Manciu Et Al., 2022

Climate Dynamics
https://doi.org/10.1007/s00382-022-06545-1
Impacts of land cover changes and global warming on climate

in Colombia during ENSO events
Astrid Manciu1,2 · Anja Rammig1 · Andreas Krause1 · Benjamin Raphael Quesada2
Received: 17 January 2022 / Accepted: 15 October 2022

© The Author(s) 2022
Abstract
Colombia is highly vulnerable to climate change which may be intensified due to the climatic effects of regional deforestation.
Here, we quantify the impact of historical (1900–2011) land cover changes (LCC) and of global warming during ENSO events
(CC) on precipitation, temperature and surface energy balance components by running the Weather Research and Forecast-
ing model WRF v3.9 at 10 km resolution. We find that historical anthropogenic CC causes a mean temperature increase of
0.77 ± 0.02 °C in Colombia, which is more pronounced in high altitudes. Precipitation is enhanced by 0.98 ± 0.30 mm/day
(+ 9%), particularly over forested areas and reduced at the Pacific coast. LCC imply a reduction of precipitation particularly
above the Andes (− 0.48 ± 0.10 mm/day) and Caribbean Coast (− 0.67 ± 0.12 mm/day), where LCC effects dampen CC
effects by 24% and 72%, respectively. La Niña tends to intensify LCC and CC effects in the Andes but dampens them at the
Coast, roughly by a factor of two compared to El Niño impacts in both regions. At the subregional level, LCC and CC can
have impacts of similar magnitude on precipitation highlighting the need to precisely account for both drivers in hydrocli-
matic assessments. Contrary to almost all observations and similar simulations with climate models, WRF simulates a cool-
ing bias after historical deforestation in Colombia, even with alternative WRF land surface models. We identify two main
sources of biases in the default WRF parametrization to explain this inaccuracy: (1) surface shortwave radiation reflected
after deforestation is overestimated; (2) associated evapotranspiration loss is underestimated. Improved model representation
and validation of tropical vegetation properties are necessary to provide more robust and confident projections.
Keywords Land cover change · Climate change · Surface energy balance · Deforestation · Regional climate modelling
1 Introduction climate by reducing carbon stocks and releasing carbon into

the atmosphere (Pan et al. 2011), leading to global warming
Since preindustrial times, mean surface air temperature has (biogeochemical feedback to deforestation) and by alterations
risen globally by 1 °C due to anthropogenic emissions of of regional energy, momentum and water fluxes (biophysical
greenhouse gases from fossil fuel burning and land use and feedbacks), which can outbalance biogeochemical effects and
land cover changes (LCC, IPCC 2018). Especially the trop- lead regionally to net cooling or additional warming (Claussen
ics are recently experiencing accelerating deforestation rates et al. 2001). Jia et al. (2019) suggest that on a global scale, bio-
(Hansen et al. 2013, Song et al. 2018). Deforestation influences geochemical effects dominate the interplay of both feedbacks.
However, the net impacts of LCC highly vary in space and
time and may have contrasting consequences depending on
* Benjamin Raphael Quesada latitude (Davin and de Noblet-Ducoudré 2010). Modelling and
[email protected]
observational studies suggest that boreal regions experience a
Astrid Manciu cooling of surface air temperature, while lower latitudes show
[email protected]
warming in response to local deforestation (Alkama and Ces-
1
TUM School of Life Sciences Weihenstephan, Technical catti 2016; Findell et al. 2017). Colombia, a megadiversity hot-
University of Munich, Hans–Carl–von–Carlowitz–Platz 2, spot and one of the top 12 deforesting countries (FAO 2015),
85354 Freising, Germany is particularly affected by such LCC impacts, which in tropi-
2
Faculty of Natural Sciences, “Interactions Climate‐ cal countries are expected to amplify the warming from CC.
Environment (ICE)” Research Group, Earth System Sciences Every year, roughly 180,000 ha (IDEAM 2017, 2018, 2019) of
Program, Universidad del Rosario, Bogotá, Colombia
13
Vol.:(0123456789)
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

A. Manciu et al.
forests in Colombia are lost due to agricultural expansion, cat- biophysical and land cover properties: Colombian Amazon,
tle ranching, urbanization and timber extraction. At the same Caribbean, Pacific Coast, Orinoco plains (Llanos orien-
time, the intensity of extreme weather events associated with tales) and Andes. The three branches (cordilleras) of the
the El Niño-Southern Oscillation (ENSO), e.g. droughts or Andes, which cross the country in the center and west,
floods, is increasing (Ávila et al. 2019; Hoyos et al. 2013). confers Colombia a unique and complex orography with
Thus, the question arises of how such climate events and LCC altitudes ranging between 5800 m and sea level, resulting
are interacting. in a wide variety of climatic and vegetation zones. Local-
Understanding deforestation and global warming impacts scale phenomena such as inter-Andean valley circulations
on climatic variables in Colombia is crucial because this coun- but also large-scale oscillations in the Atlantic and Pacific,
try concentrates very high levels of ecosystem services, terrain interactions with the Amazon and Orinoco basin influence
complexity, species richness and water provision. Modelling Colombia’s climatic variability (Poveda et al. 2011; Espi-
this region is challenging because global climate models cannot noza et al. 2020). On the inter-annual scale however, the
resolve small-scale terrain complexity inherent to the presence hydrological variability is dominated by ENSO, with El
of the three Andean cordilleras (Posada-Marín et al. 2018; Espi- Niño causing less evapotranspiration followed by less recy-
noza et al. 2020). Regional Climate Models (RCMs) can repre- cled precipitation, consequently fewer river discharges, and
sent local, small-scale phenomena more accurately than global less cloud cover which facilitates more radiation and eventu-
General Circulation Models (GCMs) or Earth System Models ally higher temperatures (Poveda et al. 2011). La Niña, the
(ESM). One of the most widely used RCMs is the Weather opposing ENSO-phase, has a reversed effect. ENSO impacts
Research and Forecasting model (WRF) which has proven its the western Andean branch first while the eastern branch is
capability to dynamically downscale global reanalysis data from affected with a little time lag and less strong (Poveda et al.
coarse to high resolution in good agreement with observations 2011). Generally speaking, Colombia experiences a drier
(e.g. Posada-Marín et al. 2018; Caldwell et al. 2009; Lee and and prolonged dry season during El Niño, and a wetter and
Berbery 2012). In complex terrain domains, WRF is considered prolonged wet season during La Niña.
to produce climatic features even better than the original reanal-
ysis data sets due to more accurate representation of orography
(e. g.Heikkilä et al. 2010; Sun et al. 2016; Soares et al. 2012; 2.2 Model settings
Gao et al. 2012; Jiménez-Esteve et al. 2018).
To date, no impact study for Colombia exists that evaluates In order to simulate climate response to LCC and CC
the impacts of LCC on climate at high spatial resolution. Thus, in Colombia, we use WRF version 3.9.1.1 (Skamarock
we present here the first high-resolution, dynamical downscal- et al. 2008). This non-hydrostatic, fully compressible
ing study investigating LCC and CC impacts in Colombia. and terrain-following sigma coordinate model (Powers
We perform sensitivity simulations with WRF v3.9, forcing et al. 2017) has been proven to perform well in high-
the model with historical land cover and climate input and resolution regional modelling studies and is one of the
compare the results against present-day conditions. Our aims world’s most widely used numerical weather prediction
are to answer the following questions: models. The simulations are executed by the Advanced
Research WRF (ARW) dynamical core of wrf.exe. It con-
(1) How do historical LCC and CC influence surface air tains basic dynamical equations for advection, Coriolis,
temperature and precipitation in different regions of pressure gradient terms, buoyancy and diffusion. Options
Colombia, during ENSO events (2009–2011)? on microphysics, radiation schemes and planetary bound-
(2) What is the dominant forcing for regional climate ary layers are set in namelist.input. Lookup tables for
changes in Colombia? soil characteristics and land cover and vegetation prop-
(3) How are land cover changes and global warming erties set important physical features like minimum and
impacts modulated by ENSO? maximum albedo per land use category, surface rough-
(4) How is the surface energy balance affected by different ness, rooting depth, which the model accesses during the
scenarios of land cover change in Colombia? model run.
The simulations are carried out on two regional
domains encompassing a large external domain (D01)
2 Materials and methods covering large parts of Central America, Northern South
America, Amazon basin, the Caribbean Sea and Eastern
2.1 Study site Pacific Ocean, along with an internal (nested) domain cov-
ering continental Colombian territory (D02, see Fig. 1),
Colombia is located in the northwest of South America and excluding the Colombian islands of San Andres, Providen-
contains five major biogeographic regions with contrasting cia and Santa Catalina off the coast of Nicaragua.
13

