availability
lions of people every year around the planet (Bryant, 1993; Wilhite, by incorporating the effect of temperature on changes in evaporation
2000). It also has serious effects on the availability of water and there- demand as does PDSI. On the other hand, it maintains the multi-
fore on economic activities such as agriculture (Lesk et al., 2016) and temporal nature and simplicity of SPI (Marcos-Garcia et al., 2017).
tourism and profound impacts on human health (Stanke et al., 2013) According to the latest report of the Intergovernmental Panel on Cli-
and ecosystems (Alary et al., 2014) that may persist over time. (Dai, mate Change (IPCC, 2014), analysis of the precipitation regime (Calbo,
2011). However, drought is a phenomenon that is not well understood 2010; Lavaysse et al., 2012), droughts (Burke and Brown, 2008; Lopez-
due to its complexity and lack of historical records (Wilhite, 2000) and Bustins et al., 2013) and the extreme temperatures that drastically in-
because it depends on numerous factors. crease evapotranspiration (ET) (Rebetez et al., 2006) and decrease soil
For this reason, the scientific community and institutions are putting moisture (Sheffield and Wood, 2008) suggest that drought episodes
a lot of effort into understanding, identifying, documenting and moni- could become more severe around the world in the 21st century (Dai,
toring this phenomenon more exhaustively. Examples are the drought 2013).
databases of the European Drought Observatory, the National Drought There are some studies that emphasize that future projections of
Mitigation Center and, the historical database of the Standardized Pre- drought may overestimate drought episodes if future soil moisture
cipitation Evapotranspiration Index (SPEI) (http://spei.csic.es/ (Berg et al., 2017) and runoff (Yang et al., 2018) simulations are not
database.html). taken into account (Berg and Sheffield, 2018), which can affect AED. In
addition, recent studies highlight the need to include CO2 concentration
1.1. Droughts types and indexes in the analysis of AED under climate conditions since an increase of the
CO2 acts contributing to the increase in temperatures that in turn affect
Precipitation is the primary controlling factor of drought but other the Vapour-Pressure Deficit (VPD). On the other hand, CO2 could in-
meteorological phenomena, such as temperature (Cook et al., 2014; crease water use efficiency by plants reducing AED and therefore miti-
Hao et al., 2017; Livneh and Hoerling, 2016), wind (McVicar et al., gate the drying (Dai et al., 2018; Roderick et al., 2015).
2012a) and relative humidity (Willett et al., 2014), can modulate its in- In this context, most European areas and the Mediterranean region
tensity (Bates et al., 2008). Through potential evapotranspiration (PET), seem to be prominent regional climate change hotspots where an incre-
it is possible to evaluate the amount of water that would evaporate and ment in the occurrence of extreme events is expected (Beniston et al.,
transpire if there was enough water available, which is very important 2007; Skaugen et al., 2004). Specifically, a possible rise in the intensity
in the evaluation of meteorological droughts. and frequency of extreme drought events is expected (Forzieri et al.,
Because drought affects so many different aspects (environmental, 2014; Hoerling et al., 2012; Iglesias et al., 2007; Marcos-Garcia et al.,
economic, social, health), a single ‘drought’ does not really exist. 2017; Paparrizos et al., 2018), especially in the summer months
Drought is often classified into four types (Wilhite, 2000; Wilhite and (Vicente-Serrano et al., 2010c), and will have significant environmental,
Glantz, 1985): meteorological, agricultural, hydrological and socioeco- social and economic impacts (Blenkinsop and Fowler, 2007).
nomic drought. The global climate models used today reproduce temperature trends
The main subject of the current study is meteorological drought, a very well, but the level of precision for large-scale precipitation patterns
type of drought characterized by below-normal precipitation over a pe- is lower than for temperature (IPCC, 2014). This has caused the climatic
riod of months to years and that should be defined as a condition rela- projections of droughts to show great uncertainty and therefore we can-
tive to the normal local condition (Dai, 2011; Paparrizos et al., 2018; not know with precision the effects of climatic change on drought sever-
Wilhite, 2000). ity at the regional level in the future (Burke and Brown, 2008). This is
On the other hand, to characterize droughts, standardized drought especially problematic in areas with high precipitation variability, such
indexes are used in the literature. These indexes are direct indicators as the Mediterranean region, where the drought patterns derived
based on climate information, defined so that the results are compara- from the results of global climate models are not consistent (Vicente-
ble in time and space since droughts of the same magnitude can have Serrano et al., 2004).
very different effects depending on the time of year and the place
where they occur (Hayes et al., 1999; Vicente-Serrano, 2016; Wilhite,
2000). 1.2. Drought in NE Spain (Aragón)
Some of these indexes are well established and have been used to
monitor climatic conditions across different locations; these include In Spain, as in the rest of Europe (Feyen and Dankers, 2009), differ-
the Palmer Drought Severity Index (PDSI; Palmer, 1965) and Standard- ent series of major droughts have been happening in recent decades.
ized Precipitation Index (SPI; Mckee et al., 1993), for example. The Lin- In addition, the literature seems to indicate a trend towards an increase
coln Declaration on Drought Indexes (Hayes et al., 2011) determined in meteorological water scarcity in the Iberian Peninsula, either due to
that SPI is the only index, from the point of view of meteorological an increase in the frequency of drought episodes or due to a change in
drought, valid for any region of the world and any time scale, being the precipitation regime (Fragoso et al., 2018; Gallego et al., 2011;
one of the most used in Europe (Spinoni et al., 2015). It is able to provide Garcia-Barron et al., 2011; Machado et al., 2011; Ojeda et al., 2017;
better spatial standardization than PDSI (Lloyd-Hughes and Saunders, Vicente-Serrano et al., 2004). This makes necessary studies at a local
2002) and indicate drought initiation and termination because they level and the development of future scenarios of droughts which are ad-
are implicit parts of the index (Sonmez et al., 2005). equate as possible for evaluating the local impacts of climate change.
SPI, however, presents some limitations such as that it neglects the Drought scenarios in Spain are also scarce: either they are from stud-
effect of temperature increase and, therefore, the effect that an increase ies conducted prior to the Fifth Assessment Report of the United Nations
in PET (Vicente-Serrano et al., 2010a) or in the atmospheric evaporative Intergovernmental Panel on Climate Change (IPCC5) and in very small
demand (AED) (Vicente-Serrano et al., 2020) can have on droughts, areas (Lopez-Bustins et al., 2013) or they use IPCC5 models but use dy-
which may affect prediction of the impact of global warming in future namic downscaling information from the European Coordinated Re-
drought conditions. It should be noted, however, that other meteorolog- gional Downscaling Experiment EUROCORDEX; (Collados-Lara et al.,
ical variables as wind speed, solar radiation and air humidity, can also 2018; Marcos-Garcia et al., 2017). The latter also evaluates only SPI
affect PET changes linked to climate change. and SPEI at the 12-month scale. However, these studies agree that the
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 3

