Bolaños

Journal of Hydrology xxx (xxxx) xxx
Contents lists available at ScienceDirect
Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol
Research papers
GRACE reveals depletion of water storage in northwestern South America

between ENSO extremes
Silvana Bolaños a, *, Juan F. Salazar a, Teresita Betancur a, Micha Werner b
a
Grupo de Ingeniería y Gestión Ambiental – GIGA, Escuela Ambiental, Facultad de Ingeniería, Universidad de Antioquia, Medellín, Colombia
b
Water Science and Engineering Department, IHE Delft Institute for Water Education, Delft, the Netherlands
A R T I C L E I N F O A B S T R A C T
This manuscript was handled by Emmanouil Dynamics of terrestrial water storage are determinant for many natural and social phenomena, with implications
Anagnostou, Editor-in-Chief for water security and environmental sustainability. Here we use 2002–2017 data from the Gravity Recovery and
Climate Experiment (GRACE) to study these dynamics in the Magdalena-Cauca river basin in northwestern South
Keywords: America. Through comparison with water balance-based estimates we assess the performance of multiple GRACE
Terrestrial water storage
products in representing water storage dynamics in the basin, identifying the Mascon product from the Jet
Groundwater depletion
Propulsion Laboratory as the best suited for further analysis. We then investigate the existence of long term
ENSO
Tropical Andes trends and show that terrestrial water storage in general and groundwater storage in particular have been
GRACE gradually depleting in the basin since around the end of 2010. GRACE data reveal that this trend is not uniform
across the basin but exhibits a clear-cut pattern in which the water depletion rate is more pronounced in the
lower parts of the basin than it is in the upper basin. We explore the mechanisms behind the identified temporal
trends and spatial patterns and show that water storage depletion largely coincides with a period between the La
Niña and El Niño extreme phases of ENSO. Likewise, the pronounced contrast between depletion rates in the
lower and higher parts of the basin largely coincides with marked biophysical differences between these regions,
including the presence of major wetland systems in the lowlands, and the highlands of the Andean mountains.
1. Introduction which, for a river basin, determines that temporal variations in terres
trial water storage (dS/dt) depend on the differences between input
Many natural and social phenomena depend on the dynamics of fluxes of precipitation (P) and output fluxes of evapotranspiration (E)
terrestrial water storage, which plays a major role in the Earth’s climate and river discharge (Q). Here S includes all components of water storage
system through its effect on the global water, energy, and biogeo at the surface (e.g. water bodies, soil moisture, and snow) and beneath it
chemical cycles (Famiglietti, 2004; Huang et al., 2013; Long et al., 2017; (groundwater).
Yang et al., 2017; Salazar et al., 2018). The dynamics of terrestrial water Estimation of dS/dt for any given region is challenging, especially
storage are, however, sensitive to global change and understanding how when hydrological data is scarce or unavailable (e.g., Voss et al., 2013;
and why these dynamics are changing in different regions of the world is Ouma et al., 2015; Hassan and Jin, 2016). Monitoring wells provide
critical to water resources management in river basins and the assessing precise but generally scattered (in space and time) data of groundwater
of present and future water security (Voss et al., 2013; Long et al., 2014); level that can be used to estimate variations in groundwater storage (e.
and is recognized as a fundamental problem of the hydrological sciences g., Nanteza et al., 2016; Huang et al., 2013). Observations of soil
in the Panta Rhei context (Montanari et al., 2013; Rodriguez et al., moisture, snow, and surface water provide additional means to directly
2018). estimate different components of total water storage. The water balance
A first order explanation of these dynamics is given by the water equation (Eq. (1)) allows for the estimation of dS/dt indirectly through
balance equation, estimates of fluxes P, E, and Q. Though this is not trivial for large river
dS basins, it can be accomplished through combining multiple observa
= P − E − Q, (1) tional techniques (e.g. gauging networks and remote sensing) and
dt
* Corresponding author.
E-mail address: [email protected] (S. Bolaños).
https://doi.org/10.1016/j.jhydrol.2020.125687
Received 30 March 2020; Received in revised form 18 October 2020; Accepted 19 October 2020
Available online 26 October 2020
0022-1694/© 2020 Elsevier B.V. All rights reserved.
Please cite this article as: Silvana Bolaños, Journal of Hydrology, https://doi.org/10.1016/j.jhydrol.2020.125687
S. Bolaños et al. Journal of Hydrology xxx (xxxx) xxx
modelling. The launch of the Gravity Recovery and Climate Experiment Magdalena-Cauca basin (MC basin henceforth) in northwestern South
(GRACE) twin satellite mission in 2002 has provided an unprecedented America. We develop five main activities (Supplementary Fig. S1). After
way of directly estimating dS/dt through high-precision measurements describing our data and methods (Section 2), we first assess the per
of the Earth’s gravity field. The fundamental idea behind these GRACE- formance of multiple available GRACE products (more details in Section
based estimates is that variations in the Earth’s gravity field are related 2) in representing water storage dynamics as compared to independent
with the dynamics of the total terrestrial water storage (TWS) defined as water balance estimates based on terrestrial observations (Section 3.1).
the sum of snow, surface water, soil moisture, and groundwater (Tapley Second, we use the best performing GRACE product to investigate the
et al., 2004; Wahr et al., 2004). existence of long term trends in the MC basin as a whole, as well as in its
Since the launch of the mission, the use of GRACE data to investigate major sub-basins (Section 3.2). Third, we use the Global Land Data
variations in TWS has attracted much attention (Wada et al., 2010; Assimilation System (GLDAS) to separate surface water (SWS) and
Famiglietti, 2014). It has helped identify different regions of the world in groundwater (GWS) contributions to trends in TWS. Fourth, we use
which terrestrial water, mainly groundwater, is being depleted, GRACE data to study the spatial variability of temporal trends in order to
including northwest India (Tiwari et al., 2009; Rodell et al., 2009), the explore the existence of physically meaningful spatial patterns. This
Central Valley in California (Famiglietti et al., 2011) and Texas (Long mapping highlights an advantage of using GRACE data, as it would not
et al., 2013), Brazil (Gonçalves et al., 2020), China (Shen et al., 2015, be possible with water balance estimates that depend on spatially non-
2014, 2013, 2019), the Middle East (Voss et al., 2013), and East Africa continuous and relatively scarce observations of water balance fluxes.
(Nanteza et al., 2016; Swenson and Wahr, 2009). Such depletion in Finally, we explore the underlying mechanisms that explain the identi
water storage poses a threat to the water security and environmental fied temporal trends and spatial patterns, hypothesising that these are
sustainability of those regions. commensurate to ENSO-related variability and pronounced biophysical
Several studies have demonstrated that water storage variations can differences in the basin (Section 4). We present our main conclusions in
be inferred from GRACE with enough resolution and accuracy to benefit Section 5.
water security assessments and water management (Landerer and
Swenson, 2012; Jiang et al., 2014; Long et al., 2015; Wang and Li, 2016). 2. Data and methods
However, there is incomplete evidence of the ability of GRACE data to
reproduce the dynamics of terrestrial water storage in all continental 2.1. The Magdalena-Cauca basin
regions of the world, thus calling for further evaluation and case-specific
studies (e.g., Long et al., 2016; Seyoum and Milewski, 2016; Abiy and The region of interest in this study is the MC basin, located in
Melesse, 2017). Moreover, variations in processing of GRACE data have northwestern South America and draining an area of 276, 000 km2
led to multiple GRACE products (e.g. JPL-Mascons and CSR-Mascons, (Fig. 1). The Magdalena river originates from its headwaters in the
Save et al., 2016; Watkins et al., 2015; Wiese et al., 2016 more details Andes at an elevation of approximately 3, 700 m above sea level, and
in Section 2) the performance of which is not equivalent in every region runs for 1, 612 km before flowing into the Western Caribbean in the
(e.g., Scanlon et al., 2016; Long et al., 2017; Shamsudduha et al., 2017). Atlantic Ocean (López et al., 2018; Restrepo and Kjerfve, 2000). The
The performance of GRACE data can be assessed through comparison of Cauca River is the main tributary of the Magdalena River, flowing along
TWS anomalies from GRACE and variations in surface water, soil the western part of the basin, and joining the Magdalena River in a
moisture, snow and groundwater storage (estimated from groundwater inner-delta wetland area called La Mojana, which is a part of the
level data, if available) (Scanlon et al., 2012; Seyoum and Milewski, Mompós Depression. The average precipitation regime of the basin is
2016). Alternatively GRACE-based TWS anomalies can be compared characterized by a markedly bi-modal cycle (Fig. 2) linked to the lat
with estimates of dS/dt through Eq. (1) using data of the fluxes data P, E itudinal migration of the Inter Tropical Convergence Zone (ITCZ) (Urrea
and Q. For instance, water balance estimates have been used to show et al., 2019), though the bi-modal cycle is more pronounced in the
that GRACE produces a realistic representation of water storage dy northern part of the basin. Seasonal variations in terrestrial water stor
namics in California’s Sacramento and San Joaquin River basins in the age also exhibit a bi-modal cycle.
U.S. (Famiglietti et al., 2011) and in East Africa (Nanteza et al., 2016). Focus on this basin is motivated by its paramount importance to
Other applications of GRACE data include estimating the water water security, and related energy and food security, in Colombia. This
balance of large river basins (Jiang et al., 2014; Seo and Lee, 2016); importance is highlighted by the fact that the MC basin is the largest
evapotranspiration (Billah et al., 2015; Lv et al., 2017; Ramillien et al., basin entirely contained within Colombia, covering around 25% of the
2006; Tang and Zhang, 2011); water budget closure (Lv et al., 2017; country’s continental area; contains most of the reservoirs of the na
Sheffield et al., 2009; Wang et al., 2014); basin storage-river flow re tional hydropower system, as well as multiple wetlands, paramos
lationships (Papa et al., 2015; Reager et al., 2014; Sproles et al., 2015); (montane wetland ecosystems unique to the northern Andes), and small
groundwater variations (Chen et al., 2016, 2011, 2017, 2009, 2013, watershed systems, which are critical to water supply across the country
2016); flood forecasting (Chen et al., 2010; Reager et al., 2014; Tang (Angarita et al., 2018; IDEAM and CORMAGDALENA, 2001). The Na
damrongsub et al., 2016), as well as vegetation greenness and drought tional Water Study (ENA after its Spanish acronym, versions 2014 and
characterization (Chen et al., 2009; Long et al., 2014; Tang et al., 2014), 2018) identifies 34 aquifer systems in the MC basin (Supplementary
glacier, ice cap changes and snow water equivalent (Chen et al., 2007; Fig. S2), some of which have been extensively exploited and are critical
Frappart et al., 2011; Li et al., 2019). In most of these, GRACE data are to the agricultural sector of the country (http://www.andi.com.co/Up