Impacts of land cover changes and global warming on climate in Colombia during ENSO events
All simulations are performed from January 1st, 2009 to for simulations lasting longer than a few days (Wang et al.
December 31st, 2011, excluding the first month from the 2017). Both domains have 35 vertical levels. D01 has a
analysis due to spin-up time, covering two major ENSO horizontal resolution of 30 km. A finer 10 km resolution for
events with extreme socio-economic consequences in D02 was chosen as a trade-off between computational time
Colombia (El Niño 2009–2010 and La Niña 2010–2011, and high resolution of fine-scale climate events. Initial and
Hoyos et al. (2013), Comisión Económica para América boundary conditions for the simulations were retrieved from
Latina y el Caribe (2012)) as well as some ENSO-neu- the 6-hourly, 0.75° × 0.75° gridded ERA-Interim reanalysis
tral periods (i.e. March-July 2009, March-June 2010 and dataset provided by the European Center for Medium-Range
May–July 2011 For this period, we find an average Oceanic Weather Forecasts (ECMWF, Dee et al. 2011).
Niño Index (ONI) value (based on the 3-month running To avoid large deviations between simulation and driving
mean of SST anomalies in the Niño 3.4 region [5°N-5°S, fields (Bowden et al. 2012) we use spectral nudging on both
120°-170°W]) of − 0.34 (as computed by the Climate Pre- domains during the whole simulation run, which has been
diction Center: https://origin.cpc.ncep.noaa.gov/products/ shown to improve simulated results especially in high-res-
analysis_monitoring/ensostuff/ONI_v5.php), meaning that olution downscaling studies (e.g. Chotamonsak et al. 2012;
our simulation period (2009–2011) is slightly biased towards Bowden et al. 2012; Heikkilä et al. 2010). Additionally, Paul
more representative of weak La Niña/Neutral periods. It et al. (2016) found a reduction in precipitation bias, specifi-
appears that Summer 2010-Spring 2011 was characterized cally in rainy regions. Nudging, which comprises relaxation
by an Eastern-Pacific (EP) La Niña event of Medium grade techniques, maintains on the one hand large-scale features
(although a strong minimum trimestrial ONI of − 1.6) while from the input data driving field and on the other hand, sim-
Summer 2009-Spring 2010 was marked by a Central-Pacific ulates small-scale features. To let the model develop its own
(CP) El Niño event of Medium grade (with an amplitude of mesoscale and synoptic structures in the surface-near tropo-
1.7 °C) (Ren et al. 2018). sphere, we decide to nudge zonal and meridional wind (U
We used two-way nesting to allow interactions between and V), temperature (T) and geopotential height (PH) only
inner and outer domain which is the recommended method above the planetary boundary layer additionally to switch-
ing off nudging in the first ten levels from the bottom of
the model. Similar to Posada-Marín et al. (2018) and Paul
et al. (2016), we use a relaxation time of approximately one
D01
hour, which corresponds to a value of 0.0003 for U, V, T and
PH nudging coefficient (guf, gt, gph). The top wave number,
which is the number of waves contained in the domain, is
D02 set to three in x and y direction, with three being the maxi-
mum one that is nudged, to capture ERA-Interim features
with wavelengths of approximately 1200 km and upwards
(the WRF domain size is about 3000 km × 2940 km in zonal
and meridional directions, respectively). Nudging is exerted
every 6 h, which coincides with the frequency of our ERA-
Interim reanalysis data.
We also rely on previous WRF studies in Colombia or in
nearby countries that tested several microphysics, cumulus,
longwave radiation, and planetary boundary layer schemes
(Arregocés et al. 2021; Núñez 2014; García 2014; Posada-
Marín et al. 2018; González-Rojí et al. 2022) to optimize
our CTRL configuration. However, the sensitivity of our
results to an exhaustive panel of WRF schemes is beyond
Topographic Height in m the scope of this study. In consequence, further physical
parametrization schemes considered include the microphys-
ics scheme of the WRF single moment 6-class (Hong and
Lim 2006), the new version of the Rapid Radiative Trans-
fer Model (RRTMG) for shortwave and longwave radiation
Fig. 1 WPS Domain Configuration and study areas – D01 spans (Iacono et al. 2008), the Noah land surface model (Tewari
3000 km × 2940 km in zonal and meridional direction and has a res- et al. 2004), the Monin–Obukhov similarity scheme for the
olution of 30 km. D02 spans 1390 km × 1900 km and has a resolu-
tion of 10 km. Focus regions for analysis are located in the southern surface-layer (Jiménez et al. 2012) and the Yonsei Univer-
Andes and at the Caribbean coast in the north of Colombia sity (YSU) planetary boundary layer (Hong et al. 2006) as
13

A. Manciu et al.
(a) (b)
Fig. 2 a Land cover in Colombia as used in WRF based on USGS-24-category classification and b Extent of land use change in Simulations 2
and 4 (given is the landcover fraction changed in percent) and selection for regional analysis
well as the Kain-Fritsch (KF) scheme cumulus convection forced by “pseudo anthropogenically-unforced”
parametrization (Kain 2004). We use adaptive time-stepping ERA-Interim data for 2009-2011, with standard WRF-
to avoid the model run crashing when calculating too high built-in land cover map
windspeed on steep slopes. (4) WITHOUT_CC_LCC simulation:
We stress here that those model settings (simulation forced by “pseudo anthropogenically-unforced”
months, schemes, boundary conditions, spin-up, nudging) ERA-Interim data for 2009-2011, preindustrial land
are similar to the ones used in other recent studies on the use/land cover
detection or attribution of regional and local climatic events, (5) 100% DEFORESTATION Simulation:
and this standard nudging setting strongly limits the overall forced by ERA-Interim data for 2009-2011 with
model internal variability across simulations (Teklay et al. 100% replacement of forest by cropland and pasture
2019; Wang et al. 2020; Glotfelty et al. 2021; Lu et al. 2021).
Meanwhile, across the simulation period, we find no signifi- Those simulations are performed with the same physical
cant drift in the differences between simulations (< 5%, not parameterization schemes and parameters. Land use cat-
shown) and we perform sensitivity experiments in Sect. 4.4 egories in the baseline simulation (simulation 1) are based
to albedo parametrization and physical schemes to explain on the USGS land-cover classification (24 categories, see
responses and potential biases. Fig. 2a). For Colombia, this means that it is largely cov-
ered by natural vegetation: evergreen broadleaf forest (EBF,
Amazonas, Chocó) and savanna (eastern Llanos). Urban and
2.3 Experimental setup agricultural land is located mostly between the central and
the eastern Andean branch, and in the north at the Caribbean
In total, we conduct five main simulations, each one repre- coast. According to the built-in USGS land cover classifi-
senting a different land cover and/or climate input, for both cation, only a few pixels are classified as urban and built-
domains: up land (Bogotá: 3 grid points; Cartagena, Cali and Bar-
ranquilla: 1 grid point, respectively). The most prominent
(1) CTRL simulation: agricultural land use type is dryland cropland and pasture.
forced by ERA-Interim data for 2009-2011, with The Caribbean coast is dominated by a cropland/woodland
standard WRF-built-in land cover map mosaic.
(2) WITHOUT_LCC simulation: To retrieve LCC effects on climate, in simulation 2, we
forced by ERA-Interim data for 2009-2011, prein- changed all categories representing urban and agricultural
dustrial land use/land cover (potential natural vegeta- land to potential natural vegetation (EBF and grassland)
tion) classes from inspecting coarse preindustrial land cover
(3) WITHOUT_CC simulation: maps suggested by Levavasseur et al. (2012) and Hengl et al.
13

(2018, simulation 2). Each grid cell consists of fractions 2.4 Statistical analysis and visualization
of more than one land cover type. Therefore, we decide to
replace the vegetation properties (minimum and maximum To identify the relative contribution of LCC and CC, the area-
values for albedo, leaf area index, emissivity, roughness weighted mean values of variables over the 3 year-simulation
length, as well as green vegetation fraction, rooting depth period are calculated at the national level as well as for two
and stomatal resistance) in the classes “Urban and Built- focus regions, namely the southern Colombian Andes (76.0°W
Up land”, “Dry cropland and pasture” as well as “Crop- – 74.8°W, 5.2°N – 1.8°N) between the central and eastern cordill-
land/Woodland mosaic” with the properties of the class eras, and the Caribbean coast (76.5°W – 74.8°W, 11°N – 7.6°N)
“Evergreen Broadleaf Forest”. The properties of the classes in northern Colombia. These two regions were chosen because
“Irrigated cropland and pasture”, “Mixed dryland/irrigated historically they were particularly affected by LCC (see Fig. 2a,
cropland and pasture” and “Cropland/Grassland Mosaic” are b). Consequently, they presumably exhibit the highest tempera-
replaced by the vegetation properties of the class “Grass- ture and/or precipitation changes from LCC in the domain.
land” (see Fig. 2b). Simulation 2 to simulation 5 were compared against the
Simulation 3 is forced by near-preindustrial (~ 1900) CTRL simulation. Hence, the effects were calculated as
ERA-Interim data. ‘Pseudo anthropogenically-unforced shown in Table 1.
(PAU)’ ERA-Interim data consisted of reconstructing his- We calculate the statistical significance of each simula-
torical climate as if there would be no anthropogenic global tion pair and associated confidence intervals at 95% using a
warming influence. To achieve this, a three-step approach Mann–Whitney-Wilcoxon non-parametric test, commonly
was employed: used in regional climate studies, without presuming the
(a) Calculate 30-year monthly mean values for the periods shape of a variable distribution.
1900–1929 (~ near-preindustrial) and 1980–2009 (~ present- Further analysis focuses on the ENSO phases. Based on
day) of variables (VAR) needed to force WRF e.g. air tem- the Multivariate ENSO Index (MEI, PSL 2020) and the Oce-
perature, surface pressure, the two wind components and sea anic Niño Index (ONI, CPC 2020), the El Niño study period
surface temperature. 30-year monthly mean values were cal- is defined from June 2009 to March 2010, and La Niña from
culated from the global reanalysis data set ERA-20C, equally July 2010 to May 2011.
provided by the ECMWF.
(b) Calculate the difference between preindustrial and
present-day values of VAR. 3 Results
ΔVAR = VARpresent−day − VARpre−industrial
3.1 Evaluation of model performance
(c) Create for each input variable new ERA-Interim PAU
data at each time step: Since WRF has already been shown to simulate tem-
peratures and precipitation satisfactorily in our model
PAU ERA - Interim = ERA - Interim − ΔVAR domain (Posada-Marín et al. (2018), we used a similar
The underlying assumption is that the regional histori- model configuration and visually evaluated model per-
cal trend for those variables is mostly due to anthropogenic formance by comparing annual temperature and pre-
global warming forcing. Simulation 4 combines the adjust- cipitation to the available observational 30-years aver-
ments made in simulation 2 and 3. The last simulation (sim- age (1981–2010) provided by the Colombian Institute
ulation 5) represents a complete deforestation (100DEF) of Hydrology, Meteorology and Environmental Studies
scenario which is forced by ERA-Interim data (as in simu- (IDEAM 2015). Underlying per-pixel data of the Fig. 3c,
lation 1 and 2) with dryland cropland and pasture replacing d were not publicly available to calculate biases, nor the
present-day EBF in the whole domain. 2009–2011 average (IDEAM, pers.comm.). However,
visually, WRF is representing spatial patterns of annual
temperatures satisfactorily (Fig. 3c). Only in the north-
ern lowlands and between the central and eastern Andes
Table 1 Method for calculating Name of effect Abbreviation Calculation performed

the individual effects of land-
cover and climate change based Historical LCC effect only LCC effect CTRL simulation – without_LCC simulation
on the performed simulation
Historical CC effect only CC effect CTRL simulation – without_CC simulation
runs
Combined effect – CTRL simulation – without_CC_LCC simulation
100% Deforestation 100DEF 100DEF simulation – CTRL simulation
13