combined use of SPI and SPEI is adequate for studying drought episodes wind (“Cierzo”) characteristic of the Ebro Valley (López et al., 2007).
in the future (Lopez-Bustins et al., 2013; Marcos-Garcia et al., 2017). Therefore, 70% of the Aragonese territory is considered semi-arid
The combined study of both indexes, SPEI and SPI could be an effec- (index value proposed by the United Nations Environment Program
tive formula for an adequate study of meteorological drought in terri- b0.5 and even 30% presents values of 0.3) (Cherlet et al., 2018).
tories with the climatology of Aragon (NE Iberian Peninsula). This
region of Spain is characterized by a continental Mediterranean climate 2.2. Datasets
with high precipitation variability and marked by very diverse orogra-
phy throughout its territory that includes areas of high mountains, val- 2.2.1. Surface observation datasets
leys and steppes (López et al., 2007). In addition, we must consider that In this study, an observational dataset (daily maximum and mini-
previous studies in Aragon have shown that an increase in temperature mum temperature and precipitation) belonging to the extensive net-
is one of the variables that will be most noticeable with climate change work of instrumental observatories owned by the Spanish
throughout this century (Gaitan et al., 2019; Ribalaygua et al., 2013a). Meteorological Agency (AEMET) (http://www.aemet.es) was used
As far as we know, drought scenarios in Aragon have not been ob- (Fig. 1). This dataset is the same as the one used in previous studies
tained to date. As has been seen, it is essential to have local scenarios (Gaitan et al., 2019; Ribalaygua et al., 2013a) in order to work with a
to determine the impact of climate change on the environmental or so- set of data that has been subjected to strict quality control (inhomoge-
cioeconomic reality of each region in order to make decisions on adap- neities, gaps, outliers, transcription errors and so on) carried out first
tation to climate change. by the Government of Aragon (López et al., 2007) and completed, in a
The goal of this study is to obtain, for the first time, meteorological second phase, by (Ribalaygua et al., 2013a). As a complement to quality
drought scenarios for Aragon (located in NE of Spain) for the 21st cen- controls, those stations with a large number of data gaps or b15 years of
tury using a statistical methodology to downscale GCMs from CMIP5. daily records were discarded.
To achieve this goal, the capacity of the GCMs to simulate the past For the simulation of future climate scenarios of precipitation, a first
observed climate was assessed (validation) and using CMIP5, precipita- set of 263 stations was used (red dots in Fig. 1a). Of these 263 stations,
tion scenarios for Aragon were generated to simulate future daily just those with data for both variables, temperature and precipitation,
precipitation. were used for the simulation of drought indexes (43 stations, Fig. 1b).
Finally, as Aragon is a region sensitive to episodes of drought caused
by a varying rate of precipitation and high temperatures, the SPI and 2.2.2. Atmospheric dataset
SPEI meteorological indexes were calculated and the frequency of oc- A set of nine climate models were selected from CMIP5, supplied by
currence of drought and its spatial distribution were simulated to iden- the Program for Climate Model Diagnosis and Intercomparison (PCMDI)
tify the drought vulnerability of the study area. Drought indexes were archives.
also verified. The global climate models called Earth System Models (ESMs) from
This study provides, for the first time, scenarios of meteorological the fifth phase of the Coupled Model Intercomparison Project (CMIP5)
drought in the NE of Spain according to CMIP5 models, useful for (Tripathi et al., 2006) have contributed to the acquisition of both dy-
predicting the impacts of climate change on the availability of water at namic and statistical downscaling techniques with less uncertainty.
a local scale and which are necessary for stakeholders to make decisions These models integrate the individual parts of the climate system (at-
on adaptation and mitigation of climate change. On the other hand, this mosphere, ocean, land and sea ice) and the exchange of energy and
region of Spain is a good indicator of many characteristic areas of south- mass between them (Knutti and Sedlacek, 2013). These models also in-
ern Europe (high mountains, river basins, steppes, etc.). clude chemical processes, land use, plant and ocean ecology and an in-
teractive carbon cycle, which enables integration of biochemical
2. Data and methodology processes into the models (Heavens et al., 2013), constituting a robust
set of coordinated climate model experiments (Carvalho et al., 2017;
2.1. Study area Chen et al., 2016; Perez et al., 2014).
The climate models (Table 1) were selected according to the time
The present study was carried out in the region of Aragon (NE of resolution (daily) of available predictor fields, because it is required
Spain) (Fig. 1). Because of its location, Aragon falls within the Western for the downscaling method used. All of the models were ESMs (Jones
Mediterranean climate area characterized by scarce precipitation with et al., 2011; Wang et al., 2009).
cool winters and hot, dry summers. Differences in latitude between This study used data from two different experiment families of
the most northern and most southern points of Aragon (340 km length GCMs: the Historical experiment (Taylor et al., 2012), which covers
and 240 km width) along with the influence of the Cantabrian and Med- much of the industrial period and can be referred to as ‘twentieth-cen-
iterranean Seas and the general atmospheric circulation as well as the tury’ simulations, and the Representative Concentration Pathway (RCP)
orographic complexity of the region (extreme altitude differences of family (Moss et al., 2010), which corresponds to different possible
over 3000 m between the plains (the Ebro River valley) and the moun- ranges of radiative forcing reached in the year 2100 with respect to
tains (the Pyrenees)), give rise to great subclimate variety, with differ- values of the pre-industrial era. This study used future projections de-
ent thermal and pluviometric regimes that condition the local climate termined by the RCP8.5 ‘high’ scenario and RCP4.5 ‘intermediate’ sce-
(López et al., 2007). nario, the core of IPCC5 experiments.
Precipitation is scarce in most of Aragon and is distributed clearly ac- In order to study the behaviour of the CMIP5 model Historical simu-
cording to relief, as the isohyets are arranged in concentric circles de- lations, we used the reanalysis dataset from the European Centre for
creasing from mountain areas to the centre of the region. Although Medium-Range Weather Forecasts (ECMWF ERA-40; http://www.
the average annual total precipitation of the Aragonese territory is ecmwf.int/research/era/do/get/) (Uppala et al., 2005) for the period
around 550 mm, there are regions for which the average is below 1958–2000 at 6-hourly time resolution and 125 km spatial resolution.
these values (for example, in the central sector of the Ebro Depression). For verification of the methodology, it was necessary to reduce the tem-
Only in the Pyrenees and, to a lesser extent, in the Iberian Mountain poral and spatial scale of the reanalysis in order to compare both ERA-40
Range, does precipitation reach important values, 1800–2000 mm, and the climate model simulations (Ribalaygua et al., 2013a; Ribalaygua
and show positive water balance values (considering the difference be- et al., 2013b). The geographical limits of the atmospheric window used
tween precipitation and AED). On the other hand, N60% of the region has were latitudes 31.5°N to 55.1°N and longitudes 27.0°W to 14.6°E, cover-
average values of AED above 1100 mm, showing a negative water bal- ing not only the geographic area under study but also the surrounding
ance, to which contributes not only the scarce rainfall but also the strong atmosphere areas which exert a meteorological influence all over the
4 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Fig. 1. Location of the study area and observatories. Aragon (Spain) in Europe. Points indicate the stations used in the study. a) Stations of precipitation (264) used in the generation of
climate regional scenarios of precipitation (verification, validation and scenarios). b) Stations used exclusively on the generation of drought indexes (43).
Map source: OpenStreetMap.

Iberian Peninsula (Ribalaygua et al., 2013a). The use of the ERA-40 data Ribalaygua et al., 2018; Rodriguez et al., 2014; Santiago et al., 2017).
set has allowed us to compare these new results with those published This methodology offers some advantages: it is computationally inex-
by Ribalaygua et al., 2013a. pensive, provides local information and allows quantifying the uncer-
tainty associated with the downscaling process (Van der Linden and
2.3. Methodologies Mitchell, 2009). Other advantages are the application of future simula-
tions consistent with observations (physically coherent between
2.3.1. Validation and generation of future precipitation scenarios them) and using local scale (because nearby data points in space are
A two-step analogue/regression statistical downscaling method de- not subjected to different climate change conditions) (Ribalaygua
veloped previously (Ribalaygua et al., 2013b) was applied to obtain fu- et al., 2013b).
ture scenarios of precipitation and drought. This method has been used Through the validation process we can, on the one hand, evaluate
in national and international projects, with good verification results the ability of each ESM to simulate the predictor fields (comparing the
(Gaitan et al., 2019; Monjo et al., 2016; Moutahir et al., 2017; downscaled Historical experiment simulation for each model with the

Table 1
Information about the nine climate models belonged to the 5 Coupled Model Intercomparison Project (CMIP5) corresponding to the fifth report of the IPCC. Models were supplied by the
Program for Climate Model Diagnosis and Intercomparison (PCMDI) archives.

GFDL-ESM2M 2°x2,5° National Oceanic and Atmospheric Administration (NOAA), E.E.U.U. Dunne et al. (2012)
daily
CanESM2 2,8°x2,8° Canadian Centre for Climate Modeling and Analysis (CC-CMA), Canadá. Chylek et al. (2011)
daily
CNRM-CM5 1,4°x1,4° CNRM (Centre National de Recherches Meteorologiques), Meteo-France, Francia. Voldoire et al. (2013)
daily
BCC-CSM1-1 1,4°x1,4° Beijing Climate Center (BCC), China Meteorological Administration, China. Xiao-Ge et al. (2013)
daily
HADGEM2-CC 1,87°x1,25° Met Office Hadley Center, United Kingdom. Collins et al. (2008)
daily
MIROC-ESM-CHEM 2,8°x2,8° Japan Agency for marine-Earth Science and Technology (JAMSTEC), Atmosphere and Ocean Research Institute Watanabe et al. (2011)
daily (AORI), and National Institute for Environmental Studies (NIES), Japan.
MPI-ESM-MR 1,8°x1,8° Max-Planck Institute for Meteorology (MPI-M), Germany. Raddatz et al. (2007);
daily Marsland et al. (2003)
MRI-CGCM3 1,2°x1,2° Meteorological Research Institute (MRI), Japan. Yukimoto et al. (2011)
daily
NorESM1-M 2,5°x1,9° Norwegian Climate Centre (NCC), Norway. Bentsen et al. (2013);
daily Iversen et al. (2013)
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 5