combined with observations of e.g. precipitation and either hydrological loads/ENA_2018-comprimido.pdf). These aquifers will likely play a
or land surface models. Through the combination improved insight is crucial role in safeguarding future water security and environmental
provided into the water balance and fluxes in basins studied, but less so sustainability (IDEAM, 2015; IDEAM, 2019). Further, the MC basin is
into the performance of GRACE data in representing the dynamics of located in a region that is a world biodiversity hotspot (Rangel-Ch, 2015;
TWS. Independent observation of TWS dynamics at the basin level and Myers et al., 2000), and has been identified as highly sensitive to ENSO-
how these are related to climatic phenomena such as ENSO is of induced climate variability (Hoyos et al., 2013; Bedoya-Soto et al.,
importance to water resources management, particularly in large and 2019). The region is regarded to be highly vulnerable to climate change
sparsely monitored basins and considering the projected intensification (Pabón Caicedo, 2012; Magrin et al., 2014).
of the ENSO phenomenon (Cai et al., 2015).
In this paper, we use GRACE data derived products to study the 2.2. GRACE and water balance components data
spatial and temporal dynamics of TWS at the basin scale, and their po
tential linkages to climate change and variability in the tropical We use 183 months from April 2002 to June 2017 of GRACE-derived
2
Fig. 1. The Magdalena-Cauca (MC) river basin in northwestern South America. Topography (shading) and gauging stations (points) used in this study.
monthly anomalies of TWS from three different products. The first standard processing steps are used (Landerer and Swenson, 2012; Ouma
product is constructed from post-processed GRACE RL05 level-3 land et al., 2015). This signal loss refers to a smoothing and signal attenuation
(L3-land) data, gridded at 1◦ (̃110 km near the Equator) and based on that result from the filtering process that is used to reduce GRACE data
the RL05 Spherical Harmonics (SH) basis function. SH Datasets are ob noise. GRACE data are noisy with noise increasing with increasing SH
tained from three sources; the Center for Space Research at University of degree (Landerer and Swenson, 2012; Scanlon et al., 2016). Correcting
Texas (CSR, Austin); the Jet Propulsion Laboratory (JPL), and Geo the loss of signal is important because it reduces the ability of GRACE
forschungs Zentrum Potsdam (GFZ) processing centers (Landerer and data to reproduce components of the observed water balance, especially
Swenson, 2012; Swenson and Wahr, 2006). The ensemble mean of these in small basins (Scanlon et al., 2016).
three post-processed GRACE datasets is used to balance the bias of any In the original GRACE data, TWS anomalies are relative to the
single post-processing technique (e.g., Sakumura et al., 2014; Voss et al., 2004–2009 time-mean baseline. For consistency, all other anomalies
2013). The second and third products are Mascons producs from JPL and used in this study were calculated with respect to the same baseline.
CSR. We denote these as JPL-Mascons and CSR-Mascons respectively. Missing data (e.g. due to battery management) were directly remedied
These are solutions gridded at 0.5◦ (̃55 km) that use regional mass by linear interpolation (e.g., Ouma et al., 2015; Xiao et al., 2015; Liesch
concentration functions (i.e. Mascons) to parameterize the gravity field. and Ohmer, 2016; Shamsudduha et al., 2012). There were 20 out of 183
Mascon products have become operational within the last years (Save missing values corresponding to the following months (Mmm-YY): Jun-
et al., 2016; Watkins et al., 2015; Wiese et al., 2016). All five solutions of 02, Jul-02, Jun-03, Jan-11, Jun-11, May-12, Oct-12, Mar-13, Aug-13,
GRACE land used in this study (CSR, JPL, GFZ, JPL mascon and CSR Sep-13, Feb-14, Jul-14, Dec-14, Jun-15, Oct-15, Nov-15, Apr-16, Sep-
mascon) are available through the GRACE Tellus page http://grace.jpl. 16, Oct-16, and Feb-17. Variations in water storage are always expressed
nasa.gov supported by the NASA Measures Program. as an equivalent water thickness (e.g. cm of water).
In order to restore the signal lost through processing, the scaling The water balance components in the MC basin are estimated using a
factors provided by each processing center to scale GRACE data and combination of data from the national network of gauging stations and
3
Fig. 2. Average 2002–2015 annual cycle of precipitation (P), discharge (Q), and variations in terrestrial water storage (dS/dt) in the MC basin as a whole, as well as
in its southern (upstream, south of 7.5 latitude) and northern (downstream, north of 7.5 latitude) parts. Precipitation and discharge data are from gauges shown in
Fig. 1. Storage estimates are based on GRACE data as clarified below (Eq. (2)).
global re-analysis and satellite datasets. In all cases, these data corre GLEAM v3B is largely driven by satellite data. Taking advantage of the
spond to the period 2002–2015, with the original values converted into multiple available data sources, an ensemble of estimates of the varia
anomalies relative to the common 2004–2009 baseline. Data from tion in terrestrial water storage through the water balance is produced
gauging stations of the national network were provided by the National (Eq. (1)). Table 1 summarizes the GRACE and water balance components
Institute of Hydrology, Meteorology and Environmental Studies of data used in this study.
Colombia (IDEAM for its Spanish acronym; available in http://www.
ideam.gov.co/), and include time series of precipitation from 1084 2.3. Comparing GRACE and water balance derived estimates of TWS
gauges, evaporation from 199 gauges, and river flow from 5 gauges variation
(Fig. 1). Precipitation data were also obtained from four products: the
TRMM Multi-satellite Precipitation Analysis—TMPA, gauge-adjusted Comparison between GRACE-derived and water balance-based esti
(Huffman et al., 2010)— that uses data from the Tropical Rainfall mates of TWS variation has been widely used as a procedure for
Measuring Mission (Huffman et al., 2007), the Global Precipitation assessing GRACE data (e.g., Becker et al., 2010; Famiglietti et al., 2011;
Climatology Centre Full Data Reanalysis Version 7 (GPCC gauge anal Voss et al., 2013; Nanteza et al., 2016; Scanlon et al., 2018). Using all
ysis) that provides gridded gauge-analysis products derived from quality sources of data for P and E, as well as records of Q near the outlet of the
controlled station data (Schneider et al., 2008), the Climate Research MC basin, we first estimate monthly values of dS/dt from the water
Unit (CRU) time-series version 3.24 of high resolution gridded data of balance equation (Eq. (1)). Then we compare these water balance-based
month-by-month variation in climate—CRU gauge analysis (Harris and estimates of dS/dt with the time derivative of TWS anomalies, expressed
Jones, 2017)—, and the Centre for Medium-Range Weather Forecasts as a forward difference derivative (TWSC) of the form
(ECMWF) ReAnalysis-Interim—ERA-Interim (Dee et al., 2011). For
dS
evapotranspiration, we use data from the MODIS Global Evapotranspi (2)
′ ′
≈ TWSCn = TWSn − TWSn− 1 ,
dt n
ration Project (MOD16) (Running et al., 2017), ERA-Interim reanalysis,
and the Global Land Evaporation Amsterdam Model (GLEAM). GLEAM where n − 1 and n denote consecutive months, and the primes indicate
combines multi-satellite observations to estimate daily actual evapora anomalies. This comparison is used to differentiate between GRACE
tion through a process-based methodology. Both GLEAM versions v3A products, with all further analyses using only the GRACE products found
and v3B (Martens et al., 2017; Miralles et al., 2011) are considered. to perform best.
GLEAM v3A is based on satellite-observed soil moisture, vegetation
optical depth, and snow-water equivalent, air temperature and radiation
from reanalysis, and a multi-source precipitation product; whereas
4
Table 1
(4)
′ ′ ′
Summary of GRACE and water balance components used in this study. GWS = TWS − SM ,
Source Variable Resolution Time period References
where SM can be estimated by means of hydrological or land surface
′
GRACE EWT 1◦ × 1◦ Apr 2002–Jan Swenson and models, such as the Global Land Data Assimilation System (GLDAS)
(CSR, JPL, 2017 Wahr, 2006 (Rodell et al., 2004). GLDAS integrates ground and satellite observation
GFZ) Landerer and
Swenson, 2012
based data sets for parameterizing, forcing and constraining multiple
JPL, CSR EWT 0.5◦ × 0.5◦ Apr 2002–Jun Watkins et al., hydrological or land surface models (LSMs) in order to realistically
mascons 2017 2015 simulate water and energy budgets and fluxes (Syed et al., 2008). GLDAS
Wiese et al., 2016 is a robust simulation system that incorporates satellite and ground-
GLDAS EWT 0.25◦ × Jan 2000–Dec Rodell et al., 2004
based observational data products using advanced land surface
0.25◦ 2017
TMPA gauge- P 0.25◦ × Jan 2000–Dec Huffman et al., modeling and data assimilation techniques with the purpose of gener
adjusted 0.25◦ 2016 2010 ating optimal fields of land surface states for soil moisture, snow, and
GPCC gauge P 0.5◦ × 0.5◦ Jan 2000–Oct Schneider et al., fluxes, at global scales and high spatial resolution (0.25–1◦ ) in near-real
analysis 2016 2008 time (Rodell et al., 2004). GLDAS outputs have been similarly used in
CRU gauge P 0.5◦ × 0.5◦ Jan 2000–Dec Harris and Jones,
analysis 2015 2017
GRACE studies (e.g., Ni et al., 2018; Ospina et al., 2018; Ouma et al.,
ERA Interim P 0.75◦ × Jan 2000–Dec Dee et al., 2011 2015; Yang and Chen, 2015).
E 0.75◦ 2015 GLDAS output from the Noah LSM is used. Noah uses parameteri
MODIS E 0.5◦ × 0.5◦ Jan 2000–Dec Running et al., zations of multiple land surface processes (e.g. runoff, infiltration, and
2014 2017
canopy conductance) to simulate the evolution of the surface energy and
GLEAM v3A E 0.25◦ × Jan 2000–Dec Martens et al.,
0.25◦ 2014 2017 water budgets in response to near-surface atmospheric forcing, as well as
Miralles et al., the dependence of this evolution on surface conditions such as vegeta
2011 tion state and soil texture Ek et al., 2003. This LSM provides the finer
GLEAM v3B E 0.25◦ × Jan 2003–Dec Martens et al., resolution (0.25◦ ) as compared to the other available LSMs, and has
0.25◦ 2015 2017
Miralles et al.,
been successfully used to study the dynamics of water storage in tropical
2011 South America (Eom et al., 2017; Han et al., 2009).
IDEAM P 1084 Jan 2000–Dec http://www.ide
E stations 2015 am.gov.co 2.5. Detection of trends and breakpoints
Q 199 stations
5 stations
To analyse trends, seasonality from GRACE data is removed by
EWT: equivalent water thickness. P: precipitation. E: evapotranspiration. Q: subtracting the monthly mean for each calendar month, and then
river discharge. applying linear regression and the non-parametric Mann–Kendall (MK)
test. The MK test is a standard rank-based procedure which allows
2.4. Isolation of GWS contribution from GRACE-derived TWS variation assessing the significance of trends (Huang et al., 2013). The Segmented
package in R (Muggeo et al., 2008) is used to study whether the water
TWS anomalies derived from GRACE can be understood as the result storage records are better described by one or two trend lines, i.e.
of adding anomalies in groundwater storage (GWS ) and all forms of
′
whether there are breakpoints at which a trend is reduced (e.g. the slope
surface water storage (SWS ) (e.g., Chen et al., 2016; Shamsudduha is reduced without changing its sign) or even reversed (the sign of the
′
et al., 2012; Famiglietti et al., 2011 and others) as expressed by slope changes). This trend analysis is performed for the whole MC basin
and its main tributaries, as well as for each GRACE grid cell in order to
(3)
′ ′ ′
TWS = GWS + SWS . explore the existence of spatial patterns in TWS variability.
Therefore, anomalies of groundwater storage GWS can be estimated
′
as the residual of TWS after subtracting SWS . SWS includes multiple