A. Manciu et al.
valley, it underestimates mean temperature by about 2 °C. 3.2 Regional impacts of CC and LCC on temperature
WRF simulates peaks higher in altitude (i.e. colder than
reanalysis) and valleys lower in altitude on average (i.e. In our simulations, global warming (CC) since the beginning
warmer than reanalysis), as expected (see e.g. Figure 11 of the last century increases regional temperature in Colom-
lower panel of Posada-Marin et al. 2018). Precipitation bia (+ 0.77 ± 0.02 °C, Fig. 4b) while historical LCC and
is generally overestimated across the whole domain but 100DEF lead to a significant weak cooling (− 0.01 ± 0.01 °C
follows the observed spatial patterns. and − 0.16 ± 0.01 °C, respectively; Fig. 4a, c). Generally,
Fig. 3 Model evaluation of a simulated mean (2009–2011) annual 2 m surface air temperature and b precipitation against c observed mean
(1981–2010) annual 2 m surface air temperature and d precipitation provided by IDEAM. The color scale is the same for both sources
13

these effects are significantly lower at the Caribbean coast deforested areas, where forest was converted to cropland
(+ 0.58 ± 0.02 °C for CC and − 0.04 ± 0.01 °C for histori- and pasture (Fig. 4a). CC effects are significantly stronger at
cal LCC) than in the Andean region (+ 0.65 ± 0.02 °C for higher altitudes (+ 0.17 °C/km above 500 m at the national
CC and − 0.07 ± 0.01 °C for historical LCC). The maps level and + 0.12 °C/km across an Andean transect, r > 0.6 for
show that the cooling effect from historical LCC is mainly both, in elevation-dependent warming vs. height diagram,
located in the Andean region over potentially historically see Fig. 5a, b).
Fig. 4 Changes in 2 m surface air temperature (in °C) due to a LCC effect, b CC effect, c 100% deforestation effect and d combined effect
(LCC + CC). Minimum and maximum values are displayed above and below the color scale
13

A. Manciu et al.
3.3 Regional impacts of CC and LCC on precipitation Caribbean coast. Precipitation increase is also significantly
higher at higher elevations, albeit less significantly than
Historical LCC and 100DEF simulations both reduce temperature (r < 0.3): by + 0.17 mm/day/km above 500 m
mean precipitation in Colombia by -0.08 ± 0.22 mm/day at the national level and by + 0.32 mm/day/km across the
(− 0.7%, not significant p > 0.05). In contrast, precipita- Andean transect (Fig. 5c, d).
tion significantly increases due to CC by 0.98 ± 0.30 mm/
day (+ 9%). Combined effects are marginally smaller 3.4 LCC and CC impacts on temperature
(+ 0.91 ± 0.31 mm/day) with negligible synergistic effects. and precipitation modulated by ENSO
In terms of precipitation response, the Caribbean Coast
is more affected by historical LCC and 100DEF than the When comparing simulated temperature changes in Colom-
Andean region. In the coastal region, LCC decreases pre- bia during ENSO events over the study period (mean over
cipitation by − 0.67 ± 0.12 mm/day (− 7%), while 100DEF 2009–2011), no remarkable differences are found during El
leads to an increase of 0.44 ± 0.13 mm/day (+ 5%). In con- Niño (07-2009/03-2010) and La Niña (06-2010/05-2011;
trast, the CC impact is by a factor of two higher in the see Fig. 7a). The same holds true for the Coast (Fig. 7b).
Andes than at the coast and at the national level. LCC In the Andean region, LCC and CC effects are slightly
and 100DEF both decrease precipitation in the Andes by enhanced in magnitude during La Niña (by − 0.01 ± 0.01 °C
− 0.48 ± 0.10 mm/day (− 4%) and − 0.18 ± 0.08 mm/day –i.e. + 14%– and by + 0.04 ± 0.01 °C –i.e. + 6%– respectively,
(− 1.5%), respectively (see Fig. 6a–d). Deforested areas compared to the full simulation period, Fig. 7c) and damp-
clearly show precipitation reduction (Fig. 6a, c), while CC ened during El Niño (− 0.02 ± 0.01 °C – i.e. − 28%– and
and combined effects show opposite trends, except for the − 0.08 ± 0.01 °C – i.e. − 12%–, respectively).
(c)
(a) E le va tio n d e p e n d e n t w a r m in g d u e to C lim a te C h a n g e
E le v a tio n d e p e n d e n t in c re a s e in d a ily p re c ip ita tio n a b o v e 5 0 0 m
66
Δ 2m surface temperature in °C
1.0
Δ Daily precipitation in mm
1 .0
.. D if d a ily p re c ip ia tio n in m m
.. D if 2 m S u rfa c e Te m p e ra tu re in C
0.60 .5 0
0
y = 0.45 + 0.00017*height y = 1.28 + 0.00017 *height

r = 0.73 r = 0.12
−3
-3
1000 2000 3000 4000
1,000 2,000 H ig h t in m 3,000 4.000
Heigth in m
1000 2000 3000 4000
1,000 2,000 Heigth
H ig h t in m
in m 3,000 4,000
(b)
E le v a tio n d e p e n d e n t w a r m in g f o r la titu d in a l b a n d 6 .5 5 8 0 9 N
(d)
E le v a tio n d e p e n d e n t in c re a s e in d a ily p re c ip ita tio n f o r la titu d in a l b a n d 6 .5 5 8 0 9 N
44
1 .0
1.0
Δ 2m surface temperature in °C
Δ Daily precipitation in mm
33
.. D if 2 m S u rfa c e Te m p e ra tu r e in C
.. D if d a ily p re c ip ia tio n in m m
0.80 .8 22
1
1
0 .6 y = 0.56 + 0.00012*height 0 y = 1.15 + 0.00032*height

0.6 0
r = 0.67 r = 0.22
−1
1000 2000 3000

-1
1000 2000 3000
1,000 2,000
H ig h t in m 3,000 1,000 2,000
H ig h t in m 3,000
Heigth in m Heigth in m
Fig. 5 Relationship between CC-induced temperature/precipitation and altitude. Temperature (a, b) and precipitation (c, d) changes across dif-
ferent altitudes above 500 m on national level (upper panels) and across a longitudinal transect (lower panels)
13

Fig. 6 Same as Fig. 4, but for changes in daily precipitation (in mm)
The CC effect on precipitation depends on the ENSO magnitude (Fig. 7f). At the coast, an opposite reaction
phases: it is weakened by El Niño (+ 0.62 ± 0.31 mm/ can be observed: CC and combined effects are stronger
day) but enhanced by La Niña (+ 1.08 ± 0.42 mm/ during El Niño and weaker during La Niña (Fig. 7e).
day, Fig. 7d). Combined effects respond in the same Here, historical LCC reduces precipitation, especially
pattern, although slightly weaker than the CC effect during La Niña (− 0.84 ± 0.13 mm/day). If the com-
(+ 0.55 ± 0.29 mm/day and + 1.0 ± 0.4 mm/day, for El plete domain would be deforested and replaced by dry-
Niño and La Niña respectively). The Andean region land cropland and pasture, it would result in a general
reacts in a similar pattern, just about twice as strong in decrease in precipitation which would hit the Andes the
13

A. Manciu et al.
(a) (d)
(b) (e)
(c) (f)
Fig. 7 Temperature (a–c) and precipitation (d–f) changes due to LCC and CC in different Colombian regions: a and d Colombian mean, b and e
coastal mean, c and f Andean mean
strongest during La Niña event (− 0.23 ± 0.09 mm/day, 3.5 Regional impacts of historical LCC and 100DEF
Fig. 7f). On the contrary, the 100% deforestation case on surface energy balance
increases precipitation at the coast across all studied
time periods, with its highest impact during La Niña Changes in the surface energy balance following histori-
(+ 0.69 ± 0.19 mm/day, Fig. 7e). cal LCC show a significant increase in ΔSW ↑ of 2.03 W/
m2 (p < 0.05, Fig. 8a), resulting in a relative cooling effect.
13

This represents a contribution of roughly 35% to the total m2, p < 0.05), pointing to a dominance of albedo-driven
magnitude of change (i.e. ΔSW ↑ divided by Δ sum of all cooling.
absolute flux changes). The 100DEF scenario with the replacement of all EBF by
A similar change in magnitude but with an opposite cropland and pasture in the model domain (Fig. 8b) reveals a
sign for ΔL (− 1.92 W/m2, p < 0.05) has a warming effect more drastic picture compared to historical LCC: a stronger
on national level, thereby balancing the cooling from increase in reflected shortwave radiation from the surface
ΔSW ↑ . Δ(H + G) decrease and contribute also to warm- and a smaller reduction in latent heat, which is strongest
ing, albeit the effect is smaller. Together, the heat fluxes on country level. The effect exists also in the focal regions,
account for 50% of the total magnitude of change (same though weaker because these areas are already largely defor-
as above: Δ non-radiative fluxes divided by Δ sum of all ested in the control simulation.
absolute flux changes). Albedo on average significantly rises by 0.008 and 0.04
Downward radiative fluxes are slightly decreased and following historical LCC and 100DEF in Colombia, respec-
contribute around 15% to the overall change. However, tively. At the coast, the albedo increases by 0.045 due to his-
on a local scale, changes in fluxes due to LCC effects are torical LCC and by 0.01 under 100DEF. The Andean region
imbalanced, especially at the coast, where the dominant shows an albedo increase of 0.02, respectively, after LCC
land cover was changed from EBF to a cropland/wood- and complete deforestation.
land mosaic ( ΔSW ↑: + 12.43 W/m2 vs. ΔL : − 9.96 W/m2;
p < 0.05). In the Andean region, where EBF was converted
to dryland cropland and pasture, these two fluxes are simi-
larly imbalanced ( ΔSW ↑: + 6.80 W/m2 vs ΔL : − 4.26 W/ 4 Discussion
This study explores the effect of both land cover change and
(a) LCC effect on different fluxes global warming particularly during ENSO events, as well as
14 their combined effects on surface air temperature, precipi-
12
10
tation and surface energy fluxes, over Colombia, a climate
8 modeling coldspot.
6
4
The performance of the WRF model in simulating general
temperature and precipitation patterns in Colombia is sat-
in W/m2
2
0
-2
isfactory for the purpose of this research (Fig. 3; Sect. 3.1).
-4 An apparent overestimation of precipitation is found com-
-6
-8
paring our WRF CTRL simulation and IDEAsM patterns
-10 (the national observational dataset, see Fig. 3). This can be
-12
explained because ERA-Interim already slightly overesti-
∆SW↑ ∆SW↓ + ∆LW↓ ∆L ∆H+∆G
mates precipitation in Colombia (Posada-Marín et al. 2018),
Colombia Costa Andina
our large-scale climatic input data to WRF. In addition,
Jin et al. (2010) and Teklay et al. (2019) explored differ-
ent model configurations with varying land surface models
(b)
100DEF effect on different fluxes (LSM) in a similar context and found that precipitation is
14 overestimated regardless of the LSM used. Another reason
12
10 for the discrepancy might be the simulation period of only
8 three years of the recent decade.
6
4
4.1 CC effect on temperature and precipitation
in W/m2
2
0
-2
-4 Interestingly, our results on temperature changes substan-
-6
-8
tially contrast with the official climate near-term projections
-10 from the Colombian IDEAM in its Third National Commu-
-12
∆SW↑ ∆SW↓ + ∆LW↓ ∆L ∆H+∆G
nication on Climate change in Colombia (TCNCC, IDEAM
et al. 2017). Historical surface air temperature increases in
Colombia Costa Andina
our study compare well with national averages published by
IDEAM (+ 0.77 °C vs. + 0.8 °C), however, spatial warming
Fig. 8 a LCC and b 100DEF effect on different surface energy fluxes:
Shortwave reflected (ΔSW ↑), downward fluxes (ΔSW ↓ + ΔLW ↓), patterns are very different (Fig. 4b, d). Our simulations show
latent heat flux (ΔL), sensible and ground heat flux (Δ H + ΔG) the highest temperature increases occurring above Andean
13