downscaled ERA-40 simulation for a common period, 1958–2000) and, recommended within the SPEI package and they also depend only on
on the other, quantify the uncertainties inherent to future climate pro- temperature and precipitation, unlike other more complex methods
jections through an ensemble strategy (Monjo et al., 2016). such as the Penman–Monteith (Smith et al., 1998) and Jensen–Haise
Bias and standard deviation at seasonal scale have been used as error methods (Jensen and Haise, 1963). Both methods only take into account
measures. This validation process presents some limitations related to the temperature, so it is assumed that the calculation of AED trends
the observational data available to be considered in the final uncertainty could have certain limitations (Irmak et al., 2012; McVicar et al.,
analysis. More information about the validation process can be 2012b; Sheffield et al., 2012). For a certain increase in the temperature,
consulted in (Ribalaygua et al., 2013b). the change in the obtained result can be higher than the one really ex-
Future local climate scenarios at local and daily scale for precipita- pected according a complete method like Penman-Monteith. Therefore,
tion were produced for nine ESMs (see Table 1) and two RCPs (RCP4.5 the role of AED on drought severity would be overestimate and this
and RCP8.5) as a previous step to calculate the drought indices. As pre- would have some effect on the drought indices obtained for future
cipitation is an essential variable in the analysis of drought, these sce- scenarios.
narios are a starting point providing initial information on future The way in which the indexes have been analysed follows the guide-
pluviometric conditions. lines of the WMO (WMO, 2017) which recommends analysis of a
The local climatic projections of precipitation belonging to CMIP5 drought episode from three main aspects – magnitude (index values),
were obtained in this study using the same methodology as that used duration (alternation between positive and negative values) and spatial
for temperature scenarios, previously described (Gaitan et al., 2019). extent – and all these aspects configure the severity of the episode.
In order to assess the capacity of the downscaling methodology to
2.3.2. Drought indexes simulate SPI and SPEI, we analysed the intensity and duration of the dif-
SPI was developed by Mckee et al. (1993) and is based on two as- ferent drought episodes shown by both indexes, comparing the SPI and
sumptions: 1) that the variability of precipitation is greater than that SPEI values calculated from the simulated ERA-40 temperature and pre-
of temperature and AED, and 2) that the rest of the variables are station- cipitation series with those obtained from the observed series for a com-
ary over time. The SPI value is defined as a numerical value that repre- mon period (1970–2000). Verification of the maximum and minimum
sents the number of standard deviations of precipitation, over the temperature and precipitation can be seen in a previous study
accumulation period in question, with respect to the average, once the (Ribalaygua et al., 2013a). The statistical measures used in the verifica-
original distribution of precipitation has been transformed into a normal tion processes were the bias, standard deviation and Pearson correla-
distribution (i.e., mean of zero and standard deviation of 1). The SPI tion. The statistical measures were calculated using R computing
values can be interpreted as the number of standard deviations by software (R Development Core Team, 2010).
which the observed anomaly deviates from the long-term mean. From the ESM simulated temperature and precipitation series (nine
SPEI developed by (Vicente-Serrano et al., 2010a) and revisited by ESMs and two RCPs), we determined the drought episodes that are ex-
(Begueria et al., 2014) is a variant of the widespread SPI; it has greater pected in Aragon during the upcoming decades of the 21st century. The
potential as a drought index since it considers the climate balance SPI and SPEI scenarios were compared to a historical period
(through the difference between monthly precipitation and AED). SPEI (1976–2005) to analyse the future changes with respect to the actual
values can be interpreted in the same way as SPI values (number of situation of these extreme events.
standard deviations by which the observed anomaly deviates from the To draw future local climate scenario maps, we used Thin Plate
long-term mean). Spline (TPS) regression from the R package ‘fields’ (Nychka et al., 2015).
Both indexes were calculated using the R package ‘SPEI’ (Version
1.7). The SPI was calculated using Gamma distribution to fit the original 3. Results
precipitation series (Organization WMO, 2012) and the SPEI was calcu-
lated using log-logistic distribution (Vicente-Serrano et al., 2015; 3.1. Validation and precipitation scenarios
Vicente-Serrano and Begueria, 2016). The parameters of these distribu-
tions were obtained by the method of unbiased probabilistic weighted The results of the validation process (comparison between the ERA-
moments (Vicente-Serrano and Begueria, 2016). The scale of SPI and 40 precipitation simulations and the historical precipitation simulations
SPEI values used in the study can be seen in Table 2. for each ESM for a common period (1958–2000)) are shown in Fig. 2, for
The period 1976–2005 was used as a reference period, which repre- both absolute (mm) and relative precipitation (%). The results are vari-
sents the last 30 years of the Historical period. Based on this reference able depending on the model and the seasonal period; however, all the
period, both the SPI and the SPEI were calculated for the period models are able to reproduce the annual cycle of precipitation as well as
2006–2100. The choice of the reference period was made to evaluate the differences between seasonal periods (maximum values in autumn
the future hidroclimatic conditions of the region with respect to the av- and spring, followed by winter and summer). The obtained bias and
erage conditions of the last 30 years of the Historical experiment. standard deviation are less than ±1 mm/day, which in relative terms
To obtain the AED values used in the calculation of SPEI, both the supposes a difference of less than or around ±10% in the worst of the
Hargreaves and Samani (1985) and Thornthwaite (1948), formulas cases.
have been used, denominated SPEI-Har and SPEI-Thor, respectively. In general terms, a big variation in the Aragon precipitation regime is
These formulas were chosen to calculate AED because they are not expected. According to scenario RCP8.5, mean variations in the
amount of precipitation are expected to be around ±10% for all seasons
of the year, except for the summer where no precipitation change is ex-
Table 2 pected. Scenario RCP4.5 shows no precipitation fluctuations throughout
SPEI/SPI Intensities Scale (Vicente-Serrano et al., 2010a). the 21st century with respect to current values (see Support informa-
tion, Figs. S1 to S4).
SPEI/SPI

≥2 Extremely wet
3.2. Generation of future local climate scenarios of drought indexes
1.5 a 2 Severely wet
0.5 a 1.5 Moderately wet
−0.5 a 0.5 Normal values 3.2.1. Verification of drought index simulation
−1.5 ≤ −0.5 Moderately dry To verify the simulation of drought indexes, the first step was to
−1.5 ≤ −2 Severely dry compare the SPI and SPEI values obtained from the observations with
≤−2 Extremely dry
those calculated from the simulated series of ERA-40.
6 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Fig. 2. Validation of precipitation. Comparison between the precipitations obtained using the downscaled Historical data of the global climate models and the downscaled reanalysis data,
for every seasonal period. Two graphs at the top: seasonal comparative between the precipitation simulated using the downscaled Historical data (colour bars) and that of the downscaled
reanalysis data (black lines) for each global climate models (see Table 1) and for the four seasons: winter (December–February; first bar of each group of four), spring (March–May, second
bar), summer (June–August; third bar) and autumn (September–November; four bar). Two graphs at the bottom: relative seasonal differences between the simulated data using the
downscaled Historical data and that of the downscaled reanalysis data. Seasonal precipitation amounts are shown on the left columns and seasonal values of the standard deviation on
the right columns. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3 shows the verification results corresponding to SPI at time 1981–1982, 1989, 1994–1995 and 1998. The same can be seen for wet
scales from 1 month (SPI-1 M) to 12 months (SPI-12 M) for the period episodes (positive SPI) as in, for example, the periods 1976–1977,
1970–2000 (Fig. 3a and b). This process allows the identification of ep- 1988 and 1996–1997.
isodes of deficit or excess precipitation recorded and simulated from Figs. 4 and S5 show the results of the verification process for SPEI
ERA-40. In addition, the number of months in the period 1970–2000 based on SPEI-Har and SPEI-Thor calculations, respectively.
in which SPI values were obtained within different intensity ranges Similar results were obtained for calculation of SPEI based on the
(Table 1) for SPI-1 M, SPI-3 M and SPI-6 M are shown (Fig. 3c, d and e). Hargreaves method (Fig. 4a and b) although in this case the correlation
As can be seen in Fig. 3a and b, the time series of the simulated SPI for obtained between the observed and simulated time series of SPEI is
ERA-40 shows, in an acceptable way, the same values presented by the slightly higher (0.80 for SPEI-1 M, 0.78 for SPEI-3 M, 0.72 for SPEI-6 M
observed SPI, with a correlation of p = 0.75 in the case of SPI-1 M, p = and 0.73 for SPEI-12 M).
0.72 for SPI- 3 M, p = 0.64 for SPI-6 M and p = 0.61 for SPI-12 M. When the Thornthwaite method is used for calculating AED in SPEI
The simulated and observed SPI values show dry episodes (negative (see Fig. S5), the temporal correlations are lower than those obtained
SPI) in similar periods, for example the periods 1970–1972, 1978, with SPEI based on Hargreaves.
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 7

Fig. 3. SPI verification. Results of the verification process for the SPI. a) Time series of the SPI index calculated from observed data at the time-scales from 1 to 12 months for the period
1970–2000, b) time series of the SPI index calculated from downscaled ERA-40 at the time-scales from 1 to 12 months for the period 1970–2000. c), d) and e) number of months
within the 1970–2000 period corresponding to each interval of the SPI intensities scale based on observed data (blue columns) and on downscaled ERA-40 (red columns) for the
time-scales of 1, 3 and 6 months. Average of the all the stations used in the simulations. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)

On the other hand, for about 65–70% of the period considered, water The SPI values obtained are hardly modified with respect to the His-
balance conditions in Aragon was considered normal (SPI/SPEI between torical period so, according to these results, the water balance character-
−0,5 and 0,5), suffering extreme wet or dry episodes for only 2–4% of istics of the region as a whole would remain similar to the current ones.
the period 1970–2000. On the other hand, the SPEI climate change scenarios, considering the
The error (bias) made in the simulation of SPI (Fig. 3c to e) and SPEI effect of AED, show a marked tendency towards increasingly negative
(Fig. 4c to e) is quite small for all of the classes considered (b ± 2 values of the index with respect to the Historical period, especially at
months). the end of the century.
Both RCPs show a similar evolution until 2060, with changes of SPEI
with respect to the Historical period of −0.6 for SPEI-3 M, −0.9 for SPEI-
3.2.2. Local climate scenarios to predict drought indexes 6 M and − 1.3 for SPEI-12 M. For the final period of the century, the var-
Figs. 5 to 9 (complemented with Figs. S6 and S7) show the results iation begins to be more pronounced under the conditions of RCP8.5,
obtained for the simulation of SPI and SPEI throughout the 21st century with changes of −1.2 for SPEI-3 M, −1.8 for SPEI-6 M and − 2.8 for
from different perspectives. SPEI-12 M, while under scenario RCP4.5, SPEI values vary slightly from
Fig. 5 shows local climate change scenarios for future SPEI (a, c and those reached in 2060.
d) and SPI (b, d and f) at 3-, 6- and 12-month scale, which have been These results are very well reflected in the time-scale evolution
predicted on the basis of the nine models (see Table 2) and here are maps, where the simulated time series from 1976 to 2100 are repre-
used to obtain a general vision of the changes in water balance for the sented, both for SPEI (Fig. 6) and SPI (Fig. 7) and under both scenarios,
Aragon region as a whole. The future projections of SPI and SPEI for RCP4.5 (Figs. 6a and 7a) and RCP8.5 (Figs. 6b and 7b) with respect to dif-
the period 2006–2100 have been made based on the reference period ferent time scales (from 1 to 12 months).
(Historical 1976–2005). When working with normalized indexes, the Fig. 6 shows a tendency towards more and more extreme SPEI
future values of the SPI and SPEI represent anomalies with respect to values, especially in the longer time scales. For time scales of up to 4
the average values of the reference period, which allows to evaluate months, an alternation between periods considered normal and dry pe-
the future evolution of the hydric conditions in Aragon with respect to riods is expected (SPEI values between −1,5 and 0,5). For longer time
the average of the last 30 years of the Historical experiment. scales, there is a tendency towards more intense and prolonged periods
8 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Fig. 4. SPEI Har verification. Results of the verification process for the SPEI based on Hargraves Evapotranspiration. a) Time series of the SPEI index calculated from observed data at the
time-scales from 1 to 12 months for the period 1970–2000, b) time series of the SPEI index calculated from downscaled ERA-40 at the time-scales from 1 to 12 months for the period
1970–2000. c), d) and e) number of months within the 1970–2000 period corresponding to each interval of the SPI intensities scale based on observed data (blue columns) and on
downscaled ERA-40 (red columns) for the time-scales of 1, 3 and 6 months. Average of the all the stations used in the simulations. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