′ ′ ′ 2.6. Optical and Synthetic Aperture Radar (SAR) images
components, though some may be negligible for a given region. For
instance, the snow water equivalent can be assumed as negligible for the To support our discussion of the spatial variability of trends, varia
MC basin due to its tropical climate regime. Soil moisture (SM) is tions in the extent of wetlands areas in the Mompós Depression are
estimated based on Palsar and Landsat images (e.g., Escobar Martínez,
included in the definition of SWS adopted here, similar to other studies
′
2011; Anaya-Acevedo et al., 2017; Yuan et al., 2015; Palomino-Ángel

(Fatolazadeh et al., 2016; Ouma et al., 2015; Chen et al., 2016), and
et al., 2019; Dabrowska-Zielinska et al., 2014; Arnesen et al., 2013). To
arguably accounts for a considerable portion of SWS variations (Huss
characterize the extent of flooding during the 2010–2011 (La Niña) wet
et al., 2017).
season 43 available SAR images were processed. The L-band HH polar
Another TWS component that may be significant is surface water
ized mode is adopted as it is known to be sensitive to water level change
storage of major water bodies such as reservoirs and lakes (e.g., Becker
beneath the vegetation (Pope et al., 1997; Rosenqvist et al., 2007; Grings
et al., 2010; Swenson and Wahr, 2009; Shamsudduha et al., 2017).
et al., 2009) while maintaining coherence for interferometric processing
Indeed, most of the major reservoirs and some large wetland systems in
(Kim et al., 2009). For the dry season, optical images are found to be
Colombia are located within the MC basin (Angarita et al., 2018). In
better suited to identify wetland extent in tropical environments (Lie
some studies, this component is subtracted from TWS to isolate GWS (e.
senberg et al., 2007; Deus, 2016). 36 Landsat images are used to char
g., Famiglietti et al., 2011; Fatolazadeh et al., 2016; Huang et al., 2016;
acterize wetlands in the Mompós Depression during the El Niño of
Liesch and Ohmer, 2016 and others). However, this requires an esti
2015–2016. Data from PALSAR and Landsat were obtained, respec
mation of water storage dynamics in such water bodies (e.g. using sur
tively, from online portals of the Alaska Satellite Facility (https://vertex.
face water altimetry data from LEGOS, Voss et al., 2013), which is not
daac.asf.alaska.edu/) and the USGS EarthExplorer (https://eart
possible for the MC basin due to lack of data. Further, water storage
hexplorer.usgs.gov/).
dynamics in large water bodies is often tightly coupled to GWS dynamics
through sub-surface flows (Winter, 1999; Ouma et al., 2015). From this
perspective, as with several other studies (e.g., Rodell et al., 2007;
Moiwo et al., 2012; Ouma et al., 2015; Hachborn et al., 2017) GWS is
′
calculated as;
5
3. Results only statistics in which JPL-Mascon is clearly worse than other product
is in the error for the time series which is best for CSR-Mascon (blue vs.
3.1. Comparison between GRACE products and water balance-based yellow circles in Fig. 3c). In summary, JPL-Mascon produces better or
estimates comparable results with respect to other GRACE-based estimates in
almost all cases (five out of six statistics). This indicates that estimates
Fig. 3 shows a comparison between the multiple GRACE-based esti based on JPL-Mascon are closer to water balance-based estimates in the
mates of TWSC and the corresponding estimates of dS/dt based on the MC basin, and hence we choose this product for further analysis.
water balance (Eqs. (1) and (2)). Three different GRACE-based estimates
are included; the mean of the Spherical Harmonics (SH) basis functions 3.2. Trends in water storage
from the CSR, JPL, and GFZ processing centers (GRACE Mean SH); and
the Mascon products from JPL (JPL-Mascon) and CSR (CSR-Mascon). GRACE data (JPL-Mascon henceforth) reveals the existence of two
Estimates derived from the water balance are established with 14 different trends in terrestrial water storage (TWS ) during the study
′
combinations that use all sources of data for P,E, and Q (Table 1). In this period for the whole MC basin (Fig. 4). There is a positive trend (i.e. an
Taylor diagram, filled and unfilled points correspond to observation- increase in water storage, 0.14 ± 0.02 cm/month) before 2010 and a
based estimates obtained respectively with all 14 combinations of data negative trend (i.e. a decrease or depletion of water storage,
sources and gauging stations from IDEAM only. − 0.37 ± 0.04 cm/month) afterwards. This positive (negative) trend is
Agreement between GRACE- and water balance-based estimates equivalent to an increase (decrease) of TWS of 41.04 ± 5.19 km3
varies among the three GRACE products. There are three clear differ (77.68 ± 7.57 km3) during 2002–2010 (2011–2017) in the MC basin. In
ences. First, correlation is worst for the GRACE Mean SH product (red in all cases, reported trends are statistically significant meaning that the
Fig. 3), both for the time series and the annual cycle. The poor corre corresponding p-value is lower than 0.05, and break points are identified
lation is particularly as a result of the lag in the first peak of the bi-modal by using the objective method described in Section 2.
climate, which for the GRACE Mean SH product occurs between The pattern consisting in a positive trend of water storage that is
May–June, while all other estimates indicate the peak in April–May. reversed in recent years is also present in major tributaries of the MC
This lag is evident in the average annual cycle (Fig. 3b) as a result of the basin and for surface and groundwater storage (Fig. 5 shows GWS and
′
pronounced lags that are observed in several years of the time series, e.g.
SM while Supplementary Fig. S3 shows the same map but for TWS ).
′ ′
2004, 2006, 2009, 2012 and 2013 (Fig. 3a).

Here we introduce the major sub-basins of the whole MC basin, and refer
Second, the bi-modality of the annual cycle is less pronounced in
to them as the middle sub-basins: the Upper-Middle Magdalena (UMM)
estimates of CSR-Mascon (yellow in Fig. 3) than it is for all other esti
and Cauca (C), and upper sub-basins: the Upper Cauca (UC) and Upper
mates. Although this is not easy to detect by visual examination of the
Magdalena (UM). Soil moisture (SM ) and groundwater (GWS ) storage
′ ′
time series (Fig. 3a), it is evident in the average annual cycle (Fig. 3b),
contribute differently to trends in total terrestrial water storage (TWS ).
′
and results in the worst representation of temporal variance (standard

deviation) obtained with this product (yellow points in Fig. 3c). At the scale of the whole basin (MC), as well as in its middle sub-basins
Third, better agreement between GRACE and the water balance- (UMM and C), trends in groundwater are more pronounced than for soil
based estimates is found for the JPL-Mascon product (blue in Fig. 3). moisture. During the decreasing trend in 2011–2017, the UMM basin
In particular, JPL-Mascon is clearly best in representing the average lost around 27.64 km3 and 12.73 km3 of groundwater and soil moisture,
annual cycle (Fig. 3b), with the highest correlation, lowest error, and respectively. During the same period, the corresponding losses of
best standard deviation (blue triangle in Fig. 3c). As for the time series of groundwater storage and soil moisture in the C basin were close to 6.30
monthly values (Fig. 3a), JPL-Mascon is comparable with GRACE mean km3 and 4.78 km3, respectively.
SH in error and standard deviation but better in correlation (blue vs. red The upper sub-basins (Fig. 5e,d) exhibit contrasting patterns when
circles in Fig. 3c), and comparable with CSR-Mascon in correlation but compared to the middle sub-basins (Fig. 5c,b) as well as the whole MC
much better in standard deviation (blue vs. yellow circles in Fig. 3c). The basin (Fig. 5a). Two main differences are that in the upper sub-basins
depletion rates during recent years have been more pronounced for
Fig. 3. Comparison between GRACE-based esti