A. Manciu et al.
cordilleras and higher elevation areas (> 1 °C, dark red in Furthermore, elevation-dependent rainfall changes under
Figs. 4b, d and 5a, b), lower temperature increases in the global warming are poorly studied (Li et al. 2017) and we
northern and western areas and intermediate increases in are only aware of one study in the tropics (Urrutia and Vuille
the south and eastern parts (e.g. Llanos and Amazon Forest). 2009, tropical Andes). This latter study does not show con-
This is in contradiction with IDEAM results where higher sensual patterns of elevation-dependent rainfall changes
elevation areas warm less (1–1.5 °C) and the Llanos or low under global warming, which depends on the latitude or the
elevation areas more but more homogeneously (> 1.5 °C, slope orientation. The authors found that “In general the
Fig. 4b, d). In other means, we show a positive elevation- projected changes in precipitation are not as pronounced as
dependent warming (EDW) while the IDEAM showed the for temperature” and that precipitation is slightly higher in
opposite result (negative EDW). Confusingly, the IDEAM the global warming scenario below 4000 m.a.s.l. and slightly
states in this very same TCNCC, that the “most significant less beyond that elevation (Urrutia and Vuille 2009). Our
increase in mean, minimum and maximum temperatures results show a slightly higher rainfall increase under his-
will be in the Andean region, especially in high elevation torical global warming in Colombian Andean cordilleras vs.
areas” (IDEAM et al. 2017). In the literature, there is grow- other low-elevation areas (Figs. 5b and 6c, d) which is a con-
ing recent evidence that the rate of warming is amplified sistent pattern with IDEAM near-term projections (IDEAM
with elevation (Pepin et al. 2015; Wang et al. 2016; Hock et al. 2017) and some mid-latitude patterns (Tibetan Plateau,
et al. 2019). However, in the tropics, there is a particular Li et al. 2017) but differs from Urrutia and Vuille (2009).
problem with observational data from mountain regions, Finally, our results suggest that CC effects are more domi-
which is both sparse and inhomogeneous (Pepin et al. 2015), nant on national climate in Colombia than LCC effects while
and EDW in the Tropical Andes is controversially discussed LCC affect rainfall substantially more than temperature at
in the literature (Vuille and Bradley 2000; Beniston et al. regional and sub-regional scales, which is consistent with
1997; Hock et al. 2019; Rangwala and Miller; 2012). At least most recent studies (Hock et al. 2019; Quesada et al. 2017a,
three elements tend to reinforce here our results: (1) most 2017b). Moreover, at the subregional level (Coast or Andean
recent literature in the tropics based on observations show sub-region of Colombia), LCC effects are of the same mag-
more warming above high-elevation vs. low-elevation areas nitude on precipitation as CC effects, which urges to sys-
(Rangwala and Miller 2012; Aguilar-Lome et al. 2019), (2) tematically account for land-use and land-cover dynamics
GCMs and RCMs mostly show positive EDW in tropical in any departmental and regional mitigation and adaption
Andes (Rangwala and Miller 2012; Hock et al. 2019), (3) to climate change plans.
WRF reproduces correctly the EDW in low latitude environ-
ments along elevational gradients (Gao et al. 2018; Expósito 4.2 LCC effect on precipitation and temperature
et al. 2015; Lin et al. 2015).
Concerning precipitation changes, global warming inten- A drying over tropical deforested areas is reported through-
sifies the global hydrological cycle (Jia et al. 2019), i.e. the out the literature (Perugini et al. 2017; Coe et al. 2017;
warming of surface air temperature features more evapora- Spracklen and Garcia-Carreras 2015; Lejeune et al. 2015;
tion of water into the atmosphere, thus more water content Saavedra et al. 2020; Sierra et al. 2021; Eiras-Barca et al.
in the air, since warmer air can hold more water vapor. How- 2020) which is in good agreement with our results, although
ever, according to an average of four emission scenarios, the magnitude of precipitation change differs. Trees usually
IDEAM et al. (2017) expect precipitation to decrease over have a deeper rooting depth and are therefore able to access
the Caribbean coast and the Colombian Amazonas (by deep-lying groundwater even in dry periods. Removing
10–40% by 2071–2100 compared to the reference period tropical forests and replacing them by crops and pastures
1976–2005) and to increase in the Andean region. Our results in reduced evaporative leaves and canopies, foliar
results suggest an overall precipitation increase in Colom- density, surface roughness, stronger winds and an increased
bia in response to historical global warming, except for the albedo. Consequently, evapotranspiration and precipitation
Pacific coast. Our findings are comparable to Skansi et al. (due to fewer clouds) are expected to diminish in tropical
(2013), who analyzed station-based time series data from agricultural areas. Furthermore, modeled (Sampaio et al.
1950–2010 in South America. Our findings show a drying 2007; Nobre et al. 1991; Zemp et al. 2017) changes in pre-
along the Pacific coast, which is confirmed at least for the cipitation show that drying is even more pronounced during
southern part by Skansi et al. (2013) but in contradiction the dry seasons and in low latitude areas.
to Carmona and Poveda (2014). In WRF, the decrease in Reviews on GCM and RCM modeling studies with
Western Pacific Colombian coast precipitation in response to a focus on Amazon deforestation report a precipitation
global warming can be explained by less tropospheric winds decrease (Spracklen and Garcia-Carreras 2015; Lejeune
coming from the Pacific which usually advect moisture, et al. 2015; Sampaio et al. 2007). Lejeune et al. (2015) report
driven by less ocean-continent thermal contrast (not shown).
13

a reduction of ~ − 0.7 mm/day (− 5 to − 10%) across 28 incorrectly simulating cooling after tropical deforestation
GCM and RCM experiments, in agreement with Sprack- and thus conducted in the next section a decomposition of
len and Garcia-Carreras (2015), who report a decrease of our results as well as sensitivity experiments modifying
16.5 ± 13% across 44 simulations of complete Amazon albedo parametrization and land surface models.
deforestation. This is similar to our findings of a 4% decrease
in precipitation (− 184 mm/year or − 0.5 mm/day in case 4.3 LCC and CC interactions with ENSO
of total deforestation over the modeling domain) but more
pronounced in magnitude. Studies investigating the interactions between LCC and
Though the comparison with other regional WRF studies ENSO in tropical South America are rather rare. The con-
(Laux et al. 2017; Paul et al. 2016; Takahashi et al. 2017; sensus, however, is that LCC enhances the ENSO signal
Quesada et al. 2017a) shows discrepancies, due to the spa- (Nobre et al. 2009; Beltrán-Przekurat et al. 2012; Bush et al.
tially specific influence of climatic phenomena, they agree 2017; Zhang et al. 2009). Tölle et al. (2017), for example,
that deforestation leads to decreasing precipitation over used the RCM COSMO-CLM and investigated conse-
large-scale deforested areas. quences of abrupt tropical deforestation during the ENSO
Our simulations show that tropical deforestation causes phases in South East Asia and concluded that they “can
a slight cooling at country level and more enhanced cool- amplify the impact of the natural mode ENSO”. Another
ing over deforested grid-cells. Such cooling after deforesta- study by Chapman et al. (2020) on the island of Borneo
tion simulated by default WRF parametrizations is at odds (Indonesia) confirms that deforestation under El Niño condi-
against all observations and climate simulations about the tions resulted in warmer and drier climate than under normal
tropical deforestation impacts on temperature (Perugini conditions. Syktus et al. (2007), whose study domain was
et al. 2017; Lawrence and Vandecar 2015; Jia et al. 2019). Australia, came to the same conclusion. Bush et al. (2017)
The IPCC Special Report on Climate Change and Land suggest that the ENSO signal has been even amplifying since
(Jia et al. 2019) concludes that there is high confidence that humans started cultivating agricultural goods, which was
large-scale deforestation leads to mean biophysical warm- buffered before by natural vegetation. The buffering effect
ing (+ 0.61 ± 0.48 °C) over the entire tropics and especially has been confirmed by Meijide et al. (2018). However, our
over deforested areas. After entire Amazon deforestation, results suggest that it depends on the region, land cover type,
GCMs and RCMs reviewed by Lejeune et al. (2015) simu- ENSO phase and variable whether the LCC effect is ampli-
lated a temperature increase by 0.8 °C on average, whereas fied or not, i.e. there is no or little influence of LCC and
WRF in our study simulates a cooling of − 0.21 ± 0.02 °C CC on temperature during ENSO, whereas precipitation
over Colombian deforested grid-cells in the 100DEF sce- changes are more affected, especially during La Niña and at
nario. Perugini et al. (2017) reviewed eleven modeling stud- the Coast. Beltrán-Przekurat et al. (2012) concluded from
ies in the tropics and found a mean temperature increase of their results that dry periods are enhanced by the occurrence
0.68 °C (forest cropland/grassland) which is in agreement of ENSO, but that the type of land cover conversion also
with three tropical observational studies they investigated plays an important role for the sign of change.
(+ 0.41 °C on average). Temperature changes derived from
satellite observations (+ 1.06 °C in Alkama and Cescatti 4.4 Decomposition of energy balance and model
(2016), + 1.34 °C in Duveiller et al. (2018)) or from forest sensitivity to albedo parametrization and land
change data (+ 0.38 °C in Prevedello et al. (2019) confirm surface models (LSM)
the warming effect of deforestation in the tropics. However,
caution is needed as observational studies do not capture At a regional scale, the energy balance is deeply modified
nonlocal biophysical impacts from deforestation (Winckler by forest losses. In the tropics, a deforestation-driven warm-
et al. 2017). ing caused by decreased latent heat flux is expected to out-
Only a few modeling studies found a cooling as a result balance the albedo-driven cooling. When decomposing the
of tropical deforestation. For instance, Lejeune et al. (2015) energy balance into its single components and comparing it
report a mean increase in surface air temperatures of + 1.2 °C to observed values provided by Duveiller et al. (2018), WRF
across 28 GCM and RCM studies conducted for the Amazon with the coupled Unified Noah LSM (Tewari et al. 2004),
basin with only 3 studies showing cooling out of 28 in total. which is used here, overestimates the change in reflected
Robertson (2019), using the HadGEM2-ES ESM, obtained shortwave radiation from the surface by a factor of 3.5 over
a cooling similar to our study (− 0.08 °C vs. − 0.01 °C) and converted grid-cells (EBF → crop/grass, see Table 2). At
showed with a decomposition of the surface energy balance the same time, the magnitude in latent heat change reduction
that albedo and evapotranspiration responses were strongly is underestimated by 41% in WRF. WRF simulates a latent
biased (overestimated and underestimated, respectively). heat flux reduction of 9.6 W/m2, while a reduction of 20 to
We suspected similar reasons being responsible for WRF 30 W/m2 would be more realistic according to Duveiller
13

