1. Introduction
Extreme hydrological events such as floods and droughts have a profound impact on society and considerably affect human safety, water supply, food production, and energy generation. In particular, severe and prolonged droughts are the cause of famine, energy rationing, and blackouts and also contribute to increased incidence of social conflict (Raleigh and Urdal 2007). Brazil, covering 8.5 × 106 km2 with a population of 200 million people, has the largest reserve of surface renewable water in the world. According to Global Land Data Assimilation System (GLDAS; Rodell et al. 2004) model ensemble outputs, the country produces about 20% of global inland water that flows into the oceans. However, water availability across the country is highly variable. Based on GLDAS, the rainy northern region, with an average precipitation rate of 6 mm day−1, produces 65% of Brazil’s total runoff. The southeastern Brazil (SE) and northeastern Brazil (NE) regions, with lower precipitation rates of 3.8 and 2.9 mm day−1, respectively, only contribute 7% and 6% of the total surface water and include most of the country’s population, 39% and 25%, respectively. The southeast, in particular, has the highest gross domestic production, primarily owing to the economic productivity of São Paulo and Rio de Janeiro, with approximately 11 million and 6 million people, respectively.
Eastern Brazil, with an area of about 2.5 × 106 km2, has been severely affected by droughts since 2012 that followed a prolonged period of above-average temperatures coupled with low precipitation rates. These conditions depleted both natural surface and groundwater storages and had far-reaching impacts on Brazil’s agricultural and electrical production and drinking and irrigation water supply. About 32% of Brazil’s water is used for agriculture, primarily in the southern and southeastern regions. Local newspapers reported massive agricultural loss in the 2014 harvest, reaching millions of dollars, because of the droughts (Rapoza 2014). In the southeastern and midwestern regions, hydropower generation, which provides more than 70% of total electricity consumed in Brazil, has been jeopardized by prolonged water depletion.
The water supply of 20 million residents of São Paulo, along with tens of millions more people in eastern Brazil, is also in jeopardy. Data provided by the utility company of São Paulo (Sabesp) show that state’s main reservoir system, the Cantareira, suffered an intense depletion and has reached historically low water storage in recent months: as of 4 September 2014, the system’s water storage was 10.7% of its total capacity, far below the 46.5%, 67.3%, and 80.6% observed in 2013, 2012, and 2011, respectively. Just as in most districts in the northeast, the water crisis has forced Sabesp and the São Paulo state government to take the emergency measure of rationing water in numerous cities (Reuters 2014).
Previous studies have taken advantage of terrestrial water storage anomalies (TWSA) derived from the Gravity Recovery and Climate Experiment (GRACE) mission (Tapley et al. 2004) in order to identify droughts and quantify their severity worldwide (Rodell et al. 2009; Famiglietti and Rodell 2013; Thomas et al. 2014). The GRACE mission, designed to measure changes in Earth’s gravimetric field, enables estimation of TWSA over large areas. TWSA is an aggregate estimate, including surface water (rivers, wetlands, and lakes) and subsurface (soil moisture and groundwater) reservoirs, along with water on the leaves and snow. GRACE data have been used to determine the evapotranspiration over land (Ramillien et al. 2006; Rodell et al. 2011), groundwater storage change (Frappart et al. 2011) and depletion (Rodell et al. 2009), impacts of severe floods on water storage (Espinoza et al. 2013), and drought detection (Famiglietti and Rodell 2013) and its characterization (Thomas et al. 2014). GRACE estimates of TWSA have also been used in data assimilation schemes (Zaitchik et al. 2008; Li et al. 2012) and in the calibration (Werth et al. 2009) and evaluation (Getirana et al. 2014) of hydrological models. In this context, the objective of this study is to characterize and quantify the extended drought currently occurring in Brazil using GRACE TWSA, complemented by GLDAS vertical flux data. Special attention is given to austral fall periods [April–June (AMJ)], which follow the rainy summer season, over southeastern and northeastern Brazil. These regions have approximate areas of 9.3 × 105 and 1.56 × 106 km2, respectively. Two numerical tests are used in order to identify major breaks in GRACE TWSA time series.
Another goal of this study is to determine the correlation between surface water storage change in large reservoirs and GRACE TWSA in an extreme drought context. A previous study evaluated the ability of GRACE to detect the Three Gorges Dam construction by correcting TWSA using model outputs (Wang et al. 2011). In this paper, however, water storage observations at 16 major reservoirs in the impacted area are directly compared against monthly GRACE TWSA.