mates of TWSC and the corresponding estimates of
dS/dt based on the water balance. (a) Time series of
monthly values during 2002–2015; (b) 14-years
(2002–2015) average annual cycle; and (c) Taylor
diagrams relating observation-based estimates
(ensemble: filled points, gauging stations from
IDEAM: unfilled points) and TWSC (GRACE) for
both the time series in (a) and the annual cycle in
(b). Grey shading in panels (a) and (b) shows the
envelope of dS/dt estimates (Eq. (1)) using multiple
data sources for the water balance components. Blue
bars show the network-averaged annual cycle of
precipitation for the same period. Green line shows
the 14-years (2002–2015) average annual cycle of
surface water storage change (SWS’ in Eq. (3)). (For
interpretation of the references to colour in this
figure legend, the reader is referred to the web
version of this article.)
6
Fig. 4. Monthly anomalies of TWS from JPL-Mascon for the whole MC basin. Solid lines show statistically significant trends (p < 0.05) which are positive before
December 2010 and negative afterwards. Seasonality is removed from data.
Fig. 5. Monthly anomalies of SM (blue) and GWS (red) and their corresponding trends (solid lines) at the MC basin as a whole and its major sub-basins. Seasonality is
removed from data. Only statistically significant trends (p < 0.05) are plotted. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
soil moisture than for groundwater, and break-point may be located in MC basin around ̃7.5 latitude (this is not an arbitrary selection but a
2008, i.e. earlier than the break-point identified in 2010 for the other result based on Fig. 6) implies that the upper sub-basins are in the upper
basins. Despite these differences, two findings are still common to the part of the basin, while the middle sub-basins include areas in both the
whole MC basin and all of its major sub-basins: (i) there has been a water higher and lower parts of the basin. Average trends have the same sign in
storage depletion trend in recent years, which in most cases (only ex both parts of the basin (positive during 2002–2010 and negative for
ceptions are for soil moisture in the upper sub-basins, Fig. 5e,d) was 2011–2017) but are much more pronounced in the lower part. During
preceded by a positive trend; and (ii) the depleting trend coincides with the positive trend period, water storage increased about 26.14 km3
a period between ENSO extremes. Common patterns and differences (trend is 0.23 ± 0.03 cm/month) in the lower part of the basin, whereas
between sub-basins are further discussed in Section 4. it increased around 13.99 km3 (trend is 0.08 ± 0.02 cm/month) in the
Temporal trends in water storage are not uniformly distributed higher part. Likewise, the decreasing trend in the lower part
across space in the MC basin. There is a marked difference between the ( − 0.60 ± 0.06 cm/month) is equivalent to a loss of ̃49.96 km3 during
higher (upstream, south of ̃7.5 latitude) and lower (downstream, north the period, while in the higher part ( − 0.24 ± 0.03 cm/month) the
of ̃7.5 latitude) parts of the basin (Fig. 6). Note that this partition of the corresponding loss is about 31.80 km3.
7
Fig. 6. Maps of statistically significant (p < 0.05) trends in TWS anomalies for time periods 2002–2010 (a) and 2011–2017 (b). Panels c–f show the temporal
evolution of TWS anomalies averaged over the northern (c, e: downstream, north of 7.5 latitude) and southern (d, f: upstream, south of 7.5 latitude) parts of the basin
during 2002–2010 (c, d) and 2011–2017 (e, f). Solid lines represent statistically significant (p < 0.05) trends.
From Fig. 6, it is evident that the monthly variability (variance) of difference between SH and Mascons is that SH solutions are global
TWS is higher for the northern time series than it is for the southern time whereas Mascons can be applied at regional to global scales (Scanlon
series, even if the trends are removed. This implies that the trends have et al., 2016). This difference is important because in the global SH so
different levels of significance. Likewise, the variability of GWS is higher lutions one cannot distinguish between land and ocean areas; hence, the
than that of SM (Fig. 5), with consequent differences in the significance generally higher land signals leak into the lower ocean signals, thereby
of the trends. Despite these differences, all trends are found to be sta reducing signal amplitudes. In contrast, land and ocean areas can be
tistically significant. explicitly defined during Mascon processing, reducing leakage errors
Differences in trends between the northern and southern parts of the relative to the SH solutions (Scanlon et al., 2016).
MC basin are not attributable to differences in the size of these areas. Also, gridded Mascon fields are provided at the improved spatial
Likewise, differences between trends shown in Fig. 5 are not attributable resolution of 0.5◦ × 0.5◦ , while SH solutions are available at a resolu
to differences in the size of sub-basins. This is because time series data in tion of 1◦ × 1◦ . The greater resolution of Mascons solutions for smaller
Figs. 5 and 6 are normalized by the size of the corresponding area. This spatial regions (Watkins et al., 2015) may also provide an explanation
area is directly related to but does not fully describe the storage capacity, for the lag between SH and Mascons solutions, particularly for the first
which depends also on the properties of soil columns and related land peak (April-May) of the largely bi-modal climate. While the upper basin
–atmosphere dynamics. Some important factors affecting these proper is clearly bi-modal (Fig. 3), the lower basin tends to a more uni-modal
ties and dynamics will be discussed in the next section. climate with an extended wet season. Whether the lag is due to the
poorer resolution of SH solutions resulting in a mixed representation of
4. Discussion the heterogeneous climate, or due to the SH solution better capturing the
dynamic of the lower basin than the upper basin is open to additional
In this section we explore the reasons and/or mechanisms behind our research.
three main findings, as well as their implications for water security and As compared to the other Mascon-type product, i.e. CSR-Mascon,
environmental sustainability. The first finding is that among different JPL-Mascon has two differences that we consider consistent with the
GRACE-based products the JPL-Mascon product exhibits a closer better performance found. First, these Mascon products differ in their
agreement with water balance-based estimates of water storage dy spatial resolution. JPL-Mascon samples the gravity field at approxi
namics in the MC basin (Section 3.1). This finding generally agrees with mately the native resolution of the GRACE mission at the Equator,
Scanlon et al., 2016 who describe several advantages of Mascon solu whereas CSR-Mascon deliberately oversamples the gravity field at the
tions relative to SH such as the increase of signal amplitude due to the Equator to increase the solution resolution at higher latitudes (Watkins
reduction of leakage from land to ocean, and the application of et al., 2015; Save et al., 2016; Scanlon et al., 2016). This may be why
geophysical data constraints during processing with little empirical post CSR maps present a spatial distribution smoother than JPL maps, so that
processing required. In contrast to the SH approach in which noise it is difficult to identify areas with special interest (e.g. Supplementary
reduction and signal restoration are applied as post processing steps, Fig. S3). The second difference between JPL-Mascon and CSR-Mascon
noise is reduced during Mascon processing by applying constraints solutions is that they represent two fundamentally different ap
(Scanlon et al., 2016). The use of mass concentrations leads to better proaches to applying constraints. Among these differences JPL-Mascon
signal-to-noise ratios of the Mascon fields as compared to the SH solu constraints are based on both GRACE data and geophysical models,
tions (Watkins et al., 2015; Shamsudduha et al., 2017). In contrast to SH whereas CSR-Mascon constraints are based on GRACE data only (Wat
products used to produce our GRACE ensemble, JPL-Mascon does not kins et al., 2015; Save et al., 2016). Hence, JPL-Mascon uses a combi
require the use of scaling factors in post-processing, which are needed in nation of near-global geophysical models along with altimetry
SH products to restore the signal lost during the processing of the observations to improve mass flux estimates globally, which may be
truncation of the gravity coefficients and filtering (Landerer and particularly important in the complex terrain of the Andes (Watkins
Swenson, 2012). JPL-Mascon (as well as CSR-Mascon) estimates et al., 2015). Scanlon et al. (2016) also show that CSR-Mascon has
terrestrial mass changes directly from inter-satellite acceleration mea slightly higher errors than all the other solutions for large basins. For
surements and can be used without further post-processing (Rowlands further comparison between SH and Mascon GRACE products we refer
et al., 2010; Watkins et al., 2015; Shamsudduha et al., 2017). Another the reader to Scanlon et al. (2016) and Shamsudduha et al. (2017).
8
Overall, our results suggest that JPL-Mascon is preferrable than other not identify the break-point in the trend in 2010 but rather in 2008
products for the assessments of water storage dynamics in the MC basin. (Fig. 5d,e). The depleting trend in these sub-basins is significant if
The second finding is the existence of significant temporal trends in starting at 2010 as well (results not shown). That the objective method
water storage that shift from positive (increase in water storage) to identifies this break-point in 2008 may, however, be related to physi
negative (water storage depletion) during the study period (Section 3.2). cally meaningful differences between the upper and middle sub-basins.
In the MC basin as a whole, as well as in its middle sub-basins (C and There were two La Niña years around 2008 (Fig. 7), and the effect of La
UMM), this shift occurs during the 2010–2011 La Niña (Figs. 4 and Niña 2010–2011 on TWS was less pronounced in the upper sub-basins
5a–c). The resulting negative trend continues until the 2015–2016 El (Fig. 5d,e) than in the middle sub-basins (Fig. 5b,c) or the whole basin
NiÑo. This is further clarified by Fig. 7 which shows the evolution of (Fig. 5a). This is evident from comparison between peak values of TWS
TWS anomalies along with precipitation anomalies and the Multivariate anomalies which occur in 2008 for the upper sub-basins. Despite this
ENSO Index (MEI; Wolter and Timlin, 1993). difference in the start of the depleting trend, there is a clear pattern
Several other similar concepts are highlighted in Fig. 7. The common to all sub-basins and the MC basin as a whole: a depleting trend
depleting trend occurs between two strong ENSO extremes that had which starts during La Niña (around either 2008 or 2010) and continues
extensive impacts in the MC basin: The 2010–2011 La Niña was related until El Niño 2015–2016, i.e. that occurs between ENSO extremes.
with severe flooding (Hoyos et al., 2013), while the 2015–2016 El Niño Although our results indicate that climate variability plays a major
was linked with the most severe drought in recent years (Hoyos et al., role in water storage dynamics, they do not allow to relate the water
2017; Martinez et al., 2017). Indeed, the maximum (TWS = 22.60 cm in storage depletion trends found with climate change. Near the end of our
′
December, 2010) and minimum (TWS = − 21.48 cm in March, 2016)