A. Manciu et al.
Table 2 Comparison of Simulation ↓/variable → ΔT2 ΔSW ↑ ΔLW ↑ ΔL Δ(H + G)

simulated deforestation
responses in selected variables Duveiller et al. (2018) + 1.34 5.47 7.53 − 16.31 4.21
using different land surface
WRF (Noah, default LSM and albedo) − 0.07 19.02 1.11 − 9.64 − 8.32
models to observed variables
from Duveiller et al. (2018) WRF (Noah, default LSM and adjusted albedo) − 0.03 7.29 2.15 − 2.77 − 5.85
WRF RUC LSM − 0.13 22.84 − 1.44 − 11.77 − 6
WRF 5 layer thermal diffusion LSM − 0.13 22.76 − 0.35 − 1.55 − 14.15
WRF Pleim-Xiu LSM 0.06 − 0.02 − 1.4 0.83 0.53
2 m surface air temperature difference is given in °C, energy fluxes in W/m2. Values from Duveiller et al.
(2018) refer to a mean value for forest—> crop/grass conversions in the tropics, values from model simula-
tions refer to grid-cells with Evergreen Broadleaf Forest—> cropland and pasture conversion
et al. (2018), Nobre et al. (1991) and Snyder et al. (2004) 0.06 °C over deforested grid-cells. Furthermore, a decom-
when forests are converted to cropland and/or pasture. Fur- position of the alternative LSM’s energy balance shows
thermore, sensible heat is decreasing by -8.4 W/m2 in WRF also changes in radiative and non-radiative fluxes that
following deforestation, instead of increasing by + 5 W/ are in stark contrast to observed changes (Duveiller et al.
m2 to + 20 W/m2 as suggested in the literature (Duveiller 2018, see Table 2). Thus, we investigated several sources
et al. 2018; Nobre et al. 1991; Sampaio et al. 2007; Llopart of deforestation-induced temperature biases in WRF,
et al. 2018; Takahashi et al. 2017; Coe et al. 2017; Oliveira which were found in recent tropical studies performed with
et al. 2019). In summary, the most important reason for the WRF in the default configuration (Glotfelty et al. 2021;
deforestation-temperature bias in WRF seems to be the high Sierra et al. 2021).
overestimation of shortwave radiation reflected from the sur- Finally, we stress here that we are confident with WRF
face after deforestation, while biases in the latent/sensible responses to CC (Sect. 4.2) as well as precipitation responses
heat partitioning tend to compensate each other. This is a to LCC: (1) vegetation model properties are the same
coherent finding compared to Sierra et al. (2021) who men- between CC simulations and climatic patterns are well rep-
tioned the WRF overestimation of the reflected shortwave resented by WRF in this region – see Sect. 3.1 and Posada-
radiation and the strong underestimation of the sensible heat Marín et al. (2018); (2) precipitation responses after LCC
flux and evaporation/latent heat after cooling in response to and tropical deforestation (100DEF) are consistent with the
forest loss in the tropical Amazon-Andes transition region. observations in this tropical region (see Sect. 4.3), (3) the
Indeed, Robertson (2019) concluded, that “to reduce biases tend to cancel out: the underestimation of evapotran-
this error the difference between the PFT’s albedo param- spiration loss after deforestation has an opposite effect with
eters should be reduced”. Following this recommendation, respect to the overestimated cooling for precipitation: the
we adjusted the land use category’s albedo parameters former tends to underestimate the magnitude of the reduced
and ran new WRF simulations. The default albedo ranges precipitation response (more evapotranspiration leads to
from 0.17–0.23 for dryland cropland and pasture, from more clouds and in turn more precipitation) while the latter
0.19–0.23 for grassland and has a constant value of 0.12 tends to overestimate it (cool air tends to hold less water
for EBF. Observed values for albedo indicate that 0.13 to thus less precipitation, Mahmood et al. (2014); (4) across
0.19 for crops, 0.13 to 0.23 for grassland and 0.11 to 0.15 the different WRF land cover schemes tested, precipitation
for tropical EBF are more appropriate (Cescatti et al. 2012, responses are not substantially affected and are systemati-
Hänchen 2017; Prentice et al. 1992; Giambelluca et al. cally negative after deforestation (not shown).
1997). After running simulation 1 and 2 again with these
adjusted values (e.g. using minimum albedo of 0.11 and
a maximum albedo of 0.15 for EBF), ΔSW ↑ is now only 5 Conclusion
slightly overestimated (33%). However, the underestima-
tion of ΔL is amplified (by 83%, see Table 2). Colombian climate regional modelling is challenging in
Additional runs were conducted to explore the model’s several respects: it has a highly uneven and mountainous
sensitivity to the LSM used. Table 2 shows that the cool- terrain highlighting the need for high-resolution model-
ing over deforested pixels is even more enhanced when ling, it is a regional climate modelling coldspot (very few
applying the Rapid Update Cycle (RUC) LSM (− 0.13 °C, studies), it is highly vulnerable territory to climate change
Benjamin et al. 2004) and the 5-Layer thermal diffusion and biodiversity loss and thus, it calls for an urgent need
scheme (− 0.13 °C, Dudhia 1996). Only the Pleim-Xiu of hydroclimatic assessments (Poveda et al. 2011). In
LSM (Pleim and Xiu 1995) simulates a warming of this study, we quantified and disentangled the impact of
13

historical LCC and CC on Colombian climate perform- this first-of-its-kind attribution study at high-resolution
ing WRF simulations during ENSO events (2009–2011), (10 km) analyzing global warming and land-cover changes
identifying the most dominant forcing and analyzing the impacts on climate in this region. Although our simulation
impacts of historical LCC on the surface energy balance period is limited (2009–2011), our results on the whole
given several model configurations. period are likely to represent the simulated impact of his-
LCC reduces precipitation everywhere over deforested torical land cover changes and global warming on Colom-
areas, but especially in the Caribbean coast in the north. bian climate during a climatological period, albeit slightly
Conversely, CC increases precipitation, specifically in the skewed towards La Niña.
Andean region. LCC effects are found to be strong driv- To avoid obvious temperature biases in WRF simula-
ers of sub-regional precipitation changes: dampening CC tions, in-depth analysis and further development of land
effects magnitude by 72% and 24% on average in Coastal cover classification and the representation of vegetation
Caribbean and Andean regions, respectively. Combining properties used by WRF is necessary. This includes a more
historical LCC and CC, temperature increases by 0.76 °C, realistic and up-to-date representation of tropical land
especially in high altitudes. Conflicting with most other cover data. USGS land use classification in WRF stand-
modeling studies and observations, WRF simulates a tem- ard simulations dates back to the early 90s. Since then,
perature decrease following historical LCC and 100DEF, tropical deforestation and land use rose tremendously. A
which is intensified over deforested areas. thorough review of vegetation properties parametrization
Our results suggest that CC effects are more dominant is necessary to assure a good representation of tropical
on regional climate in Colombia than LCC effects but both land–atmosphere feedbacks. We implemented alternative
effects are comparable in magnitude over deforested areas albedo values for Colombia’s most important land use cat-
or at the subregional level. La Niña tends to enhance the egories and tested its sensitivity, which showed a substan-
studied effects across the country, whereas El Niño weak- tial improvement of simulated reflected shortwave radia-
ens them, except at the coast where the pattern is mostly tion. However, parametrizations of e.g. roughness length
inverted. In the Andes, La Niña tends to intensify LCC and and stomatal resistance, important parameters which drive
CC effects roughly to a factor of two compared to El Niño evapotranspiration, were not investigated, potentially
impacts in both cases. explaining that simulated latent heat fluxes were even
Our simulations indicate a serious bias in WRF con- more biased. To understand the role of the ENSO phases
cerning an incorrect simulation of cooling after tropi- better, longer simulation studies are needed, in order to
cal deforestation, recently found as well in other tropi- have a more appropriate climatological period and an aver-
cal regions (Glotfelty et al. 2021; Sierra et al. 2021). We age of more ENSO events. In addition, a follow-up study
further investigate this apparent bias because of its great could investigate climate extreme indices like maximum
importance for climate studies using WRF with default daily temperature and maximum daily precipitation to
parametrization in tropical contexts and probably beyond. provide evidence for LCC and CC impacts on extreme
We show here that WRF coupled with the Unified Noah weather events. Although land use data has been used for
LSM, RUC LSM and the 5-layer thermal diffusion LSM regional climate simulations in the Neotropics (e.g. Sierra
overestimates reflected surface shortwave radiation et al. 2021; Saavedra et al. 2020) and although a more
response and underestimates evapotranspiration-driven recent update land use data already exists (years 2000s,
warming in the tropics which leads us to the assumption Eva et al. 2004), a newer high-resolution land-cover data-
that vegetation properties are not appropriately para- set for Colombia spanning a climatological period would
metrized for tropical regions in all investigated LSMs in represent a great input for more robust climate and hydro-
WRF. We therefore recommend using default vegetation logical assessment in Colombia. Finally, it should be
properties of WRF with caution for temperature assess- noted that the official climatic projections by the TCNCC
ments. Improved model representation of vegetation prop- (IDEAM et al. 2017) contains methodological shortcom-
erties and more validation data in tropical areas are needed ings, inconsistencies and need urgent reassessment by
to provide more robust and confident climate projections at the national communities of meteorology and climatology
regional-to-local scale. We stress here that, although our (Arias et al. 2022).
study has limits concerning the limited sensitivity experi-
ments across physical schemes or imprecise representa-
Funding Open Access funding provided by Colombia Consortium. We
tion of land-cover characteristics, the model settings are deeply acknowledge the services of the High Performance Computing
tuned to limit internal model variability maximizing the center hosted at Universidad del Rosario (Colombia) and Advanced
signals response, following similar WRF climatic studies Computer Laboratory for Research as well as the useful technical help
(see Sects. 2.2 and 2.4). Meanwhile, we provide statis- provided (https://www.urosar io.edu.co/Labor atorio-Comput acion-
Avanzada-Investigacion/Infraestructura/"). A.Manciu acknowledges
tical significance to ensure robust messages throughout
13