of drought, with SPEI values of up to −3 at the end of the century. The according to RCP4.5 and considerably more intense under RCP8.5 with
pattern obtained is similar under both RCPs, being more pronounced values from −1.8 at 3 months to −4 at 12 months. The north-west
in the case of RCP8.5. area of the region, which is expected to be most affected by drought ep-
In the time-scale map corresponding to SPI (Fig. 7), the same pattern isodes, deserves special attention. The Pyrenees zone is the one that will
as that obtained for SPEI is not appreciated; in this case, alternating dry clearly suffer the fewest expected drought episodes; under RCP4.5 it is
and wet periods are observed for all time scales, these being somewhat expected to remain in normal water balance conditions while under
more extensive as we move along the time scales. The same pattern is RCP8.5, at most, SPEI will reach values of −1.5 (at the end-century
observed under both RCPs, the signal being slightly stronger in the and at 12-month time scale).
case of RCP8.5. The SPI spatial maps (Figs. S6 and S7) show how the region will re-
As a complement to the previous results, which allowed the extrac- main under normal water balance conditions, highlighting the Ebro
tion of results for the water regime of Aragon as a whole, the spatial basin at the end of the 21st century and under RCP8.5, where more neg-
maps of both indexes are shown. Figs. 8 and 9 show the climate scenar- ative values of SPI (around −1) are appreciated, but which are still
ios for mean SPEI according to RCP4.5 and RCP8.5, respectively. These within the range considered normal for the region.
figures show the temporal evolution for four time scales: 1 month It is important to emphasize that if the average value of the SPEI/SPI
(first row), 3 months (second row), 6 months (third row) and 12 tends to increasingly negative values and if this is a constant trend in the
months (fourth row). The temporal periods chosen were 2011–2040 future, the conditions considered normal today will evolve towards new
(present), 2041–2070 (mid-century) and 2071–2100 (end-century). values considered normal (Vicente-Serrano et al., 2020).
Figs. S6 and S7 show the same information but for mean SPI. This study has been carried out for each of the observatories used in
The results for SPEI vary considerably between different points in the the study and for each of the climatic models, which reveals that the en-
Aragon region. Coinciding with what was said before, it is observed how tire region is going to be affected by episodes of drought despite its lo-
the SPEI values become more extreme as the 21st century and time cation and height. As an example, the temporal evolution of both
scales advance. The Ebro Valley area is the one that will be subject to indexes obtained according to the MPI-ESM-MR climate model and
more intense episodes of precipitation shortage at the end of the 21st under both RCPs is shown for the observatories of Zaragoza (Figs. 10
century, with SPEI values from −1 at 3 months to −2 at 12 months and 11) and Cedrillas-Huesca (Figs. S8 and S9).
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 9

Fig. 5. Simulated SPEI and SPI for the twenty-first century. Values are displayed as absolute increase compared to the amount simulated for the 1976–2005 Historical period for the time
scales 3 months (a and b), 6 months (c and d) and 12 months (e and f). The vertical dotted line marks the end of the Historical data (2005). Data grouped for every RCP simulation of every
global climate model selected and for the last 30 years of every station. The ensemble median (solid lines) and the 10th–90th percentile (shaded areas) values are displayed.

The choice of these observatories was based on the Climate Atlas of positive SPI for the Cedrillas-Huesca observatory are more intense and
Aragon (López et al., 2007), since they are two of the reference points prolonged than those predicted for Zaragoza.
used in the climatic characterization of the region. The choice of these
observatories was also made based on their location; the Zaragoza ob-
servatory is located in the Zaragoza airport station at a height of 263 4. Discussion
m while the Cedrillas-Huesca observatory is located in the northern
area of the region at a height of 1347 m. In addition, the Zaragoza airport These results offer the possibility of having future climate projec-
station is considered representative of the variability of temperatures in tions based on recently updated data, allowing the evaluation of how
Aragon (Roldán et al., 2011). drought could affect the region of Aragon, both spatially and temporar-
The expected temporal evolution of SPEI throughout the 21st cen- ily, and can be taken as a reference to analyse its impact on multiple sec-
tury is consistent with that explained above, but as it is a single climatic tors. Temporally, drought increases to the end of the century; at the
model and uses a single observatory, the alternation between wet and territory level, the area most affected will be the central area of the
dry periods can be seen more clearly at a time scale of 1 to 3 months. Ebro Valley, where most of the population in the area is concentrated.
Also, as we move forward in the time scales, this alternation softens, The difficulty of developing impact studies and quantifying their
resulting in periods of more intense and prolonged precipitation short- damage as a result of periods of water scarcity comes mainly from the
age while, for SPI, the alternation between wet and dry periods is ob- lack of observed values and studies at a local level with future projec-
served for all time scales. This highlights, as for SPI, how periods with tions, hence the need to publish studies of these characteristics.
10 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Fig. 6. SPEI Time series under RCP4.5 and RCP8.5 along the 21st century at time-scales from 1 to 12 months. Data grouped for every RCP simulation of every global climate model and for
every station. Both emissions scenarios are represented: RCP4.5 (figure a) and RCP8.5 (figure b).

In this study, climate change scenarios of drought indexes for the re- the results of future climate scenarios for Zaragoza (representative ob-
gion of Aragon, Spain, based on nine ESMs corresponding to CMIP5 have servatory of Aragon, Roldán et al., 2011) and Cedrillas-Huesca (Support
been generated for the first time. information) are shown.
The evolution of two indexes, SPI and SPEI, has been obtained
throughout this century and also over the territory, which has allowed
us to observe that while SPI, which only considers precipitation, 4.1. Precipitation scenarios used for the simulation of drought indexes
shows few changes, SPEI, that considers temperature and incorporates
the effects of AED, shows a tendency towards periods of increasingly in- For the simulation of precipitation, ESMs have been used instead of
tense drought, especially when considering accumulated periods of lon- climatic models. ESMs are the most powerful climatic models to date
ger duration and those at the end of the century. Therefore, in the and incorporate significant improvements (Flato et al., 2014) that
current climate change context it is essential to take into account the ef- allow better accuracy in climate simulation, as can be seen in the good
fect of temperature in the study of droughts. results obtained in the validation process.
Figs. 6 and 7 represent a novel representation of the evolution of Validation of the ESMs has shown good results for simulating precip-
drought, allowing identification, simultaneously, of the intensity of the itation. Both the obtained bias and standard deviation are less than ±1
episodes and their duration in different periods of accumulation. mm/day, which in relative terms supposes differences of less than or
One of the strengths of this study is the use of local climate scenarios around ±10% in the worst cases; however, those values are within the
(at the observatory level) to generate future drought indexes. Having order of natural variability of precipitation. These results are better
this information will facilitate decision-making in the face of expected than those obtained for the generation of scenarios of the fourth IPCC re-
changes based on what is expected to occur at each observatory and port published by (Ribalaygua et al., 2013a) particularly in the summer
not in the region a whole. As an example of the study at local level, months, a particularly critical time in Aragon.
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 11

Fig. 7. SPI Time series under RCP4.5 and RCP8.5 along the 21st century at time-scales from 1 to 12 months. Data grouped for every RCP simulation of every global climate model and for
every station. Both emissions scenarios are represented: RCP4.5 (figure a) and RCP8.5 (figure b).