′
study period there is an evident positive trend in TWS after around 2016.
values of TWS anomalies found during the whole study period occurs This may be interpreted as recovery of the TWS after the strong El Niño
during these ENSO extremes. of 2015–2016. We do not isolate this trend in our analysis because it
Fig. 7 also shows that weaker La Niña (e.g. 2007–2008) and El Niño occurs during a very short period of time given the available period of
(e.g. 2009–2010) events can be related with positive and negative record. The period of record available also means that we cannot be
anomalies of TWS, respectively. This is consistent with La Niña and El conclusive on the existence of a climate change signal in water storage
Niño being typically associated with positive and negative anomalies of dynamics in the MC basin (e.g. long-term persistent water storage
precipitation in this region, witnessed by the positive/negative anoma depletion).
lies of precipitation mirroring negative/positive values of MEI. The This highlights the importance of long term monitoring of water
correlation between deseasonalized TWS anomalies and MEI is also storage dynamics, which in the case of GRACE and the MC basin de
negative (-0.71), indicating that negative (positive) values of MEI during pends on international efforts beyond the reach of Colombia. Such
La Niña (El Niño) are related with positive (negative) anomalies of TWS. continued monitoring is critical for distinguishing between climate
This correlation between MEI and TWS anomalies exhibits the same variability and the potential effects of climate change. The implications
pattern as the correlation between MEI and precipitation found for for water security and environmental sustainability and water resources
northwestern South America (Waylen and Poveda, 2002; Poveda, 2004; management planning are quite different for positive/negative trends in
Guarin and Poveda, 2013), as well as in other tropical regions in which TWS modulated by ENSO, than for a persistent depletion rate as a result
La Niña and El Niño are related with positive/negative precipitation of climate change (e.g., Buytaert and De Bièvre, 2012; IDEAM et al.,
anomalies (Phillips et al., 2012; Awange et al., 2014; Zhang et al., 2015). 2015). More frequent and/or intense El Niño events as a result of climate
The correlation between TWS and MEI improves slightly (up to − 0.73) if change (Duque-Villegas et al., 2019; Cai et al., 2015; Cai et al., 2014)
the time series are lagged by two months, consistent with the notion that could, however, exacerbate longer term depletion rates in the region.
the effects of ENSO on the hydrology of the region are lagged (Guarin Other signals of climate change have been identified in the region,
and Poveda, 2013; Ni et al., 2018). These results concur with the including the shrinkage of tropical glaciers (Poveda and Pineda, 2009)
conclusion that the MC basin is in a region highly sensitive to ENSO and long-term trends in monthly hydro-climatic series including
extremes (Poveda, 2004), and are consistent with the premise that the streamflow, precipitation, and temperature (Carmona and Poveda,
observed trends in water storage (Figs. 4 and 5) are strongly modulated 2014).
by ENSO extremes. The third finding is that the magnitude (not the sign) of temporal
It is notable that in the upper sub-basins the objective method does trends exhibit a pronounced difference between the lower and higher
Fig. 7. Monthly values of TWS anomalies (blue line, left axis), precipitation anomalies (gray bars, left axis), and Multivariate ENSO Index (right axis). Shading
indicates whether the ENSO phase is Neutral (gray), El Niño (red), or La Niña (blue). (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
9
parts of the basin (Fig. 6). This spatial heterogeneity largely coincides wetland dynamics and TWS (Xie et al., 2016; O’Loughlin et al., 2018;
with the pronounced differences between these parts of the basin. The Lee et al., 2011). Wetlands are inextricably linked to other key players of
higher part of the basin (to the South) is in the Andes mountains and is the terrestrial water balance such as rivers and aquifers. During wet
characterized by complex terrain with pronounced elevation gradients conditions, extensive wetlands in a flat terrain or depression (e.g. the
(elevation ranges from approximately 41 to 5500 m.a.s.l). There are also Mompós depression) can enhance TWS via retention of water from
pronounced gradients in biophysical properties including biodiversity increased precipitation and river discharges including floods. La Niña,
(Josse et al., 2011), weather and climate regimes (Poveda, 2004; IDEAM and particularly the strong 2010–2011 event, exacerbated the wet
and CORMAGDALENA, 2001), geomorphology and soils (IDEAM and conditions in the lower part of the MC basin. In contrast, during dry
CORMAGDALENA, 2001). This Andean part of the MC basin is also conditions, wetlands can enhance the reduction of TWS via direct
characterized by the presence of dams and reservoirs (Angarita et al., evaporation from extensive areas of free surface water, especially if the
2018), strategic ecosystems including paramos (Ortiz et al., 2005) and availability of energy is relatively high, such as in tropical regions.
tropical forests (Rodriguez and Armenteras, 2005), as well as tropical
glaciers on the highest peaks (Rabatel et al., 2013); all of which can exert 5. Conclusions
a significant influence on the hydrology of the basin at multiple scales.
Indeed, the progressing loss of paramos, tropical forests and glaciers has We assess the performance of selected GRACE products in repre
raised concerns on how these changes will affect future water security in senting the dynamics of water storage in the northwestern South
the tropical Andes (Bradley et al., 2006; Buytaert et al., 2011; Vuille, America, more specifically in the Magdalena-Cauca (MC) basin in
2013). Colombia. The potential of GRACE derived products in understanding of
Compared to the higher part of the MC basin, the lower part con these dynamics, and their relation to climate variability, is significant to
stitutes much flatter terrain characterized by the presence of extensive water resources planning in large, sparsely monitored river basins. Our
wetland systems, mainly in the Mompós depression (Angarita et al., results show that GRACE-based estimates of the water storage are
2018), and a drier climate with some areas experiencing increasing comparable to estimates obtained through the water balance equation
aridity and desertification processes (Jaramillo-Mejia and Chernichov using independent sources of the constituent components of the water
sky, 2019; Etter et al., 2008). Fig. 8 illustrates how the spatial extent of balance; precipitation, evaporation, and river discharge. The perfor
these wetlands was largely reduced between 2010 and 2016, consistent mance of the different available GRACE products is, however, not uni
with water depletion during the same period. The Mompós wetlands form. The Mascon product developed by the Jet Propulsion Laboratory
system are severely impacted by droughts and floods related to ENSO (JPL-Mascon) is found to be best suited for describing the spatial and
extremes. Indeed, while the 2010–2011 La Niña caused severe flooding temporal variability of Terrestrial Water Storage (TWS) in the tropical
and concomitant damage due to e.g. the breach of a large artificial levee MC basin.
in the Cauca River (Nardini and Idarraga, 2016), the 2015–2016 El Niño The JPL-Mascon data reveals the existence of trends in terrestrial
caused a severe drought with significant impacts on local water security water storage in the MC basin that have shifted from positive (i.e. in
and related economic impacts (Martinez et al., 2017). crease of water storage) to negative (i.e. water storage depletion) during
We interpret the result of the more pronounced positive (before the study period. These trends and shifts are evident not only in the total
2010) and negative (after 2010) trends in the lower part of the basin as a terrestrial water storage but also in soil moisture and groundwater, and
higher sensitivity of water storage in this region to external forcing due have not only occurred in the MC basin as a whole but are also evident in
to ENSO variability. This higher sensitivity can be related to the high its major sub-basins. Compared to soil moisture, the contribution of
variability of water storage in the extensive wetlands of the region groundwater to these trends in total water storage is generally larger.
(Fig. 8). Indeed, previous studies have found a close relation between We find that identified shifts in trends are clearly related to ENSO, with
Fig. 8. Wetlands area superimposed over anomalies of TWS around the beginning (2010) and end (2016) of the water depletion trend in the MC basin. Wetlands
areas are based on Palsar and Landsat images (see Section 2.6 for details).
10
the most pronounced trend in the period of analysis occurring between floodplain using ALOS/PALSAR ScanSAR images. Remote Sens. Environ. 130,
51–61.
the strong 2010–2011 La Niña and 2015–2016 El Niño events. The
Awange, J., Forootan, E., Kuhn, M., Kusche, J., Heck, B., 2014. Water storage changes
trends in water storage are, however, not uniform across the MC basin, and climate variability within the Nile Basin between 2002 and 2011. Adv. Water
exhibiting a pronounced contrast between the higher and the lower Resour. 73, 1–15.
parts of the basin, i.e. between the mountainous (Andean) and flatter Becker, M., LLovel, W., Cazenave, A., Güntner, A., Crétaux, J.-F., 2010. Recent
hydrological behavior of the East African great lakes region inferred from GRACE,
(near the Caribbean sea) parts of the basin. We interpret the more pro satellite altimetry and rainfall observations. C.R. Geosci. 342, 223–233.
nounced trends in the lower part of the basin as a higher sensitivity of Bedoya-Soto, J.M., Poveda, G., Trenberth, K.E., Vélez-Upegui, J.J., 2019. Interannual
water storage to ENSO variability, related with the presence of extensive hydroclimatic variability and the 2009–2011 extreme ENSO phases in Colombia:
from Andean glaciers to Caribbean lowlands. Theoret. Appl. Climatol. 135,
wetlands that can exacerbate both water retention during wet conditions 1531–1544.
related to La Niña and water loss during dry conditions related to El Billah, M.M., Goodall, J.L., Narayan, U., Reager, J., Lakshmi, V., Famiglietti, J.S., 2015.
NiÑo. This has implications for future planning and management of A methodology for evaluating evapotranspiration estimates at the watershed-scale
using GRACE. J. Hydrol. 523, 574–586.
water resources, particularly considering projections of increased fre Bradley, R.S., Vuille, M., Diaz, H.F., Vergara, W., 2006. Threats to water supplies in the
quency and severity of ENSO events. tropical Andes. Science 312, 1755–1756.
These results provide a first quantitative assessment of GRACE data Buytaert, W., Cuesta-Camacho, F., Tobón, C., 2011. Potential impacts of climate change
on the environmental services of humid tropical alpine regions. Glob. Ecol. Biogeogr.
for this basin, contributing to the global assessment of GRACE as a 20, 19–33.
reliable source of data for the study of terrestrial water resources, and Buytaert, W., De Bièvre, B., 2012. Water for cities: the impact of climate change and
highlighting the importance of continuing monitoring efforts like demographic growth in the tropical Andes. Water Resour. Res. 48.
Cai, W., Borlace, S., Lengaigne, M., Van Rensch, P., Collins, M., Vecchi, G.,
GRACE.
Timmermann, A., Santoso, A., McPhaden, M.J., Wu, L., et al., 2014. Increasing
frequency of extreme El Niño events due to greenhouse warming. Nat. Clim. Change
CRediT authorship contribution statement 4, 111–116. https://doi.org/10.1038/nclimate2100.
Cai, W., Santoso, A., Wang, G., Yeh, S.-W., An, S.-I., Cobb, K.M., Collins, M.,
Guilyardi, E., Jin, F.-F., Kug, J.-S., et al., 2015. ENSO and greenhouse warming. Nat.
Silvana Bolaños: Conceptualization, Methodology, Formal analysis, Clim. Change 5, 849–859. https://doi.org/10.1038/nclimate2743.
Investigation, Writing - original draft. Juan F. Salazar: Conceptualiza Carmona, A.M., Poveda, G., 2014. Detection of long-term trends in monthly hydro-
climatic series of Colombia through Empirical Mode Decomposition. Clim. Change
tion, Investigation, Writing - review & editing. Teresita Betancur: 123, 301–313.
Conceptualization, Investigation, Writing - review & editing. Micha Chen, J., Wilson, C., Tapley, B., Blankenship, D., Ivins, E., 2007. Patagonia icefield
Werner: Writing - review & editing. melting observed by gravity recovery and climate experiment (GRACE). Geophys.
Res. Lett. 34.
Chen, J., Wilson, C., Tapley, B., Yang, Z., Niu, G., 2009. 2005 drought event in the
Amazon River basin as measured by GRACE and estimated by climate models.
Declaration of Competing Interest J. Geophys. Res.: Solid Earth 114.
Chen, J., Wilson, C.R., Tapley, B.D., Scanlon, B., Güntner, A., 2016. Long-term
The authors declare that they have no known competing financial groundwater storage change in Victoria, Australia from satellite gravity and in situ
observations. Global Planet. Change 139, 56–65.
interests or personal relationships that could have appeared to influence