A. Manciu et al.
TUM university and the Hanns-Seidel-Stiftung for the Mobility Northern Colombia during 2016 dry season. Dyn Atmos Oceans
Fellowship as well as the Universidad del Rosario for supervision. 96:101261. https://doi.org/10.1016/j.dynatmoce.2021.101261
B.Quesada acknowledges Climat AmSud program (project 21-CLI- Ávila Á, Guerrero F, Escobar Y, Justino F (2019) Recent precipitation
MAT-13), Colombian-french researcher association COLIFRI in the trends and floods in the Colombian Andes. Water. https://d oi.o rg/
framework of the French FSPI fund on Renewable Energy ecosystem 10.3390/w11020379
in Orinoco Region project and the Starting Grant of Universidad del Beltrán-Przekurat A, Pielke RA Sr, Eastman JL, Coughenour MB
Rosario for funding this work. (2012) Modelling the effects of land-use/land-cover changes on
the near-surface atmosphere in southern South America. Int J
Data availability For this research, we did not use any new data for Climatol 32:1206–1225. https://doi.org/10.1002/joc.2346
simulating the studied effects. The ERA-Interim reanalysis dataset from Beniston M, Diaz HF, Bradley RS (1997) Climatic change at high
Dee et al. (2011), with which we forced the model, and the ERA-20C elevation sites: an overview. Clim Change 36:233–251. https://
dataset from Dee et al. (2016), that was used to create the forcing data doi.org/10.1023/a:1005380714349
without Global Warming influence, is available for download from Benjamin SG et al (2004) An hourly assimilation-forecast cycle: the
the European Center for Medium-Range Weather Forecasts (ECMWF, RUC. Mon Weather Rev 132:495–518
available at https://www.ecmwf.int/en/forecasts/datasets/browse-reana Bowden JH, Otte TL, Nolte CG, Otte MJ (2012) Examining interior
lysis-datasets) upon creation of a cost-free user account. The newly grid nudging techniques using two-way nesting in the WRF
simulated and processed data in this research is available in the Center model for regional climate modeling. J Clim 25:2805–2823.
for Open Science repository under the following link: https://osf.io/ https://doi.org/10.1175/jcli-d-11-00167.1
pcrk5/?view_only=5f80e189351b4845bd9df72b28ed87fe. Bush MB, Correa-Metrio A, van Woesik R, Shadik CR, McMichael
CNH (2017) Human disturbance amplifies Amazonian El Nino-
Declarations southern oscillation signal. Glob Chang Biol 23:3181–3192.
https://doi.org/10.1111/gcb.13608
Conflict of interest The authors declare no conficts of interest or com- Caldwell P, Chin H-NS, Bader DC, Bala G (2009) Evaluation of a
peting interests. WRF dynamical downscaling simulation over California. Clim
Chang 95:499–521. https://doi.org/10.1007/s10584-009-9583-5
Open Access This article is licensed under a Creative Commons Attri- Carmona AM, Poveda G (2014) Detection of long-term trends in
bution 4.0 International License, which permits use, sharing, adapta- monthly hydro-climatic series of Colombia through empirical
tion, distribution and reproduction in any medium or format, as long mode decomposition. Clim Chang 123:301–313. https://doi.
as you give appropriate credit to the original author(s) and the source, org/10.1007/s10584-013-1046-3
provide a link to the Creative Commons licence, and indicate if changes Cescatti A et al (2012) Intercomparison of MODIS albedo retrievals
were made. The images or other third party material in this article are and in situ measurements across the global FLUXNET net-
included in the article's Creative Commons licence, unless indicated work. Remote Sens Environ 121:323–334. https://doi.org/10.
otherwise in a credit line to the material. If material is not included in 1016/j.rse.2012.02.019
the article's Creative Commons licence and your intended use is not Chapman S et al (2020) Compounding impact of deforestation on
permitted by statutory regulation or exceeds the permitted use, you will Borneo’s climate during El Niño events. Environ Res Lett.
need to obtain permission directly from the copyright holder. To view a https://doi.org/10.1088/1748-9326/ab86f5
copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Chotamonsak C, Salathe EP, Kreasuwan J, Chantara S (2012) Evalu-
ation of precipitation simulations over Thailand using a WRF
regional climate model. Chiang Mai J Sci 39:623–638
Claussen M, Brovkin V, Ganopolski A (2001) Biogeophysical versus
biogeochemical feedbacks of large-scale land cover change.
References Geophys Res Lett 28:1011–1014. https://d oi.o rg/1 0.1 029/
2000gl012471
Aguilar-Lome J, Espinoza-Villar R, Espinoza J-C, Rojas-Acuña J, Wil- Coe MT, Brando PM, Deegan LA, Macedo MN, Neill C, Silvério
lems BL, Leyva-Molina W-M (2019) Elevation-dependent warm- DV (2017) The forests of the Amazon and cerrado moderate
ing of land surface temperatures in the Andes assessed using regional climate and are the key to the future. Trop Conserv
MODIS LST time series (2000–2017). Int J Appl Earth Obs Sci. https://doi.org/10.1177/1940082917720671
Geoinf 77:119–128. https://doi.org/10.1016/j.jag.2018.12.013 Comisión Económica para América Latina y el Caribe C (2012)
Alkama R, Cescatti A (2016) Biophysical climate impacts of recent Valoración de daños y pérdidas. Ola invernal en Colombia
changes in global forest cover. Science 351:600–604. https://doi. 2010–2011. Bogotá: Misión BID - Cepal
org/10.1126/science.aac8083 CPC (2020) Historical El Nino / La Nina episodes (1950-present)
Arias PA, Villegas LD, Mesa OJ, Pabón JD, Arango J, Santiago, [Online]. Boulder, Colorado, USA: NOAA. National Weather
Arango-Aramburo S, Armenteras D, Barahona R, Berrouet L, Service. Climate Prediction Center. Available: https://origin.
Barco J, Cardona Y, Carvajal-Escobar Y,Ceballos-Bonilla LI, cpc.n cep.n oaa.g ov/p roduc ts/a nalys is_m onito ring/e nsost uff/
Cerón WL, Escobar JH, González C, Hoyos IC, Hoyos N, Jara- ONI_v5.php. Accessed 10 Oct 2019
millo C, Jiménez-Sánchez G, Martínez JA, Poveda G, Quesada Davin EL, de Noblet-Ducoudré N (2010) Climatic impact of global-
B, Quintero-Vallejo E, Rendón Am, Ruiz-Carrascal D, Salazar A, scale deforestation: radiative versus nonradiative processes. J
Salazar JF, Sánchez A, Vieira SC, Villegas JC, Villegas-Palacio Clim 23:97–112. https://doi.org/10.1175/2009jcli3102.1
C, Zuluaga MD (2022) Implicaciones metodológicas e incon- Dee DP et al (2011) The ERA-interim reanalysis: configuration and
sistencias de la Tercera Comunicación Nacional sobre Cambio performance of the data assimilation system. Q J R Meteorol
Climático de Colombia. Revista de la Academia Colombiana de Soc 137:553–597. https://doi.org/10.1002/qj.828
Ciencias Exactas Físicas y Naturales 46(180):769–794. https:// Dee DP et al (2016) ERA-20C: an atmospheric reanalysis of the
doi.org/10.18257/raccefyn.1705 twentieth century. J Clim 29:4083–4097. https://doi.org/10.
Arregocés HA, Rojano R, Restrepo G (2021) Sensitivity analysis of 1175/jcli-d-15-0556.1
planetary boundary layer schemes using the WRF model in
13