The results obtained for the processes of verification of the method- (Gaitan et al., 2019), SPEI has been chosen; its formulation is similar
ology (Ribalaygua et al., 2013a) and validation of the ESMs are good to that of SPI and allows the comparison of both indexes and evaluation
enough to allow the use of local climatic scenarios generated under of the future behaviour of drought episodes considering the effects of fu-
these conditions in impact studies and analysis of extreme episodes ture temperature changes. Both indexes have been verified and used
such as periods of precipitation shortage. previously in Aragon (Vicente-Serrano et al., 2010a). We have only
Future precipitation scenarios show, under RCP8.5 conditions, a used the temperature in the calculation of AED because the absence of
slight decrease in precipitation throughout the 21st century for all sea- observed historical data of variables such as radiation or humidity
sons of the year, except for the summer months where there is hardly does not allow us a correct validation process of certain indices such
any variation compared to current values of precipitation in the region. as Penman that include these variables.
Under RCP4.5 conditions, less pessimistic than the previous one, barely
any precipitation changes are expected at any time of the year.
These results are consistent with those published by AEMET (www. 4.2.1. Verification results
aemet.es) and directly by the IPCC (Mukherjee et al., 2018), although In general, the results of the verification process show good correla-
the latter show the direct outputs of the ESMs and do not carry the tions between the observed and simulated time series for both indexes
added value of applying downscaling techniques. for the period 1970–2000, higher ones being obtained for SPEI. This is
consistent with the results published by Vicente-Serrano (2013); they
obtained higher correlations for the calculation of SPEI than SPI, espe-
4.2. Consideration of the simulation of drought indexes cially for the summer months, which are the most critical in the region
of Aragon.
SPI is considered by experts in this field as one of the few indexes ap- The temporal series based on observations are satisfactorily rep-
plicable in any region of the world for any time scale (Hayes et al., 2011) resented by the temporal series based on simulations, recreating al-
and with multiple advantages of application compared to other indexes most all dry and wet episodes of importance. It is observed how both
of widespread use such as PDSI (Dracup et al., 1980; Guttman, 1998; the simulated SPI and SPEI tend, for the majority of times, to present
Hayes et al., 2011; Hayes et al., 1999; Vicente-Serrano et al., 2010b). In dry and humid periods of greater intensity than those observed, es-
the context of climate change with significant temperature variations pecially for longer time scales, as occurred in 1976–1977 for positive
12 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Fig. 8. Time-scales SPEI maps under RCP4.5. Geographical representation of the expected evolution of the SPEI for Aragon in the periods 2041–2070 and 2071–2100 compared to the
reference Historical Period (1971–2000) in terms of absolute values according to the RCP4.5 at different time-scales. The rows show the four time-scales analysed in the study (1
months, 3 months, 6 months and 12 months) and the columns, the three temporal periods (2011–2040, 2041–2070 and 2071–2100). The maps are generated by interpolating the
available stations over the territory.

values of the indexes and in 1981–1982 for negative ones. In general, 4.2.2. Future scenarios
the number of months of the period 1970–2000 located within each The uncertainties associated with both processes, verification and
of the classes defined for SPI/SPEI has been simulated very validation, should be considered when interpreting future scenarios.
satisfactorily. For drought projections the focus should be on changes in the frequency
The dry and wet periods detected in this study coincide with or are and magnitude of cases located at the lower tail of the distribution as
approximate to those published previously (Vicente-Serrano and was recommended by Vicente-Serrano et al. (2019).
Lopez-Moreno, 2005) based on SPI (dry episodes: 1986–1987, 1989 Future meteorological drought scenarios based on SPI barely show
and 1994–1997; wet episodes: 1976–1980), in the Climate Atlas of water balance variations with respect to normal values, regardless of
Aragon (López et al., 2007) based on the precipitation regime (dry epi- the time scale considered and the region of Aragon, except for the
sodes: 1970, 1985, 1993 and 1995), by Spinoni (Spinoni et al., 2015) Ebro Valley where there is a slight sign of drought at the end of the
based on a combined 12-month index (dry episodes: 1979–1980 and 21st century and under the conditions of RCP8.5.
1995–1998) and by Tselepidaki (Tselepidaki et al., 1992) from a These results were expected due to precipitation scenarios barely
European study (dry episode: 1989), among others. In some cases, the showing changes throughout the 21st century.
years are not exactly the same because of the use of different drought When considering other climatic variables, such as temperature, the
and temporal scale indexes. drought scenarios based on SPEI show a clear trend towards
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 13

Fig. 9. Time-scales SPEI maps under RCP8.5. Geographical representation of the expected evolution of the SPEI for Aragon in the periods 2041–2070 and 2071–2100 compared to the
reference Historical Period (1971–2000) in terms of absolute values according to the RCP8.5 at different time-scales. The rows show the four time-scales analysed in the study (1
months, 3 months, 6 months and 12 months) and the columns the three temporal periods (2011–2040, 2041–2070 and 2071–2100). The maps are generated by interpolating the
available stations over the territory.

increasingly dry periods and longer droughts, especially in the Ebro area limitations under global warming conditions(Mishra and Singh, 2010;
and south-west of the region. According to the trends shown by the Mishra and Singh, 2011; Vicente-Serrano et al., 2010a) and it is for
temperature and precipitation scenarios obtained for Aragon, the re- this reason that, when considering AED in the calculation of SPEI, such
sults obtained were expected. The fact that the results obtained at the different results are obtained, especially at the end of the century and
12-month scale are more intense than those of 1–3 months is partly a not only under the conditions of RCP8.5, that some authors consider
result of the way in which drought indices are formulated and the less realistic (Hausfather and Peters, 2020), but also of RCP4.5.
autoregressive component of its metric so that when the timescale in- Some studies recommend the use of PET and add value against
creases, changes in the frequency of drought conditions increase more global warming (Hu and Willson, 2000; Vicente-Serrano et al., 2010a,
in comparison to changes in the mean state. Although, recently, Tsakiris and Vangelis, 2005). Recent studies (Vicente-Serrano et al.,
Vicente-Serrano et al. (2019) showed that these changes are indepen- 2019; Vicente-Serrano et al., 2020) suggest using AED in the future
dent of the metric with which these indices have been calculated, study of droughts, as well as analyzing the impact caused by the in-
changes in the frequency of drought conditions increase more in com- crease in CO2 (Yang et al., 2019). Probably, considering the response
parison to changes in the mean state. that vegetation could have to an increase in CO2 and, therefore, in the
The lack of consideration of variables such as temperature, wind or evapotranspiration process, could provide some variation in the future
humidity in the calculation of SPI means that this index presents certain evolution of drought episodes that should be explored in future studies.
14 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Fig. 10. Time series for Zaragoza under MPI-ESM-MR RCP4.5. Evolution of the SPEI (first column) and the SPI (second column) based on the MPI-MR-SM model and under the RCP 4.5 at
different time-scales - 1 month (first row), 3 months (second row), 6 months (third row) and 12 months (fourth row)- for Zaragoza.

The results of future drought scenarios presented here show results scenarios of the fifth IPCC report and providing the added value of work-
in line with those obtained in other studies where it is concluded that ing at the local scale by applying a downscaling technique.
the Mediterranean regions will experience an increase in the severity The scenarios obtained in this study indicate that the Ebro Valley, the
and frequency of droughts (Stagge et al., 2015) as a result of a slight de- most populated area in the region that includes the largest city, Zara-
crease in precipitation and an abrupt increase in temperatures (EEA, goza with N650.000 people, will be most susceptible to future periods
2010; Stagge et al., 2015) and which represent an increase in water of extreme drought and will suffer periods of drought of greater inten-
scarcity (Estrela et al., 2012). More specifically in the region of Aragon, sity and duration, especially at the end of this century, which will
the ECCE project, based on dynamic downscaling and scenarios of the have consequences in sectors such as health, water management, econ-
fourth IPCC report (Ministerio de Medio Ambiente, 2005), showed a fu- omy and society in general (Lee et al., 2017).
ture decline of the Ebro runoff, and Cook (Cook et al., 2014), based on It is remarkable that, in previous publications (Ribalaygua et al.,
scenarios of the fifth IPCC report but without downscaling, obtained 2013a), we detected that the highest values of maximum temperature,
an increase in drought episodes based on SPEI. especially at the end of the century and in summer (around 40 °C) as
well as the greatest intensity of heatwaves will also take place in this
area, so it will be especially vulnerable and these data should be consid-
4.2.3. Impact on the territory ered in the development of specific measures for adapting to climate
Although there have been studies on the Aragon area, none present change.
as complete a picture as this study, combining drought evaluation with Adaptation to climate change in each region requires studies applied
SPI and SPEI (that is, considering the effect of global warming) based on to the climatic dynamics of each territory, so downscaling quality
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 15

Fig. 11. Time series for Zaragoza under MPI-ESM-MR RCP8.5. Evolution of the SPEI (first column) and the SPI (second column) based on the MPI-MR-SM model and under the RCP 8.5 at
different time-scales - 1 month (first row), 3 months (second row), 6 months (third row) and 12 months (fourth row)- for Zaragoza.