Chen, J.L., Wilson, C.R., Tapley, B.D., 2010. The 2009 exceptional Amazon flood and
the work reported in this paper. interannual terrestrial water storage change observed by GRACE. Water Resour. Res.
46.
Dabrowska-Zielinska, K., Budzynska, M., Tomaszewska, M., Bartold, M., Gatkowska, M.,
Acknowledgments Malek, I., Turlej, K., Napiorkowska, M., 2014. Monitoring wetlands ecosystems using
ALOS PALSAR (L-Band, HV) supplemented by optical data: a case study of Biebrza
This research was carried out within the financial support from the Wetlands in Northeast Poland. Remote Sens. 6, 1605–1633.
Dee, D.P., Uppala, S., Simmons, A., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U.,
Ministry of Science, Technology, and Innovation of Colombia – MIN Balmaseda, M., Balsamo, G., Bauer, D.P., et al., 2011. The ERA-Interim reanalysis:
CIENCIAS (previously COLCIENCIAS) through the National Doctorates configuration and performance of the data assimilation system. Q. J. R. Meteorol.
Program Nº727/2015. We kindly acknowledge the Universidad de Soc. 137, 553–597.
Deus, D., 2016. Integration of ALOS PALSAR and Landsat data for land cover and forest
Antioquia and the project Evidence4Policy (project reference 106471),
mapping in Northern Tanzania. Land 5, 43.
as part of the IHE Delft Partnership Programme for Water and Devel Duque-Villegas, M., Salazar, J.F., Rendón, A.M., 2019. Tipping the ENSO into a
opment (DUPC2), for supporting this study. Data used in this manuscript permanent El Niño can trigger state transitions in global terrestrial ecosystems. Earth
Syst. Dyn. 10, 631–650.
are freely available in the web pages mentioned in the Data and Methods
Ek, M., Mitchell, K., Lin, Y., Grunmann, P., Rogers, E., Gayno, G., Koren, V., Tarpley, J.,
section and cited in the references. We acknowledge the careful and 2003. Implementation of the upgraded Noah land-surface model in the NCEP
constructive reviews of Editor Jesus Mateo-Lazaro, Clement Guilloteau operational mesoscale Eta model. J. Geophys. Res. 108, 530–543.
and two anonymous reviewers, who have helped improve the Eom, J., Seo, K.-W., Ryu, D., 2017. Estimation of Amazon River discharge based on EOF
analysis of GRACE gravity data. Remote Sens. Environ. 191, 55–66.
manuscript. Escobar Martínez, J.F. Ph.D. thesis.
Etter, A., McAlpine, C., Possingham, H., 2008. Historical patterns and drivers of
landscape change in Colombia since 1500: a regionalized spatial approach. Ann.
Appendix A. Supplementary data
Assoc. Am. Geogr. 98, 2–23.
Famiglietti, J.S., 2004. Remote sensing of terrestrial water storage, soil moisture and
Supplementary data associated with this article can be found, in the surface waters. Washington DC Am. Geophys. Union Geophys. Monogr. Ser. 150,
197–207.
online version, at https://doi.org/10.1016/j.jhydrol.2020.125687.
Famiglietti, J.S., 2014. The global groundwater crisis. Nat. Clim. Change 4, 945.
Famiglietti, J.S., Lo, M., Ho, S.L., Bethune, J., Anderson, K., Syed, T.H., Swenson, S.C., de
References Linage, C.R., Rodell, M., 2011. Satellites measure recent rates of groundwater
depletion in California’s Central Valley. Geophys. Res. Lett. 38.
Fatolazadeh, F., Voosoghi, B., Naeeni, M.R., 2016. Wavelet and Gaussian approaches for
Abiy, A.Z., Melesse, A.M., 2017. Evaluation of watershed scale changes in groundwater
estimation of groundwater variations using GRACE data. Groundwater 54, 74–81.
and soil moisture storage with the application of GRACE satellite imagery data.
Feng, W., Zhong, M., Lemoine, J.-M., Biancale, R., Hsu, H.-T., Xia, J., 2013. Evaluation of
Catena 153, 50–60.
groundwater depletion in North China using the Gravity Recovery and Climate
Anaya-Acevedo, J.A., Escobar-Martínez, J.F., Massone, H., Booman, G., Quiroz-
Experiment (GRACE) data and ground-based measurements. Water Resour. Res. 49,
Londoño, O.M., Cañón-Barriga, C.C., Montoya-Jaramillo, L.J., Palomino-Ángel, S.,
2110–2118.
2017. Identification of wetland areas in the context of agricultural development
Frappart, F., Ramillien, G., Famiglietti, J.S., 2011. Water balance of the Arctic drainage
using Remote Sensing and GIS. Dyna 84, 186–194.
system using GRACE gravimetry products. Int. J. Remote Sens. 32, 431–453.
Angarita, H., Wickel, A.J., Sieber, J., Chavarro, J., Maldonado-Ocampo, J.A., Herrera-
Gonçalves, R.D., Stollberg, R., Weiss, H., Chang, H.K., 2020. Using GRACE to quantify
R, G.A., Delgado, J., Purkey, D., 2018. Basin-scale impacts of hydropower
the depletion of terrestrial water storage in Northeastern Brazil: the Urucuia Aquifer
development on the Mompós Depression wetlands, Colombia. Hydrol. Earth Syst.
System. Sci. Total Environ. 705, 135845.
Sci. 22, 2839.
Arnesen, A.S., Silva, T.S., Hess, L.L., Novo, E.M., Rudorff, C.M., Chapman, B.D.,
McDonald, K.C., 2013. Monitoring flood extent in the lower Amazon River
11
Grings, F., Salvia, M., Karszenbaum, H., Ferrazzoli, P., Kandus, P., Perna, P., 2009. Long, D., Scanlon, B.R., Longuevergne, L., Sun, A.Y., Fernando, D.N., Save, H., 2013.
Exploring the capacity of radar remote sensing to estimate wetland marshes water GRACE satellite monitoring of large depletion in water storage in response to the
storage. J. Environ. Manage. 90, 2189–2198. 2011 drought in Texas. Geophys. Res. Lett. 40, 3395–3401.
Guarín Giraldo, G.W., Poveda, G., 2013. Variabilidad Espacial y Temporal del Long, D., Shen, Y., Sun, A., Hong, Y., Longuevergne, L., Yang, Y., Li, B., Chen, L., 2014.
Almacenamiento de Agua en el Suelo en Colombia. Revista de la Academia Drought and flood monitoring for a large karst plateau in Southwest China using
Colombiana de Ciencias Exactas, Físicas y Naturales 37, 89–113. extended GRACE data. Remote Sens. Environ. 155, 145–160.
Hachborn, E., Berg, A., Levison, J., Ambadan, J.T., 2017. Sensitivity of GRACE-derived López López, P., Immerzeel, W.W., Rodríguez Sandoval, E.A., Sterk, G., Schellekens, J.,
estimates of groundwater-level changes in southern Ontario, Canada. Hydrogeol. J. 2018. Spatial downscaling of satellite-based precipitation and its impact on
25, 2391–2402. discharge simulations in the Magdalena River basin in Colombia. Front. Earth Sci. 6.
Han, S.-C., Kim, H., Yeo, I.-Y., Yeh, P., Oki, T., Seo, K.-W., Alsdorf, D., Luthcke, S.B., Lv, M., Ma, Z., Yuan, X., Lv, M., Li, M., Zheng, Z., 2017. Water budget closure based on
2009. Dynamics of surface water storage in the Amazon inferred from measurements GRACE measurements and reconstructed evapotranspiration using GLDAS and water
of inter-satellite distance change. Geophys. Res. Lett. 36. use data for two large densely-populated mid-latitude basins. J. Hydrol. 547,
Harris, I., Jones, P., 2017. CRU TS3.24: climatic Research Unit (CRU) Time-Series (TS) 585–599.
version 3.24 of high resolution gridded data of month-by-month variation in climate Magrin, G., Marengo, J., Boulanger, J.-P., Buckeridge, M., Castellanos, E., Poveda, G.,
(Jan. 1901–Dec. 2015). Centre for Environmental Data Analysis. Scarano, F., Vicuña, S., 2014. Regional Aspects: Central and South America. In
Hassan, A., Jin, S., 2016. Water storage changes and balances in Africa observed by Climate Change 2014: Impacts, Adaptation, and Vulnerability, Contribution of
GRACE and hydrologic models. Geodesy Geodyn. 7, 39–49. Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on
Hoyos, N., Correa-Metrio, A., Sisa, A., Ramos-Fabiel, M., Espinosa, J., Restrepo, J., Climate Change, Volume II. Cambridge University Press.
Escobar, J., 2017. The environmental envelope of fires in the Colombian Caribbean. Martens, B., Gonzalez Miralles, D., Lievens, H., Van Der Schalie, R., De Jeu, R.A.,
Appl. Geogr. 84, 42–54. Fernández-Prieto, D., Beck, H.E., Dorigo, W., Verhoest, N., 2017. GLEAM v3:
Hoyos, N., Escobar, J., Restrepo, J., Arango, A., Ortiz, J., 2013. Impact of the 2010–2011 Satellite-based land evaporation and root-zone soil moisture. Geoscientific Model
La Niña phenomenon in Colombia, South America: the human toll of an extreme Develop. 10, 1903–1925.
weather event. Appl. Geogr. 39, 16–25. Martínez, R., Zambrano, E., Nieto, J.J., Hernández, J., Costa, F., 2017. Evolución,
Huang, J., Pavlic, G., Rivera, A., Palombi, D., Smerdon, B., 2016. Mapping groundwater vulnerabilidad e impactos económicos y sociales de El Niño 2015–2016 en América
storage variations with GRACE: a case study in Alberta, Canada. Hydrogeol. J. 24, Latina. Investigaciones geográficas, (pp. 65–78).
1663–1680. Miralles, D., Holmes, T., De Jeu, R., Gash, J., Meesters, A., Dolman, A., 2011. Global
Huang, Y., Salama, M., Krol, M.S., Van Der Velde, R., Hoekstra, A.Y., Zhou, Y., Su, Z., land-surface evaporation estimated from satellite-based observations. Hydrol. Earth
2013. Analysis of long-term terrestrial water storage variations in the Yangtze River Syst. Sci. 15, 453.
basin. Hydrol. Earth Syst. Sci. 17, 1985. Moiwo, J.P., Lu, W., Tao, F., 2012. GRACE, GLDAS and measured groundwater data
Huffman, G.J., Adler, R.F., Bolvin, D.T., Nelkin, E.J., 2010. The trmm multi-satellite products show water storage loss in Western Jilin, China. Water Sci. Technol. 65,
precipitation analysis (tmpa). In: Satellite Rainfall Applications for Surface 1606–1614.
Hydrology. Springer, pp. 3–22. Montanari, A., Young, G., Savenije, H., Hughes, D., Wagener, T., Ren, L.,
Huffman, G.J., Bolvin, D.T., Nelkin, E.J., Wolff, D.B., Adler, R.F., Gu, G., Hong, Y., Koutsoyiannis, D., Cudennec, C., Toth, E., Grimaldi, S., et al., 2013. Panta Rhei-
Bowman, K.P., Stocker, E.F., 2007. The TRMM multisatellite precipitation analysis everything flows: change in hydrology and society-the IAHS scientific decade
(TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine 2013–2022. Hydrol. Sci. J. 58, 1256–1275.
scales. J. Hydrometeorol. 8, 38–55. Muggeo, V.M., et al., 2008. Segmented: an R package to fit regression models with
Huss, M., Bookhagen, B., Huggel, C., Jacobsen, D., Bradley, R.S., Clague, J.J., Vuille, M., broken-line relationships. R news 8, 20–25.
Buytaert, W., Cayan, D.R., Greenwood, G., et al., 2017. Toward mountains without Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A., Kent, J., 2000.
permanent snow and ice. Earth’s Future 5, 418–435. Biodiversity hotspots for conservation priorities. Nature 403, 853–858.
IDEAM, 2015. Estudio Nacional del Agua 2014. Nanteza, J., de Linage, C., Thomas, B., Famiglietti, J., 2016. Monitoring groundwater
IDEAM, 2019. Estudio Nacional del Agua 2018. storage changes in complex basement aquifers: an evaluation of the GRACE satellites
IDEAM, & CORMAGDALENA, 2001. Estudio Ambiental de la Cuenca Magdalena – Cauca over East Africa. Water Resour. Res. 52, 9542–9564.
y Elementos para su Ordenamiento Territorial – Resumen Ejecutivo. Nardini, A., Franco Idarraga, F., 2016. A Qualitative Hydro-Geomorphic Prediction of the
IDEAM, PNUD, MADS, DNP, CANCILLERÍA, 2015. Nuevos Escenarios de Cambio Destiny of the Mojana Region (Magdalena-Cauca Basin, Colombia), to Inform Large
Climático para Colombia 2011-2100 Herramientas Científicas para la Toma de Scale Decision Making. Resources 5, 22.
Decisiones–Enfoque Nacional-Regional: Tercera Comunicación Nacional de Cambio Ni, S., Chen, J., Wilson, C.R., Li, J., Hu, X., Fu, R., 2018. Global terrestrial water storage
Climático. Bogotá DC: IDEAM. changes and connections to ENSO events. Surveys Geophys. 39, 1–22.
Jaramillo-Mejía, M.C., Chernichovsky, D., 2019. Impact of desertification and land O’Loughlin, F., Neal, J.C., Schumann, G., Beighley, E., Bates, P.D., 2018. Inundation
degradation on Colombian children. Int. J. Public Health 64, 67–73. Extents and Volumes of the Congo’s Wetlands. In AGU Fall Meeting Abstracts.
Jiang, D., Wang, J., Huang, Y., Zhou, K., Ding, X., Fu, J., 2014. The review of GRACE data Ortiz, M.A.D.-G., González, J.D.N., López, T.S., 2005. Páramos: hidrosistemas sensibles.
applications in terrestrial hydrology monitoring. Adv. Meteorol. Revista de ingeniería 64–75.
Josse, C., Cuesta, F., Navarro, G., Barrena, V., Becerra, M.T., Cabrera, E., Chacón- Ospina, M.D.L., Vargas, J.C.A., 2018. Monitoring runoff coefficients and groundwater
Moreno, E., Ferreira, W., Peralvo, M., Saito, J., et al., 2011. Physical geography and levels using data from GRACE, GLDAS, and hydrometeorological stations: analysis of
ecosystems in the tropical Andes. SK Herzog, R. Martínez, PM Jørgensen y H. Tiessen a Colombian foreland basin. Hydrogeol. J.
(comps.), Climate Change and Biodiversity in the Tropical Andes. São José dos Ouma, Y.O., Aballa, D., Marinda, D., Tateishi, R., Hahn, M., 2015. Use of GRACE time-
Campos y París: Instituto Interamericano para la Investigación del Cambio Global y variable data and GLDAS-LSM for estimating groundwater storage variability at
Comité Científico sobre Problemas del Medio Ambiente. small basin scales: a case study of the Nzoia River Basin. Int. J. Remote Sens. 36,
Kim, J.-W., Lu, Z., Lee, H., Shum, C., Swarzenski, C.M., Doyle, T.W., Baek, S.-H., 2009. 5707–5736.
Integrated analysis of PALSAR/Radarsat-1 InSAR and ENVISAT altimeter data for Pabón Caicedo, J.D., 2012. Cambio climático en Colombia: tendencias en la segunda
mapping of absolute water level changes in Louisiana wetlands. Remote Sens. mitad del siglo XX y escenarios posibles para el siglo XXI. Revista de la Academia
Environ. 113, 2356–2365. Colombiana de Ciencias Exactas, Físicas y Naturales 36, 261–278.
Landerer, F.W., Swenson, S., 2012. Accuracy of scaled GRACE terrestrial water storage Palomino-Ángel, S., Anaya-Acevedo, J.A., Simard, M., Liao, T.-H., Jaramillo, F., 2019.