Dudhia J (1996) A mulit-layer soil temperature model for MM5. Hengl T, Walsh MG, Sanderman J, Wheeler I, Harrison SP, Prentice
The Sixth PSU/NCAR Mesoscale Model Users' Workshop, IC (2018) Global mapping of potential natural vegetation: an
Boulder, Colorado assessment of machine learning algorithms for estimating land
Duveiller G, Hooker J, Cescatti A (2018) The mark of vegetation potential. PeerJ 6:e5457. https://doi.org/10.7717/peerj.5457
change on earth’s surface energy balance. Nat Commun 9:679. Hock R et al (2019) High mountain areas [Online]. Available: https://
https://doi.org/10.1038/s41467-017-02810-8 www.ipcc.ch/site/assets/uploads/sites/3/2019/11/06_SROCC_
Eiras-Barca J, Dominguez F, Yang Z, Chug D, Nieto R, Gimeno L, Ch02_FINAL.pdf. Accessed 27 Aug 2020
Miguez-Macho G (2020) Changes in South American hydro- Hong S-Y, Noh Y, Dudhia J (2006) A new vertical diffusion pack-
climate under projected Amazonian deforestation. Ann N Y age with an explicit treatment of entrainment processes. Mon
Acad Sci. https://doi.org/10.1111/nyas.14364 Weather Rev 134:2318–2341. https://doi.org/10.1175/mwr31
Espinoza JC, Garreaud R, Poveda G, Arias PA, Molina-Carpio J, 99.1
Masiokas M, Viale M, Scaff L (2020) Hydroclimate of the Hong SY, Lim JOJ (2006) The WRF single–moment 6–class micro-
Andes part I: main climatic features. Front Earth Sci. https:// physics scheme (WSM6). J Korean Meteor Soc 42: 129–151.
doi.org/10.3389/feart.2020.00064 https://w ww2.m mm.u car.e du/w rf/u sers/p hys_refs/M ICRO_
Eva HD et al (2004) A land cover map of South America. Global PHYS/WSM6.pdf
Change Biolology. https://doi.org/10.1111/j.1529-8817.2003. Hoyos N, Escobar J, Restrepo JC, Arango AM, Ortiz JC (2013) Impact
00774.x of the 2010–2011 La Niña phenomenon in Colombia, South
Expósito FJ, González A, Pérez JC, Díaz JP, Taima D (2015) High- America: the human toll of an extreme weather event. Appl
resolution future projections of temperature and precipitation Geogr 39:16–25. https://doi.org/10.1016/j.apgeog.2012.11.018
in the Canary islands. J Clim 28:7846–7856. https://doi.org/ Iacono MJ, Delamere JS, Mlawer EJ, Shephard MW, Clough SA,
10.1175/jcli-d-15-0030.1 Collins WD (2008) Radiative forcing by long-lived greenhouse
FAO (2015) Deforestation in Colombia [Online]. Available: www. gases: calculations with the AER radiative transfer models. J
globalforestwatch.org. Accessed 19 Feb 2020 Geophys Res. https://doi.org/10.1029/2008jd009944
Findell KL, Berg A, Gentine P, Krasting JP, Lintner BR, Maly- IDEAM (2015) Atlas Climatológico de Colombia – Interactivo
shev S, Santanello JA Jr, Shevliakova E (2017) The impact [Online]. Bogotá, D.C., Colombia. Accessed 3 Aug 2019
of anthropogenic land use and land cover change on regional IDEAM (2017) Resultados Monitoreo de la deforestacion [Online].
climate extremes. Nat Commun 8:989. https://doi.org/10.1038/ Available: http://www.ideam.gov.co/documents/24277/0/Prese
s41467-017-01038-w ntaci%9 7n+E strat egia+I nteg r al+d e+C ontro l+a+l a+D efor
Gao Y, Fu JS, Drake JB, Liu Y, Lamarque JF (2012) Projected changes estaci%9 7n/1 73f79 bf-3 e68-4 cbc-9 387-8 0123d 09b5e 2. Accessed
of extreme weather events in the eastern United States based on 29 June 2019
a high resolution climate modeling system. Environ Res Lett. IDEAM (2018) Resulatdos monitoreo de la deforestacion [Online].
https://doi.org/10.1088/1748-9326/7/4/044025 Available: http://documentacion.ideam.gov.co/openbiblio/bvirt
Gao Y, Chen F, Lettenmaier DP, Xu J, Xiao L, Li X (2018) Does ual/023835/Result ados_Monitoreo_Deforestacion_2017.pdf.
elevation-dependent warming hold true above 5000 m elevation? Accessed 29 June 2019
Lessons from the Tibetan Plateau. npj Clim Atmos Sci. https:// IDEAM (2019) Resultados Monitoreo de la deforestacion [Online].
doi.org/10.1038/s41612-018-0030-z Available: https://p idama zonia.c om/s ites/d efaul t/f iles/l ista
García J (2014) Validación de la capacidad del modelo WRF “Weather do/Actualizacion_cifras2018FINALDEFORESTACION.pdf.
Research and Forecasting” para pronosticar lluvia intensa, Accessed 3 Oct 2019
usando el método orientado a objetos y tablas de contingencia. IDEAM, PNUD, MADS, DNP, CANCILLERÍA (2017) Tercera
[Master's Thesis, Universidad Nacional de Colombia] Comunicación Nacional De Colombia a La Convención Marco
Giambelluca TW, Hölscher D, Bastos TX, Frazão RR, Nullet MA, De Las Naciones Unidas Sobre Cambio Climático (CMNUCC).
Ziegler AD (1997) Observations of Albedo and radiation bal- Tercera Comunicación Nacional de Cambio Climático. Bogotá,
ance over postforest land surfaces in the eastern Amazon basin. Colombia
J Clim 10:919–928. https://doi.org/10.1175/1520-0442(1997) IPCC (2018) Summary for Policymakers. In: Masson-Delmotte V et al
010%3c0919:Ooaarb%3e2.0.Co;2 (eds) Global Warming of 1.5 °C. An IPCC special report on the
Glotfelty T, Ramírez-Mejía D, Bowden J, Ghilardi A, West JJ (2021) impacts of global warming of 1.5 °C above pre-industrial levels
Limitations of WRF land surface models for simulating land and related global greenhouse gas emission pathways, in the con-
use and land cover change in Sub-Saharan Africa and develop- text of strengthening the global response to the threat of climate
ment of an improved model (CLM-AF v. 1.0). Geosci Model change, sustainable development, and efforts to eradicate poverty
Dev 14:3215–3249. https://doi.org/10.5194/gmd-14-3215-2021 Jia G et al (2019) Land–climate interactions. Climate change and land:
González-Rojí SJ, Messmer M, Raible CC, Stocker TF (2022) Sensitiv- an IPCC special report on climate change, desertification, land
ity of precipitation in the highlands and lowlands of Peru to phys- degradation, sustainable land management, food security, and
ics parameterization options in WRFV3.8.1. Geosci Model Dev greenhouse gas fluxes in terrestrial ecosystems. Intergovernmen-
15(7):2859–2879. https://doi.org/10.5194/gmd-15-2859-2022 tal Panel on Climate Change
Hänchen L (2017) Modelling impacts of land-cover changes on tem- Jiménez PA, Dudhia J, González-Rouco JF, Navarro J, Montávez JP,
perature Estimation of direct biophysical and biochemical effects García-Bustamante E (2012) A revised scheme for the WRF sur-
using a semi-empirical approach. Master of Science, Leopold- face layer formulation. Mon Weather Rev 140:898–918. https://
Franzens Universität Innsbruck doi.org/10.1175/mwr-d-11-00056.1
Hansen MC et al (2013) High-resolution global maps of 21st-century Jiménez-Esteve B, Udina M, Soler MR, Pepin N, Miró JR (2018) Land
forest cover change. Science 342:850–853. https://doi.org/10. use and topography influence in a complex terrain area: a high
1126/science.1244693 resolution mesoscale modelling study over the eastern Pyrenees
Heikkilä U, Sandvik A, Sorteberg A (2010) Dynamical downscal- using the WRF model. Atmos Res 202:49–62. https://d oi.o rg/1 0.
ing of ERA-40 in complex terrain using the WRF regional cli- 1016/j.atmosres.2017.11.012
mate model. Clim Dyn 37:1551–1564. https://doi.org/10.1007/
s00382-010-0928-6
13

A. Manciu et al.
Jin J, Miller NL, Schlegel N (2010) Sensitivity study of four land Paul S, Ghosh S, Oglesby R, Pathak A, Chandrasekharan A, Ram-
surface schemes in the WRF model. Adv Meteorol 2010:1–11. sankaran R (2016) Weakening of Indian summer monsoon
https://doi.org/10.1155/2010/167436 rainfall due to changes in land use land cover. Sci Rep 6:32177.
Kain JS (2004) The Kain-Fritsch convective parameterization: an https://doi.org/10.1038/srep32177
update. J Appl Meteorol 43:170–181. https://doi.org/10.1175/ Pepin N et al (2015) Elevation-dependent warming in mountain
1520-0450(2004)043%3c0170:Tkcpau%3e2.0.Co;2 regions of the world. Nat Clim Chang 5:424–430. https://doi.
Laux P, Nguyen PNB, Cullmann J, Kunstmann H (2017) Impacts of org/10.1038/nclimate2563
land-use/land-cover change and climate change on the regional Perugini L, Caporaso L, Marconi S, Cescatti A, Quesada B, de Nob-
climate in the central Vietnam. Land use and climate change let-Ducoudré N, House JI, Arneth A (2017) Biophysical effects
interactions in central Vietnam. Springer on temperature and precipitation due to land cover change.
Lawrence D, Vandecar K (2015) Effects of tropical deforestation on Environ Res Lett. https://doi.org/10.1088/1748-9326/aa6b3f
climate and agriculture. Nat Clim Chang 5:27–36. https://doi. Pleim JE, Xiu A (1995) Development and testing of a surface flux
org/10.1038/nclimate2430 and planetary boundary layer model for application in mes-
Lee S-J, Berbery EH (2012) Land cover change effects on the climate oscale models. J Appl Meteorol 34:16–32. https://doi.org/10.
of the La Plata basin. J Hydrometeorol 13:84–102. https://doi. 1175/1520-0450-34.1.16
org/10.1175/jhm-d-11-021.1 Posada-Marín JA, Rendón AM, Salazar JF, Mejía JF, Villegas JC
Lejeune Q, Davin EL, Guillod BP, Seneviratne SI (2015) Influ- (2018) WRF downscaling improves ERA-Interim representa-
ence of Amazonian deforestation on the future evolution of tion of precipitation around a tropical Andean valley during El
regional surface fluxes, circulation, surface temperature and Niño: implications for GCM-scale simulation of precipitation
precipitation. Clim Dyn 44:2769–2786. https://doi.org/10.1007/ over complex terrain. Clim Dyn 52:3609–3629. https://doi.org/
s00382-014-2203-8 10.1007/s00382-018-4403-0
Levavasseur G, Vrac M, Roche DM, Paillard D (2012) Statistical mod- Poveda G, Álvarez DM, Rueda ÓA (2011) Hydro-climatic variability
elling of a new global potential vegetation distribution. Environ over the Andes of Colombia associated with ENSO: a review
Res Lett. https://doi.org/10.1088/1748-9326/7/4/044019 of climatic processes and their impact on one of the Earth’s
Li X, Wang L, Guo X, Chen D (2017) Does summer precipitation trend most important biodiversity hotspots. Clim Dyn 36:2233–
over and around the Tibetan Plateau depend on elevation? Int J 2249. https://doi.org/10.1007/s00382-010-0931-y
Climatol 37:1278–1284. https://doi.org/10.1002/joc.4978 Powers JG et al (2017) The weather research and forecasting
Lin C-Y, Chua Y-J, Sheng Y-F, Hsu H-H, Cheng C-T, Lin Y-Y (2015) model: overview, system efforts, and future directions. Bull
Altitudinal and latitudinal dependence of future warming in Tai- Am Meteor Soc 98:1717–1737. https:// d oi. o rg/ 1 0. 1 175/
wan simulated by WRF nested with ECHAM5/MPIOM. Int J bams-d-15-00308.1
Climatol 35:1800–1809. https://doi.org/10.1002/joc.4118 Prentice IC, Cramer W, Harrison SP, Leemans R, Monserud RA, Solo-
Llopart M, Reboita M, Coppola E, Giorgi F, da Rocha R, de Souza D mon AM (1992) Special paper: a global biome model based on
(2018) Land use change over the amazon forest and its impact plant physiology and dominance, soil properties and climate. J
on the local climate. Water. https://doi.org/10.3390/w10020149 Biogeogr 19:117–134. https://doi.org/10.2307/2845499
Lu S, Guo W, Xue Y, Huang F, Ge J (2021) Simulation of summer Prevedello JA, Winck GR, Weber MM, Nichols E, Sinervo B (2019)
climate over Central Asia shows high sensitivity to differ- Impacts of forestation and deforestation on local temperature
ent land surface schemes in WRF. Clim Dyn 57:2249–2268. across the globe. PLoS One 14:e0213368. https://doi.org/10.
https://doi.org/10.1007/s00382-021-05876-9 1371/journal.pone.0213368
Mahmood R et al (2014) Land cover changes and their biogeophysi- PSL (2020) Multivariate ENSO index version 2 (MEI.v2) [Online].
cal effects on climate. Int J Climatol 34:929–953. https://doi. NOAA. Physical Sciences Laboratory. Available: https://psl.
org/10.1002/joc.3736 noaa.gov/enso/mei/. Accessed 4 Oct 2019
Meijide A, Shekhar Badu C, Moyano F, Tiralla N, Gunawan D, Quesada B, Arneth A, de Noblet-Ducoudré N (2017a) Atmospheric,
Knohl A (2018) Impact of forest conversion to oil palm and radiative, and hydrologic effects of future land use and land cover
rubber plantations on microclimate and the role of the 2015 changes: a global and multimodel climate picture. J Geophys Res
ENSO event. Agric For Meteorol 252:208–219. https://doi.org/ Atmos 122:5113–5131. https://doi.org/10.1002/2016jd025448
10.1016/j.agrfor met.2018.01.013 Quesada B, Devaraju N, Noblet-Ducoudré N, Arneth A (2017b) Reduc-
Nobre CA, Sellers PJ, Shukla J (1991) Amazonian deforestation and tion of monsoon rainfall in response to past and future land use
regional climate change. J Clim 4:957–988. https://doi.org/10. and land cover changes. Geophys Res Lett 44:1041–1050. https://
1175/1520-0442(1991)004%3c0957:ADARCC%3e2.0.CO;2 doi.org/10.1002/2016gl070663
Nobre P, Malagutti M, Urbano DF, de Almeida RAF, Giarolla E Rangwala I, Miller JR (2012) Climate change in mountains: a
(2009) Amazon deforestation and climate change in a coupled review of elevation-dependent warming and its possible
model simulation. J Clim 22:5686–5697. https://doi.org/10. causes. Clim Change 114:527–547. https://doi.org/10.1007/
1175/2009jcli2757.1 s10584-012-0419-3
Núñez G (2014) Análisis del efecto de la modificación de la geodata- Ren HL, Zuo J, Deng Y (2018) Statistical predictability of Niño indices
base (uso del suelo, elevación) en el clima simulado por Weather for two types of ENSO. Clim Dyn 52(9–10):5361–5382. https://
Research and Forecasting - WRF para la Sabana de Bogotá. doi.org/10.1007/s00382-018-4453-3
[Master's Thesis, Universidad Nacional de Colombia] Robertson E (2019) The local biophysical response to land-use change
Oliveira G, Brunsell NA, Moraes EC, Shimabukuro YE, Santos TV, in HadGEM2-ES. J Clim 32:7611–7627. https://d oi.o rg/1 0.1 175/
Randow C, Aguiar RG, Aragao LEOC (2019) Effects of land- jcli-d-18-0738.1
cover changes on the partitioning of surface energy and water Saavedra M, Junquas C, Espinoza JC, Silva Y (2020) Impacts of
fluxes in Amazonia using high-resolution satellite imagery. topography and land use changes on the air surface temperature
Ecohydrology. https://doi.org/10.1002/eco.2126 and precipitation over the central Peruvian Andes. Atmos Res.
Pan Y et al (2011) A large and persistent carbon sink in the world’s https://doi.org/10.1016/j.atmosres.2019.104711
forests. Science 333:988–993. https://doi.org/10.1126/science. Sampaio G, Nobre C, Costa MH, Satyamurty P, Soares-Filho BS, Car-
1201609 doso M (2007) Regional climate change over eastern Amazonia
13