studies are essential for this. However, these results at the local level are cumulative periods of longer duration are considered. This trend is de-
also useful for the whole of the southern Iberian Peninsula and central tected only in the future drought scenarios based on SPEI (which in ad-
Europe, since Aragon brings together geographical and climatic features dition to precipitation, considers AED), while in the SPI-based scenarios
representative also of these other areas. it is softened. These results reinforce the need to study these extreme
phenomena in a context of climate change, considering the
5. Conclusions temperature.
At the territory level, spatial representation allowed us to discover
The generation for the first time of climate change scenarios of that the area that will be most affected by longer and more intense pe-
drought indexes for the region of Aragon (Spain) based on nine ESMs riods of drought, but also the greatest decrease in precipitation (around
and two RCPs from CMIP5 has allowed us to obtain simultaneously 10%), is the Ebro Valley, the area that concentrates most of the popula-
the most accurate representation to date of the magnitude, duration tion as well as the main economic activities of the zone. The results have
and intensity of meteorological drought episodes and their duration in also allowed, for the first time, the study of future drought indexes at the
different periods of accumulation in this area of Spain. The use of differ- observatory level, specifically for the most populous city, Zaragoza.
ent drought indices and drought time-scales and its graphic representa- To cope effectively with the impacts of these extreme events that are
tion is a relevant novelty in the scientific literature. expected in the present century, it is essential to be able to generate
This has allowed the detection of a clear trend towards increasingly local scenarios that accurately describe climate change at the territory
intense periods of drought, especially at the end of the century when level. On the one hand, our results not only confirm a trend already
16 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

described in the Mediterranean area of an increase in the severity and EEA, 2010. The European environment: State and outlook 2010: Thematic assessment –
Adapting to Climate Change. European Environment Agency, Copenhagen http://
frequency of droughts but can also serve as a model and sentinel for www.eea.europa.eu/soer/europe/adapting-to-climate-change.
similar areas, since it has very varied climatic and orographic conditions. Estrela, T., Perez-Martin, M.A., Vargas, E., 2012. Impacts of climate change on water re-
sources in Spain. Hydrological Sciences Journal-Journal Des Sciences Hydrologiques
57, 1154–1167.
Feyen, L., Dankers, R., 2009. Impact of global warming on streamflow drought in Europe.
Declaration of competing interest J. Geophys. Res.-Atmos. 114.
Flato, G., Marotzke, J., Abiodun, B., Braconnot, P., Chou, S.C., Collins, W., et al., 2014. Eval-
uation of Climate Models.
The authors declare that they have no known competing financial Forzieri, G., Feyen, L., Rojas, R., Florke, M., Wimmer, F., Bianchi, A., 2014. Ensemble
interests or personal relationships that could have appeared to influ- projections of future streamflow droughts in Europe. Hydrol. Earth Syst. Sci.
ence the work reported in this paper. 18, 85–108.
Fragoso, M., Carraca, M.D., Alcoforado, M.J., 2018. Droughts in Portugal in the 18th
century: a study based on newly found documentary data. Int. J. Climatol. 38,
Appendix A. Supplementary data 5522–5541.
Gaitan, E., Monjo, R., Portoles, J., Pino-Otin, M.R., 2019. Projection of temperatures and
Supplementary data to this article can be found online at https://doi. heat and cold waves for Aragon (Spain) using a two-step statistical downscaling of
CMIP5 model outputs. Sci. Total Environ. 650, 2778–2795.
org/10.1016/j.scitotenv.2020.140094. Gallego, M.C., Trigo, R.M., Vaquero, J.M., Brunet, M., Garcia, J.A., Sigro, J., et al., 2011. Trends
in frequency indices of daily precipitation over the Iberian Peninsula during the last
References century. J. Geophys. Res.-Atmos. 116.
Garcia-Barron, L., Aguilar, M., Sousa, A., 2011. Evolution of annual rainfall irregularity in
Alary, V., Messad, S., Aboul-Naga, A., Osman, M.A., Daoud, I., Bonnet, P., et al., 2014. Live- the southwest of the Iberian Peninsula. Theor. Appl. Climatol. 103, 13–26.
lihood strategies and the role of livestock in the processes of adaptation to drought in Guttman, N.B., 1998. Comparing the Palmer Drought Index and the standardized precip-
the Coastal Zone of Western Desert (Egypt). Agric. Syst. 128, 44–54. itation index. J. Am. Water Resour. Assoc. 34, 113–121.
Hao, Z.C., Hao, F.H., Singh, V.P., Ouyang, W., 2017. Quantitative risk assessment of the ef-
Bates, B., Kundzewicz, Z., Wu, S., Palutikof, J., 2008. Climate Change and Water. IPCC Sec-
fects of drought on extreme temperature in eastern China. J. Geophys. Res.-Atmos.
retariat, Geneva (210 pp).
122, 9050–9059.
Begueria, S., Vicente-Serrano, S.M., Reig, F., Latorre, B., 2014. Standardized precipitation
Hargreaves, G.L., Samani, Z.A., 1985. Reference crop evapotranspiration from temperature.
evapotranspiration index (SPEI) revisited: parameter fitting, evapotranspiration
Appl. Eng. Agric. 1985 (1), 96–99.
models, tools, datasets and drought monitoring. Int. J. Climatol. 34, 3001–3023.
Hausfather, Z., Peters, G.P., 2020. Emissions - the ‘business as usual’ story is misleading.
Beniston, M., Stephenson, D.B., Christensen, O.B., Ferro, C.A.T., Frei, C., Goyette, S., et al.,
Nature 577, 618–620.
2007. Future extreme events in European climate: an exploration of regional climate
Hayes, M., Svoboda, M., Wall, N., Widhalm, M., 2011. The Lincoln Declaration on Drought
model projections. Clim. Chang. 81, 71–95.
Indices. Bull. Am. Meteorol. Soc. 92, 485–488.
Bentsen, M., Bethke, I., Debernard, J.B., Iversen, T., Kirkevag, A., Seland, O., et al., 2013. The Hayes, M.J., Svoboda, M.D., Wilhite, D.A., Vanyarkho, O.V., 1999. Monitoring the 1996
Norwegian Earth System Model, NorESM1-M - part 1: description and basic evalua- drought using the standardized precipitation index. Bull. Am. Meteorol. Soc. 80,
tion of the physical climate. Geosci. Model Dev. 6, 687–720. 429–438.
Berg, A., Sheffield, J., 2018. Climate change and drought: the soil moisture perspective. Heavens, N., Ward, D., Natalie, M., 2013. Studying and projecting climate change with
Current Climate Change Reports 4, 180–191. Earth System Models. Nat. Educ. Knowl. 4 (5), 4.
Berg, A., Sheffield, J., Milly, P.C.D., 2017. Divergent surface and total soil moisture projec- Hoerling, M., Eischeid, J., Perlwitz, J., Quan, X.W., Zhang, T., Pegion, P., 2012. On the in-
tions under global warming. Geophys. Res. Lett. 44, 236–244. creased frequency of Mediterranean drought. J. Clim. 25, 2146–2161.
Blenkinsop, S., Fowler, H.J., 2007. Changes in European drought characteristics projected Hu, Q., Willson, G.D., 2000. Effects of temperature anomalies on the Palmer Drought Se-
by the PRUDENCE regional climate models. Int. J. Climatol. 27, 1595–1610. verity Index in the central United States. Int. J. Climatol. 20, 1899–1911.
Bryant, E.A., 1993. Natural hazards. Int. J. Climatol. 13, 344–346. https://doi.org/10.1002/ Iglesias, A., Garrote, L., Flores, F., Moneo, M., 2007. Challenges to manage the risk of water
joc.3370130310 E. Cambridge University Press, Cambridge. ISBN 0 521 37295 X. scarcity and climate change in the Mediterranean. Water Resour. Manag. 21,
Burke, E.J., Brown, S.J., 2008. Evaluating uncertainties in the projection of future drought. 775–788.
J. Hydrometeorol. 9, 292–299. IPCC, 2014. Climate change 2014: impacts, adaptation, and vulnerability. Part A: global
Calbo, J., 2010. Possible climate change scenarios with specific reference to Mediterranean and sectoral aspects. Contribution of Working Group II to the Fifth Assessment Report
regions. In: Sabater, S., Barcelo, D. (Eds.), Water Scarcity in the Mediterranean: Per- of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cam-
spectives Under Global Change. 8, pp. 1–13. bridge, United Kingdom and New York, NY, USA, p. 1132.
Carvalho, D., Rocha, A., Gomez-Gesteira, M., Santos, C.S., 2017. Potential impacts of climate Irmak, S., Kabenge, I., Skaggs, K.E., Mutiibwa, D., 2012. Trend and magnitude of changes in
change on European wind energy resource under the CMIP5 future climate projec- climate variables and reference evapotranspiration over 116-yr period in the Platte
tions. Renew. Energy 101, 29–40. River Basin, central Nebraska-USA. J. Hydrol. 420, 228–244.
Chen, Y.D., Li, J.F., Zhang, Q., 2016. Changes in site-scale temperature extremes over Iversen, T., Bentsen, M., Bethke, I., Debernard, J.B., Kirkevag, A., Seland, O., et al., 2013. The
China during 2071-2100 in CMIP5 simulations. J. Geophys. Res.-Atmos. 121, Norwegian Earth System Model, NorESM1-M - part 2: climate response and scenario
2732–2749. projections. Geosci. Model Dev. 6, 389–415.
Cherlet, M., Hutchinson, C., Reynolds, J., Hill, J., Sommer, S., von Maltitz, G. (Eds.), 2018. Jensen, M.E., Haise, H.R., 1963. Estimating evapotranspiration from solar radiation. J. Irrig.
World Atlas of Desertification. Publication Office of the European Union, Drain. Div. 89, 15–41.
Luxembourg (United Nations Environment Program). Jones, C.D., Hughes, J.K., Bellouin, N., Hardiman, S.C., Jones, G.S., Knight, J., et al., 2011. The
HadGEM2-ES implementation of CMIP5 centennial simulations. Geosci. Model Dev. 4,
Chylek, P., Li, J., Dubey, M., Wang, M., Lesins, G., 2011. Observed and model simulated 20th
543–570.
century Arctic temperature variability: Canadian Earth System Model CanESM2.
Knutti, R., Sedlacek, J., 2013. Robustness and uncertainties in the new CMIP5 climate
Atmos. Chem. Phys. Discuss. 11, 22893–22907. https://doi.org/10.5194/acpd-11-
model projections. Nat. Clim. Chang. 3, 369–373.
22893-2011.
Lavaysse, C., Vrac, M., Drobinski, P., Lengaigne, M., Vischel, T., 2012. Statistical downscal-
Collados-Lara, A.J., Pulido-Velazquez, D., Pardo-Iguzquiza, E., 2018. An integrated statisti-
ing of the French Mediterranean climate: assessment for present and projection in an
cal method to generate potential future climate scenarios to analyse droughts. Water
anthropogenic scenario. Nat. Hazards Earth Syst. Sci. 12, 651–670.
10.
Lee, S.H., Yoo, S.H., Choi, J.Y., Bae, S., 2017. Assessment of the impact of climate change on
Collins, W., Bellouin, N., Doutriaux-Boucher, M., Gedney, N., Hinton, T., Jones, C.D.,
drought characteristics in the Hwanghae Plain, North Korea using time series SPI and
Liddicoat, S., Martin, G., O'Connor, F., Rae, J., Senior, C., Totterdell, I., Woodward, S.,
SPEI: 1981-2100. Water 9.
Reichler, T., Kim, J., Halloran, P., 2008. Evaluation of the HadGEM2 Model. Hadley Cen-
Lesk, C., Rowhani, P., Ramankutty, N., 2016. Influence of extreme weather disasters on
tre Technical Note. HCTN 74. Met Office Hadley Centre, Exeter, UK.
global crop production. Nature 529, 84.
Cook, B.I., Smerdon, J.E., Seager, R., Coats, S., 2014. Global warming and 21st century dry- Livneh, B., Hoerling, M.P., 2016. The physics of drought in the US Central Great Plains.
ing. Clim. Dyn. 43, 2607–2627. J. Clim. 29, 6783–6804.
Dai, A.G., 2011. Drought under global warming: a review. Wiley Interdisciplinary Lloyd-Hughes, B., Saunders, M.A., 2002. A drought climatology for Europe. Int. J. Climatol.
Reviews-Climate Change 2, 45–65. 22, 1571–1592.
Dai, A.G., 2013. Increasing drought under global warming in observations and models (vol López, F., Cabrera, M., Cuadrat, J.M., 2007. Atlas Climático de Aragón. first ed. J. Factory,
3, pg 52, 2013). Nat. Clim. Chang. 3, 171. Spain.
Dai, A.G., Zhao, T.B., Chen, J., 2018. Climate change and drought: a precipitation and evap- Lopez-Bustins, J.A., Pascual, D., Pla, E., Retana, J., 2013. Future variability of droughts in
oration perspective. Current Climate Change Reports 4 (3), 301–312. https://doi.org/ three Mediterranean catchments. Nat. Hazards 69, 1405–1421.
10.1007/s40641-018-0101-6 Sep. Machado, M.J., Benito, G., Barriendos, M., Rodrigo, F.S., 2011. 500 Years of rainfall variabil-
Dracup, J.A., Lee, K.S., Paulson, E.G., 1980. On the statistical characteristics of drought ity and extreme hydrological events in southeastern Spain drylands. J. Arid Environ.
events. Water Resour. Res. 16, 289–296. 75, 1244–1253.
Dunne, J.P., John, J.G., Adcroft, A.J., Griffies, S.M., Hallberg, R.W., Shevliakova, E., et al., 2012. Marcos-Garcia, P., Lopez-Nicolas, A., Pulido-Velazquez, M., 2017. Combined use of relative
GFDL'sESM2global coupled climate carbon Earth System Models. PartI:physical for- drought indices to analyze climate change impact on meteorological and hydrological
mulation and baseline simulation characteristics. J. Clim. 25, 6646–6665. droughts in a Mediterranean basin. J. Hydrol. 554, 292–305.
 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094 17