estimates. Water Resour. Res. 48. Analysis of floodplain dynamics in the Atrato River Colombia using SAR
Lee, H., Beighley, R.E., Alsdorf, D., Jung, H.C., Shum, C., Duan, J., Guo, J., Yamazaki, D., interferometry. Water 11, 875.
Andreadis, K., 2011. Characterization of terrestrial water dynamics in the Congo Papa, F., Frappart, F., Malbeteau, Y., Shamsudduha, M., Vuruputur, V., Sekhar, M.,
Basin using GRACE and satellite radar altimetry. Remote Sens. Environ. 115, Ramillien, G., Prigent, C., Aires, F., Pandey, R.K., et al., 2015. Satellite-derived
3530–3538. surface and sub-surface water storage in the Ganges-Brahmaputra River Basin.
Li, J., Chen, J., Ni, S., Tang, L., Hu, X., 2019. Long-term and inter-annual mass changes of J. Hydrol.: Regional Stud. 4, 15–35.
Patagonia Ice Field from GRACE. Geodesy Geodyn. 10, 100–109. Phillips, T., Nerem, R., Fox-Kemper, B., Famiglietti, J., Rajagopalan, B., 2012. The
Liesch, T., Ohmer, M., 2016. Comparison of GRACE data and groundwater levels for the influence of ENSO on global terrestrial water storage using GRACE. Geophys. Res.
assessment of groundwater depletion in Jordan. Hydrogeol. J. 24, 1547–1563. Lett. 39.
Liesenberg, V., Galvão, L.S., Ponzoni, F.J., 2007. Variations in reflectance with Pope, K.O., Rejmankova, E., Paris, J.F., Woodruff, R., 1997. Detecting seasonal flooding
seasonality and viewing geometry: implications for classification of Brazilian cycles in marshes of the Yucatan Peninsula with SIR-C polarimetric radar imagery.
savanna physiognomies with MISR/Terra data. Remote Sens. Environ. 107, Remote Sens. Environ. 59, 157–166.
276–286. Poveda, G., 2004. La hidroclimatología de Colombia: una síntesis desde la escala inter-
Long, D., Chen, X., Scanlon, B.R., Wada, Y., Hong, Y., Singh, V.P., Chen, Y., Wang, C., decadal hasta la escala diurna. Rev. Acad. Colomb. Cienc 28, 201–222.
Han, Z., Yang, W., 2016. Have GRACE satellites overestimated groundwater Poveda, G., Pineda, K., 2009. Reassessment of Colombia’s tropical glaciers retreat rates:
depletion in the Northwest India Aquifer? Scientific Rep. 6, 24398. are they bound to disappear during the 2010–2020 decade? Adv. Geosci. 22,
Long, D., Longuevergne, L., Scanlon, B.R., 2015. Global analysis of approaches for 107–116.
deriving total water storage changes from GRACE satellites. Water Resour. Res. 51, Rabatel, A., Francou, B., Soruco, A., Gomez, J., Cáceres, B., Ceballos, J.L., Basantes, R.,
2574–2594. Vuille, M., Sicart, J.-E., Huggel, C., Scheel, M., Lejeune, Y., Arnaud, Y., Collet, M.,
Long, D., Pan, Y., Zhou, J., Chen, Y., Hou, X., Hong, Y., Scanlon, B.R., Longuevergne, L., Condom, T., Consoli, G., Favier, V., Jomelli, V., Galarraga, R., Ginot, P.,
2017. Global analysis of spatiotemporal variability in merged total water storage Maisincho, L., Mendoza, J., Ménégoz, M., Ramirez, E., Ribstein, P., Suarez, W.,
changes using multiple GRACE products and global hydrological models. Remote Villacis, M., Wagnon, P., 2013. Current state of glaciers in the tropical Andes: a
Sens. Environ. 192, 198–216. multi-century perspective on glacier evolution and climate change. Cryosphere 7,
12
81–102. https://doi.org/10.5194/tc-7-81-2013. URL:/https://www.the-cryosphere. Sproles, E., Leibowitz, S., Reager, J., et al., 2015. GRACE storage-runoff hystereses reveal
net/7/81/2013/. the dynamics of regional watersheds. Hydrol. Earth Syst. Sci. 19.
Ramillien, G., Frappart, F., Güntner, A., Ngo-Duc, T., Cazenave, A., Laval, K., 2006. Time Swenson, S., Wahr, J., 2006. Post-processing removal of correlated errors in GRACE data.
variations of the regional evapotranspiration rate from Gravity Recovery and Geophys. Res. Lett. 33.
Climate Experiment (GRACE) satellite gravimetry. Water Resour. Res. 42. Swenson, S., Wahr, J., 2009. Monitoring the water balance of Lake Victoria, East Africa,
Rangel-Ch, J.O., 2015. La biodiversidad de Colombia: significado y distribución regional. from space. J. Hydrol. 370, 163–176.
Revista de la Academia Colombiana de Ciencias Exactas, Físicas y Naturales 39, Syed, T.H., Famiglietti, J.S., Rodell, M., Chen, J., Wilson, C.R., 2008. Analysis of
176–200. terrestrial water storage changes from GRACE and GLDAS. Water Resour. Res. 44.
Reager, J., Thomas, B., Famiglietti, J., 2014. River basin flood potential inferred using Tang, J., Cheng, H., Liu, L., 2014. Assessing the recent droughts in Southwestern China
GRACE gravity observations at several months lead time. Nat. Geosci. 7, 588. using satellite gravimetry. Water Resour. Res. 50, 3030–3038.
Restrepo, J., Kjerfve, B., 2000. Magdalena river: interannual variability (1975–1995) and Tang, Q., Zhang, X., 2011. Mapping evapotranspiration using MODIS and GRACE data.
revised water discharge and sediment load estimates. J. Hydrol. 235, 137–149. In: 2011 19th International Conference on Geoinformatics. IEEE. pp. 1–4.
Rodell, M., Chen, J., Kato, H., Famiglietti, J.S., Nigro, J., Wilson, C.R., 2007. Estimating Tangdamrongsub, N., Ditmar, P., Steele-Dunne, S., Gunter, B., Sutanudjaja, E., 2016.
groundwater storage changes in the Mississippi River basin (USA) using GRACE. Assessing total water storage and identifying flood events over Tonlé Sap basin in
Hydrogeol. J. 15, 159–166. Cambodia using GRACE and MODIS satellite observations combined with
Rodell, M., Houser, P., Jambor, U., Gottschalck, J., Mitchell, K., Meng, C.-J., hydrological models. Remote Sens. Environ. 181, 162–173.
Arsenault, K., Cosgrove, B., Radakovich, J., Bosilovich, M., et al., 2004. The global Tapley, B.D., Bettadpur, S., Watkins, M., Reigber, C., 2004. The gravity recovery and
land data assimilation system. Bull. Am. Meteorol. Soc. 85, 381–394. climate experiment: mission overview and early results. Geophys. Res. Lett. 31.
Rodell, M., Velicogna, I., Famiglietti, J.S., 2009. Satellite-based estimates of groundwater Tiwari, V., Wahr, J., Swenson, S., 2009. Dwindling groundwater resources in northern
depletion in India. Nature 460, 999. India, from satellite gravity observations. Geophys. Res. Lett. 36.
Rodríguez, E., Salazar, J.F., Villegas, J.C., Mercado-Bettín, D., 2018. Assessing changes in Urrea, V., Ochoa, A., Mesa, O., 2019. Seasonality of Rainfall in Colombia. Water Resour.
extreme river flow regulation from non-stationarity in hydrological scaling laws. Res. 55, 4149–4162.
J. Hydrol. 562, 492–501. Voss, K.A., Famiglietti, J.S., Lo, M., De Linage, C., Rodell, M., Swenson, S.C., 2013.
Rodríguez, N., Armenteras, D., 2005. Ecosistemas naturales de la cuenca del río Groundwater depletion in the Middle East from GRACE with implications for
Magdalena. In: Los sedimentos del río Magdalena: reflejo de la crisis ambiental. transboundary water management in the Tigris-Euphrates-Western Iran region.
Universidad Eafit, pp. 79–98. Water Resour. Res. 49, 904–914.
Rosenqvist, A., Shimada, M., Ito, N., Watanabe, M., 2007. ALOS PALSAR: a pathfinder Vuille, M., 2013. Climate change and water resources in the tropical Andes. Technical
mission for global-scale monitoring of the environment. IEEE Trans. Geosci. Remote Report Inter-American Development Bank.
Sens. 45, 3307–3316. Wada, Y., Van Beek, L.P., Van Kempen, C.M., Reckman, J.W., Vasak, S., Bierkens, M.F.,
Rowlands, D., Luthcke, S., McCarthy, J., Klosko, S., Chinn, D., Lemoine, F., Boy, J.-P., 2010. Global depletion of groundwater resources. Geophys. Res. Lett. 37.
Sabaka, T., 2010. Global mass flux solutions from GRACE: a comparison of Wahr, J., Swenson, S., Zlotnicki, V., Velicogna, I., 2004. Time-variable gravity from
parameter estimation strategies-Mass concentrations versus Stokes coefficients. GRACE: First results. Geophys. Res. Lett. 31.
J. Geophys. Res.: Solid Earth 115. Wang, H., Guan, H., Gutiérrez-Jurado, H.A., Simmons, C.T., 2014. Examination of water
Running, S., Mu, Q., Zhao, M., 2017. MOD16A2 MODIS/Terra Net Evapotranspiration 8- budget using satellite products over Australia. J. Hydrol. 511, 546–554.
Day L4 Global 500m SIN Grid V006. NASA EOSDIS Land Processes DAAC.https:// Wang, S., Li, J., 2016. Terrestrial water storage climatology for Canada from GRACE
doi.org/10.5067/MODIS/MOD16A2, 6. satellite observations in 2002–2014. Can. J. Remote Sens. 42, 190–202.
Sakumura, C., Bettadpur, S., Bruinsma, S., 2014. Ensemble prediction and Watkins, M.M., Wiese, D.N., Yuan, D.-N., Boening, C., Landerer, F.W., 2015. Improved
intercomparison analysis of GRACE time-variable gravity field models. Geophys. methods for observing Earth’s time variable mass distribution with GRACE using
Res. Lett. 41, 1389–1397. spherical cap mascons. J. Geophys. Res.: Solid Earth 120, 2648–2671.
Salazar, J.F., Villegas, J.C., Rendón, A.M., Rodríguez, E., Hoyos, I., Mercado-Bettín, D., Waylen, P., Poveda, G., 2002. El Niño-Southern Oscillation and aspects of western South
Poveda, G., 2018. Scaling properties reveal regulation of river flows in the Amazon American hydro-climatology. Hydrol. Process. 16, 1247–1260.
through a forest reservoir. Hydrol. Earth Syst. Sci. 22, 1735–1748. Wiese, D.N., Landerer, F.W., Watkins, M.M., 2016. Quantifying and reducing leakage
Save, H., Bettadpur, S., Tapley, B.D., 2016. High-resolution CSR GRACE RL05 mascons. errors in the JPL RL05M GRACE mascon solution. Water Resour. Res. 52,
J. Geophys. Res.: Solid Earth 121, 7547–7569. 7490–7502.
Scanlon, B.R., Longuevergne, L., Long, D., 2012. Ground referencing GRACE satellite Winter, T.C., 1999. Relation of streams, lakes, and wetlands to groundwater flow
estimates of groundwater storage changes in the California Central Valley. Water systems. Hydrogeol. J. 7, 28–45.
Resources Research, USA, p. 48. Wolter, K., Timlin, M.S., 1993. Monitoring ENSO in COADS with a seasonally adjusted
Scanlon, B.R., Zhang, Z., Save, H., Sun, A.Y., Schmied, H.M., van Beek, L.P., Wiese, D.N., principal. In: Proc. of the 17th Climate Diagnostics Workshop, Norman, OK, NOAA/
Wada, Y., Long, D., Reedy, R.C., et al., 2018. Global models underestimate large NMC/CAC, NSSL, Oklahoma Clim. Survey, CIMMS and the School of Meteor., Univ.
decadal declining and rising water storage trends relative to GRACE satellite data. of Oklahoma, 52, vol. 57.
Proc. Natl. Acad. Sci. 115, E1080–E1089. Xiao, R., He, X., Zhang, Y., Ferreira, V., Chang, L., 2015. Monitoring groundwater
Scanlon, B.R., Zhang, Z., Save, H., Wiese, D.N., Landerer, F.W., Long, D., variations from satellite gravimetry and hydrological models: a comparison with in-
Longuevergne, L., Chen, J., 2016. Global evaluation of new GRACE mascon products situ measurements in the Mid-Atlantic region of the United States. Remote Sens. 7,
for hydrologic applications. Water Resour. Res. 52, 9412–9429. 686–703.
Schneider, U., Fuchs, T., Meyer-Christoffer, A., Rudolf, B., 2008. Global precipitation Xie, Z., Huete, A., Ma, X., Restrepo-Coupe, N., Devadas, R., Clarke, K., Lewis, M., 2016.
analysis products of the GPCC. Global Precipitation Climatology Centre (GPCC), Landsat and GRACE observations of arid wetland dynamics in a dryland river system
DWD, Internet Publikation, p. 112. under multi-decadal hydroclimatic extremes. J. Hydrol. 543, 818–831.
Seo, J.Y., Lee, S.-I., 2016. Integration of GRACE, ground observation, and land-surface Yang, P., Chen, Y., 2015. An analysis of terrestrial water storage variations from GRACE
models for groundwater storage variations in South Korea. Int. J. Remote Sens. 37, and GLDAS: the Tianshan Mountains and its adjacent areas, central Asia. Quatern.
5786–5801. Int. 358, 106–112.
Seyoum, W.M., Milewski, A.M., 2016. Monitoring and comparison of terrestrial water Yang, P., Xia, J., Zhan, C., Qiao, Y., Wang, Y., 2017. Monitoring the spatio-temporal
storage changes in the northern high plains using GRACE and in-situ based changes of terrestrial water storage using GRACE data in the Tarim River basin
integrated hydrologic model estimates. Adv. Water Resour. 94, 31–44. between 2002 and 2015. Sci. Total Environ. 595, 218–228.
Shamsudduha, M., Taylor, R., Longuevergne, L., 2012. Monitoring groundwater storage Yuan, T., Lee, H., Jung, H., 2015. Toward estimating wetland water level changes based
changes in the highly seasonal humid tropics: Validation of GRACE measurements in on hydrological sensitivity analysis of PALSAR backscattering coefficients over
the Bengal Basin. Water Resour. Res. 48. different vegetation fields. Remote Sens. 7, 3153–3183.
Shamsudduha, M., Taylor, R.G., Jones, D., Longuevergne, L., Owor, M., Tindimugaya, C., Zhang, Z., Chao, B., Chen, J., Wilson, C., 2015. Terrestrial water storage anomalies of
2017. Recent changes in terrestrial water storage in the Upper Nile Basin: an Yangtze River Basin droughts observed by GRACE and connections with ENSO.
evaluation of commonly used gridded GRACE products. Hydrol. Earth Syst. Sci. 21, Global Planet. Change 126, 35–45.
4533–4549. Zhao, Q., Zhang, B., Yao, Y., Wu, W., Meng, G., Chen, Q., 2019. Geodetic and
Sheffield, J., Ferguson, C.R., Troy, T.J., Wood, E.F., McCabe, M.F., 2009. Closing the hydrological measurements reveal the recent acceleration of groundwater depletion
terrestrial water budget from satellite remote sensing. Geophys. Res. Lett. 36. in North China Plain. J. Hydrol. 575, 1065–1072.
Shen, H., Leblanc, M., Tweed, S., Liu, W., 2015. Groundwater depletion in the Hai River
Basin, China, from in situ and GRACE observations. Hydrol. Sci. J. 60, 671–687.
13