caused by pasture and soybean cropland expansion. Geophys Res of rainfall and temperature over Lake Tana Basin, Ethiopia. Heli-
Lett. https://doi.org/10.1029/2007gl030612 yon 5:e02469. https://doi.org/10.1016/j.heliyon.2019.e02469
Sierra JP et al (2021) Deforestation impacts on Amazon-Andes Tewari M, Chen F, Wang W, Dudhia J, LeMone MA, Mitchell K, Ek
hydroclimatic connectivity. Clim Dyn. https://doi.org/10.1007/ M, Gayno G, Wegiel J, Cuenca RH (2004) Implementation and
s00382-021-06025-y verification of the unified NOAH land surface model in the WRF
Skamarock WC et al (2008) A description of the advanced research model. 20th conference on weather analysis and forecasting/16th
WRF version 3. NCAR Tech Note NCAR/TN-475+STR. https:// conference on numerical weather prediction.
doi.org/10.5065/D68S4MVH Tölle MH, Engler S, Panitz H-J (2017) Impact of abrupt land cover
Skansi MDLM et al (2013) Warming and wetting signals emerging changes by tropical deforestation on southeast Asian climate
from analysis of changes in climate extreme indices over South and agriculture. J Clim 30:2587–2600. https://d oi.org/10.1175/
America. Glob Planet Chang 100:295–307. https://doi.org/10. jcli-d-16-0131.1
1016/j.gloplacha.2012.11.004 Urrutia R, Vuille M (2009) Climate change projections for the tropical
Snyder PK, Foley JA, Hitchman MH, Delire C (2004) Analyzing Andes using a regional climate model: temperature and precipita-
the effects of complete tropical forest removal on the regional tion simulations for the end of the 21st century. J Geophys Res.
climate using a detailed three-dimensional energy budget: an https://doi.org/10.1029/2008jd011021
application to Africa. J Geophys Res Atmos. https://doi.org/10. Vuille M, Bradley RS (2000) Mean annual temperature trends and
1029/2003jd004462 their vertical structure in the tropical Andes. Geophys Res Lett
Soares PMM, Cardoso RM, Miranda PMA, de Medeiros J, Belo- 27:3885–3888. https://doi.org/10.1029/2000gl011871
Pereira M, Espirito-Santo F (2012) WRF high resolution Wang W et al (2017) WRF-ARW V3: User’s Guide.
dynamical downscaling of ERA-interim for Portugal. Clim Dyn Wang Q, Fan X, Wang M (2016) Evidence of high-elevation amplifi-
39:2497–2522. https://doi.org/10.1007/s00382-012-1315-2 cation versus Arctic amplification. Sci Rep 6:19219. https://doi.
Song XP, Hansen MC, Stehman SV, Potapov PV, Tyukavina A, org/10.1038/srep19219
Vermote EF, Townshend JR (2018) Global land change from Wang X, Tolksdorf V, Otto M, Scherer D (2020) WRF-based dynami-
1982 to 2016. Nature 560:639–643. https://doi.org/10.1038/ cal downscaling of ERA5 reanalysis data for high mountain Asia:
s41586-018-0411-9 towards a new version of the high Asia refined analysis. Int J
Spracklen DV, Garcia-Carreras L (2015) The impact of Amazonian Climatol 41:743–762. https://doi.org/10.1002/joc.6686
deforestation on Amazon basin rainfall. Geophys Res Lett Winckler J, Reick CH, Pongratz J (2017) Robust identification of
42:9546–9552. https://doi.org/10.1002/2015gl066063 local biogeophysical effects of land-cover change in a global
Sun X, Xue M, Brotzge J, McPherson RA, Hu X-M, Yang X-Q (2016) climate model. J Clim 30:1159–1176. https://doi.org/10.1175/
An evaluation of dynamical downscaling of Central Plains sum- jcli-d-16-0067.1
mer precipitation using a WRF-based regional climate model at Zemp DC, Schleussner CF, Barbosa HMJ, Rammig A (2017) Defor-
a convection-permitting 4 km resolution. J Geophys Res Atmos estation effects on Amazon forest resilience. Geophys Res Lett
121:13801–13825. https://doi.org/10.1002/2016jd024796 44:6182–6190. https://doi.org/10.1002/2017gl072955
Syktus J, Deo RC, McAlpine CA, McGowan HA, Phinn S (2007) Zhang H, Li Y, Gao X (2009) Potential impacts of land-use on climate
Impact of land cover change on climate and El Niño in Australia. variability and extremes. Adv Atmos Sci 26:840–854. https://d oi.
MODSIM 2007: International Congress on Modelling and Simu- org/10.1007/s00376-009-8047-1
lation. The University of Canterbury, Christchurch, New Zealand
Takahashi A, Kumagai TO, Kanamori H, Fujinami H, Hiyama T, Hara Publisher's Note Springer Nature remains neutral with regard to
M (2017) Impact of tropical deforestation and forest degradation jurisdictional claims in published maps and institutional affiliations.
on precipitation over borneo island. J Hydrometeorol 18:2907–
2922. https://doi.org/10.1175/jhm-d-17-0008.1
Teklay A, Dile YT, Asfaw DH, Bayabil HK, Sisay K (2019) Impacts of
land surface model and land use data on WRF model simulations
13

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
[email protected]

Manciu Et Al., 2022

Uploaded by

Copyright:

Available Formats

Manciu Et Al., 2022

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Manciu Et Al., 2022

Uploaded by

Copyright:

Available Formats

Climate Dynamics

Impacts of land cover changes and global warming on climate

Received: 17 January 2022 / Accepted: 15 October 2022

1 Introduction climate by reducing carbon stocks and releasing carbon into

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Table 1 Method for calculating Name of effect Abbreviation Calculation performed

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

y = 0.45 + 0.00017*height y = 1.28 + 0.00017 *height

0 .6 y = 0.56 + 0.00012*height 0 y = 1.15 + 0.00032*height

1000 2000 3000

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Fig. 6 Same as Fig. 4, but for changes in daily precipitation (in mm)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Table 2 Comparison of Simulation ↓/variable → ΔT2 ΔSW ↑ ΔLW ↑ ΔL Δ(H + G)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

You might also like

Table 1 Method for calculating Name of effect Abbreviation Calculation performed

y = 0.45 + 0.00017height y = 1.28 + 0.00017 height

0 .6 y = 0.56 + 0.00012height 0 y = 1.15 + 0.00032height

Fig. 6 Same as Fig. 4, but for changes in daily precipitation (in mm)

Table 2 Comparison of Simulation ↓/variable → ΔT2 ΔSW ↑ ΔLW ↑ ΔL Δ(H + G)