Marsland, S.J., Haak, H., Jungclaus, J.H., Latif, M., Roske, F., 2003. The Max-Planck-Institute arid Areas. Proceedings of a Consultants Meeting. International Atomic Energy
global ocean/sea ice model with orthogonal curvilinear coordinates. Ocean Model 5, Agency (IAEA) (IAEA-TECDOC–1026. Ref. number: 29062763).
91–127. Sonmez, F.K., Komuscu, A.U., Erkan, A., Turgu, E., 2005. An analysis of spatial and temporal
Mckee, T., Doesken, N., Kleist, J., 1993. The relationship of drought frequency and duration dimension of drought vulnerability in Turkey using the standardized precipitation
times scales. 8th Conference on Applied Climatology. American Meteorological Soci- index. Nat. Hazards 35, 243–264.
ety, Anaheim, California, pp. 179–184 January 17–22. Spinoni, J., Naumann, G., Vogt, J.V., Barbosa, P., 2015. The biggest drought events in Europe
McVicar, T.R., Roderick, M.L., Donohue, R.J., Li, L.T., Van Niel, T.G., Thomas, A., et al., 2012a. from 1950 to 2012. Journal of Hydrology-Regional Studies 3, 509–524.
Global review and synthesis of trends in observed terrestrial near-surface wind Stagge, J.H., Kohn, I., Tallaksen, L.M., Stahl, K., 2015. Modeling drought impact occurrence
speeds: implications for evaporation. J. Hydrol. 416, 182–205. based on meteorological drought indices in Europe. J. Hydrol. 530, 37–50.
McVicar, T.R., Roderick, M.L., Donohue, R.J., Van Niel, T.G., 2012b. Less bluster ahead? Stanke, C., Kerac, M., Prudhomme, C., Medlock, J., Murray, V., 2013. Health effects of
Ecohydrological implications of global trends of terrestrial near-surface wind speeds. drought: a systematic review of the evidence. PLoS Curr., 5 Jun 5. pii: ecurrents.
Ecohydrology 5, 381–388. dis.7a2cee9e980f91ad7697b570bcc4b004. https://doi.org/10.1371/currents.
Ministerio de Medio Ambiente, 2005. Assessment Report of the Preliminary Impacts in dis.7a2cee9e980f91ad7697b570bcc4b004.
Spain Due to Climate Change. Centro de publicaciones del Ministerio de Medio Taylor, K.E., Stouffer, R.J., Meehl, G.A., 2012. An overview of CMIP5 and the experiment de-
Ambiente, Madrid. sign. Bull. Am. Meteorol. Soc. 93, 485–498.
Mishra, A.K., Singh, V.P., 2010. A review of drought concepts. J. Hydrol. 391, 204–216.
Thornthwaite, C., 1948. An approach toward a rational classification of climate. Geogr.
Mishra, A.K., Singh, V.P., 2011. Drought modeling - a review. J. Hydrol. 403, 157–175. Rev. 38, 55–94.
Monjo, R., Gaitan, E., Portoles, J., Ribalaygua, J., Torres, L., 2016. Changes in extreme precip-
Tripathi, S., Srinivas, V.V., Nanjundiah, R.S., 2006. Dowinscaling of precipitation for climate
itation over Spain using statistical downscaling of CMIP5 projections. Int. J. Climatol.
change scenarios: a support vector machine approach. J. Hydrol. 330, 621–640.
36, 757–769.
Tsakiris, G., Vangelis, H., 2005. Establishing a drought index incorporating evapotranspira-
Moss, R.H., Edmonds, J.A., Hibbard, K.A., Manning, M.R., Rose, S.K., van Vuuren, D.P., et al.,
tion. European Water 9 (10), 3–11 2005.
2010. The next generation of scenarios for climate change research and assessment.
Nature 463, 747–756. Tselepidaki, I., Zarifis, B., Asimakopoulos, D.N., 1992. Low precipitation over Greece during
Moutahir, H., Bellot, P., Monjo, R., Bellot, J., Garcia, M., Touhami, I., 2017. Likely effects of 1989–1990. Theor. Appl. Climatol. 46, 115–121.
climate change on groundwater availability in a Mediterranean region of Southeast- Uppala, S.M., Kallberg, P.W., Simmons, A.J., Andrae, U., Bechtold, V.D., Fiorino, M., et al.,
ern Spain. Hydrol. Process. 31, 161–176. 2005. The ERA-40 re-analysis. Q. J. R. Meteorol. Soc. 131, 2961–3012.
Mukherjee, S., Mishra, A., Trenberth, K.E., 2018. Climate change and drought: a perspec- Van der Linden, P., Mitchell, J. (Eds.), 2009. ENSEMBLES: Climate Change and Its Impacts.
tive on drought indices. Current Climate Change Reports 4, 145–163. Summary of Research and Results From the ENSEMBLES Project. Met Office Hadley
Nychka D, Furrer R, Paige J, Sain S, 2015. Fields: tools for spatial data. R package ver- Centre, FitzRoy Road, Exeter EX1 3PB, UK 160 pp.
sion 9.0. (URL: https://doi.org/10.5065/D6W957CT). www.image.ucar.edu/ Vicente-Serrano, S.M., 2013. In: Martínez, C.C.-L., Rodríguez, F.V. (Eds.), Spatial and Tem-
fields. poral Evolution of Precipitation Droughts in Spain in the Last Century in Adverse
Ojeda, M.G.V., Gamiz-Fortis, S.R., Castro-Diez, Y., Esteban-Parra, M.J., 2017. Evaluation of Weather in Spain. WCRP Spanish Committee, Madrid, Spain, pp. 283–296.
WRF capability to detect dry and wet periods in Spain using drought indices. Vicente-Serrano, S.M., 2016. Foreword: drought complexity and assessment under cli-
J. Geophys. Res.-Atmos. 122, 1569–1594. mate change conditions. Cuadernos De Investigacion Geografica 42, 7–11.
Organization WMO, 2012. In: Svoboda, M., Hayes, M., Wood, D. (Eds.), Standardized Pre- Vicente-Serrano, S.M., Begueria, S., 2016. Comment on ‘Candidate distributions for clima-
cipitation Index User Guide. WMO, Geneva. tological drought indices (SPI and SPEI)’ by James H. Stagge et al. Int. J. Climatol. 36,
Palmer, W., 1965. «Meteorological Drought». Research Paper No. 45. U.S. Department of 2120–2131.
Commerce Weather Bureau febrero de. (58 páginas). Available in National Climatic Vicente-Serrano, S.M., Lopez-Moreno, J.I., 2005. Hydrological response to different
Data Center de NOAA. http://www.ncdc.noaa.gov/temp-and-precip/drought/docs/ time scales of climatological drought: an evaluation of the Standardized Precip-
palmer.pdf. itation Index in a mountainous Mediterranean basin. Hydrol. Earth Syst. Sci. 9,
Paparrizos, S., Maris, F., Weiler, M., Matzarakis, A., 2018. Analysis and mapping of present 523–533.
and future drought conditions over Greek areas with different climate conditions. Vicente-Serrano, S.M., Gonzalez-Hidalgo, J.C., de Luis, M., Raventos, J., 2004. Drought pat-
Theor. Appl. Climatol. 131, 259–270. terns in the Mediterranean area: the Valencia region (eastern Spain). Clim. Res. 26,
Perez, J., Menendez, M., Mendez, F.J., Losada, I.J., 2014. Evaluating the performance of 5–15.
CMIP3 and CMIP5 global climate models over the north-east Atlantic region. Clim. Vicente-Serrano, S.M., Begueria, S., Lopez-Moreno, J.I., 2010a. A multiscalar drought index
Dyn. 43, 2663–2680. sensitive to global warming: the standardized precipitation evapotranspiration