Bolaños

Uploaded by

Copyright:

Available Formats

Bolaños

Uploaded by

Document Information

Original Title

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

Bolaños

Uploaded by

Copyright:

Available Formats

Journal of Hydrology xxx (xxxx) xxx

Contents lists available at ScienceDirect

GRACE reveals depletion of water storage in northwestern South America

as the residual of TWS after subtracting SWS . SWS includes multiple

2011; Anaya-Acevedo et al., 2017; Yuan et al., 2015; Palomino-Ángel

2004, 2006, 2009, 2012 and 2013 (Fig. 3a).

and results in the worst representation of temporal variance (standard

Fig. 3. Comparison between GRACE-based esti

December, 2010) and minimum (TWS = − 21.48 cm in March, 2016)

You might also like

Bolaños

Uploaded by

Copyright:

Available Formats

Bolaños

Uploaded by

Document Information

Original Title

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

Bolaños

Uploaded by

Copyright:

Available Formats

Journal of Hydrology xxx (xxxx) xxx

Contents lists available at ScienceDirect

GRACE reveals depletion of water storage in northwestern South America

as the residual of TWS after subtracting SWS . SWS includes multiple

2011; Anaya-Acevedo et al., 2017; Yuan et al., 2015; Palomino-Ángel

2004, 2006, 2009, 2012 and 2013 (Fig. 3a).

and results in the worst representation of temporal variance (standard

Fig. 3. Comparison between GRACE-based esti­

December, 2010) and minimum (TWS = − 21.48 cm in March, 2016)

You might also like

Fig. 3. Comparison between GRACE-based esti