R Development Core Team, 2010. R: A Language and Environment for Statistical Comput- index. J. Clim. 23, 1696–1718.
ing. R Foundation for Statistical Computing, Vienna, Austria 3-900051-07-0. http:// Vicente-Serrano, S.M., Begueria, S., Lopez-Moreno, J.I., Angulo, M., El Kenawy, A., 2010b. A
www.R-project.org (Accessed date: 6 July 2011). new global 0.5 degrees gridded dataset (1901–2006) of a multiscalar drought index:
Raddatz, T.J., Reick, C.H., Knorr, W., Kattge, J., Roeckner, E., Schnur, R., et al., 2007. Will the comparison with current drought index datasets based on the Palmer Drought Sever-
tropical land biosphere dominate the climate-carbon cycle feedback during the ity Index. J. Hydrometeorol. 11, 1033–1043.
twenty-first century? Clim. Dyn. 29, 565–574. Vicente-Serrano, S.M., Lasanta, T., Gracia, C., 2010c. Aridification determines changes in
Rebetez, M., Mayer, H., Dupont, O., Schindler, D., Gartner, K., Kropp, J.P., et al., 2006. Heat forest growth in Pinus halepensis forests under semiarid Mediterranean climate con-
and drought 2003 in Europe: a climate synthesis. Ann. For. Sci. 63, 569–577. ditions. Agric. For. Meteorol. 150, 614–628.
Ribalaygua, J., Rosa Pino, M., Portoles, J., Roldan, E., Gaitan, E., Chinarro, D., et al., 2013a. Vicente-Serrano, S.M., Van der Schrier, G., Begueria, S., Azorin-Molina, C., Lopez-Moreno,
Climate change scenarios for temperature and precipitation in Aragon (Spain). Sci. J.I., 2015. Contribution of precipitation and reference evapotranspiration to drought
Total Environ. 463, 1015–1030. indices under different climates. J. Hydrol. 526, 42–54.
Ribalaygua, J., Torres, L., Portoles, J., Monjo, R., Gaitan, E., Pino, M.R., 2013b. Description Vicente-Serrano, S.M., Dominguez-Castro, F., McVicar, T.R., Tomas-Burguera, M., Pena-
and validation of a two-step analogue/regression downscaling method. Theor. Appl. Gallardo, M., Noguera, I., et al., 2019. Global characterization of hydrological and me-
Climatol. 114, 253–269. teorological droughts under future climate change: the importance of timescales,
Ribalaygua, J., Gaitan, E., Portoles, J., Monjo, R., 2018. Climatic change on the Gulf of vegetation-CO2 feedbacks and changes to distribution functions. Int. J. Climatol.
Fonseca (Central America) using two-step statistical downscaling of CMIP5 model https://doi.org/10.1002/joc.6350.
outputs. Theor. Appl. Climatol. 132, 867–883.
Vicente-Serrano, S.M., McVicar, T.R., Miralles, D.G., Yang, Y.T., Tomas-Burguera, M., 2020.
Roderick, M.L., Greve, P., Farquhar, G.D., 2015. On the assessment of aridity with changes
Unraveling the influence of atmospheric evaporative demand on drought and its re-
in atmospheric CO2. Water Resour. Res. 51 (7), 5450–5463. https://doi.org/10.1002/
sponse to climate change. Wiley Interdisciplinary Reviews-Climate Change 11.
2015wr017031 Jul.
Voldoire, A., Sanchez-Gomez, E., Melia, D.S.Y., Decharme, B., Cassou, C., Senesi, S., et al.,
Rodriguez, R., Navarro, X., Casas, M.C., Ribalaygua, J., Russo, B., Pouget, L., et al., 2014. In-
2013. The CNRM-CM5.1 global climate model: description and basic evaluation.
fluence of climate change on IDF curves for the metropolitan area of Barcelona
Clim. Dyn. 40, 2091–2121.
(Spain). Int. J. Climatol. 34, 643–654.
Wang, B., Zhou, T.J., Yu, Y.Q., 2009. A view of earth system model development. Acta
Roldán, E., Gómez, M., Pino-Otín, R., Esteban, M., Díaz, J., 2011. Determinación de zonas
Meteorologica Sinica 23, 1–17.
isotérmicas y selección de estaciones meteorológicas representativas en Aragón
como base para la estimación del impacto del cambio climático sobre la posible Watanabe, S., Hajima, T., Sudo, K., Nagashima, T., Takemura, T., Okajima, H., et al., 2011.
relación entre mortalidad y temperatura. Rev Esp Salud Pública 85, 457–469. MIROC-ESM 2010: model description and basic results of CMIP5-20c3m experiments.
Santiago, J.M., Munoz-Mas, R., Solana-Gutierrez, J., de Jalon, D.G., Alonso, C., Martinez- Geosci. Model Dev. 4, 845–872.
Capel, F., et al., 2017. Waning habitats due to climate change: the effects of changes Wilhite, D., 2000. Drought as a natural hazard: concepts and definitions. In: Wilhite, D.A.
in streamflow and temperature at the rear edge of the distribution of a cold-water (Ed.), Droughts: Global Assessment. Routledge, London, pp. 3–18.
fish. Hydrol. Earth Syst. Sci. 21. Wilhite, D.A., Glantz, M.H., 1985. Understanding the drought phenomenon: the role of
Sheffield, J., Wood, E.F., 2008. Projected changes in drought occurrence under future definitions. Water Int. 10 (3), 111–120.
global warming from multi-model, multi-scenario, IPCC AR4 simulations. Clim. Dyn. Willett, K.M., Dunn, R.J.H., Thorne, P.W., Bell, S., de Podesta, M., Parker, D.E., et al., 2014.
31, 79–105. HadISDH land surface multi-variable humidity and temperature record for climate
Sheffield, J., Wood, E.F., Roderick, M.L., 2012. Little change in global drought over the past monitoring. Clim. Past 10, 1983–2006.
60 years. Nature 491, 435. WMO (World Meteorological Organization), 2017. Statement on the State of the Global
Skaugen, T., Astrup, M., Roald, L.A., Forland, E., 2004. Scenarios of extreme daily precipita- Climate in 2016.N° 1189.2017. 978-92-63-11189-0.
tion for Norway under climate change. Nord. Hydrol. 35, 1–13. Xiao-Ge, X., Tong-Wen, W., Jie, Z., 2013. Introduction of CMIP5 experiments carried out
Smith, M., Allen, R., Pereira, L., 1998. Revised FAO Methodology for Crop-water Re- with the climate system models of Beijing Climate Center. Adv. Clim. Chang. Res. 4,
quirements. Management of Nutrients and Water in Rainfed Arid and Semi- 41–49. https://doi.org/10.3724/SP.J.1248.2013.041.
18 E. Gaitán et al. / Science of the Total Environment 740 (2020) 140094

Yang, Y.T., Zhang, S.L., McVicar, T.R., Beck, H.E., Zhang, Y.Q., Liu, B., 2018. Disconnection be- Yukimoto, S., Yoshimura, H., Hosaka, M., Sakami, T., Tsujino, H., Hirabara, M., Tanaka, T.,
tween trends of atmospheric drying and continental runoff. Water Resour. Res. 54, Deushi, M., Obata, A., Nakano, H., Adachi, Y., Shindo, E., Yabu, S., Ose, T., Kitoh, A.,
4700–4713. 2011. Meteorological research institute- earth system model v1 (MRI-ESM1)—
Yang, Y.T., Roderick, M.L., Zhang, S.L., McVicar, T.R., Donohue, R.J., 2019. Hydrologic impli- model description. Technical Report of MRI. vol. 64.
cations of vegetation response to elevated CO2 in climate projections. Nat. Clim.
Chang. 9, 44.