2. Datasets and methods
Version RL05 spherical harmonics fields (Landerer and Swenson 2012) of GRACE monthly mass grids produced by the University of Texas Center for Space Research (CSR) are used in this study. Numerous analyses have considered the CSR data (e.g., Xavier et al. 2010; Famiglietti and Rodell 2013; Voss et al. 2013; Thomas et al. 2014). These data are smoothed using a 200-km half-width Gaussian filter and are provided on a 1° global grid at a monthly time step. As of the development of this study, the RL05 product is available from April 2002 to January 2015 (with the mean value of 2004–09 removed). When analyzing GRACE data, there is a trade-off between spatial resolution and accuracy, since 1.5 × 105 km2 is the approximate minimum area that can be resolved before errors overwhelm the signal (Rowlands et al. 2005; Swenson et al. 2006). GRACE data accuracy has been estimated to be less than 1 cm in equivalent water height, when averaged over areas larger than about 4 × 105 km2, and errors increase as the area under observation decreases (Swenson et al. 2003). In this sense, GRACE data are suitable for hydrological studies over northeastern and southeastern Brazil.
Similar to what Thomas et al. (2014) proposed as a means to assess drought severity, the relative water deficit across eastern Brazil is quantified by removing the seasonal cycle from the TWSA time series. In this sense, the water deficit is estimated by subtracting their respective mean seasonal cycles (MSCs) from the TWSA series. MSC is determined from the average of each month, January–December, over the period of data availability.
To detect the presence of breaks and changes in the MSC-detrended TWSA time series over eastern Brazil, two methods were used: the Pettitt (1979) method and the Hubert segmentation (Hubert et al. 1989). The Pettitt method is a nonparametric test based on changes in the average and the range of the series subdivided into subseries. Hubert segmentation detects the multiple breaks in time series by verifying whether differences in average and standard deviation among periods are significant. These tests have supported previous studies in characterizing time series of different hydroclimatological variables (e.g., Hubert et al. 1989; Espinoza et al. 2009).
To better understand the hydrological processes over the region and to support the water deficit evaluation, precipitation P, evapotranspiration (ET), total runoff R, and soil moisture S derived from four GLDAS land surface models (LSMs) were used as supplementary data. The models are CLM (Dai et al. 2003), Mosaic (Koster and Suarez 1996), Noah (Ek et al. 2003), and VIC (Liang et al. 1994). These products are provided at a 1° resolution and monthly time step from 1979 to present. Precipitation is a satellite–gauge merged product based on the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) 5-day-interval real-time data. CMAP uses satellite- and reanalysis-based precipitation estimates weighted to fit gauge-based observations. The quality of the analysis is strongly dependent on the amount of gauge data available as well as the accuracy of the satellite estimates. Comparisons against the Global Precipitation Climatology Project (GPCP; Adler et al. 2003) over South America indicate that CMAP underestimates average precipitation rates by 0.18 and 0.10 mm day−1 in January and July, respectively (Bosilovich et al. 2008). The pentad precipitation is disaggregated to 6-hourly resolution using precipitation fields from the National Centers for Environmental Prediction (NCEP) Global Data Assimilation System (GDAS; Derber et al. 1991). Air temperature is derived from GDAS.
Daily water storage observations at 16 reservoirs located within southeastern Brazil are used to quantify impacts of surface water storage change on GRACE TWSA. Water storage time series are made available by the Brazilian Operator of Electric System (ONS) from August 2005 to present. Based on ONS data, these reservoirs correspond to about 37% of total water stored in man-made reservoirs in Brazil. Table 1 lists the reservoirs and their main characteristics. GRACE TWSA time series were extracted from a 3 × 3 gridcell window (~9 × 104 km2) consisting of the overlying grid cell and the eight adjacent cells. These time series are compared against monthly averaged ground-based reservoir water storage change (RWSC) located within these windows. GRACE TWSA time series were extracted from 3 × 3 gridcell boxes in order to be compared against monthly averaged water storage change in reservoirs located within these windows.
Selected reservoirs within southeastern Brazil. Water extent was derived from Landsat imagery. Amplitudes were computed as the difference between the highest and lowest values in the monthly time series. TWSA (km3) was derived from 3 × 3 gridcell boxes where reservoirs are located.
3. Results
The water deficit’s spatial expansion for AMJ from 2012 to 2014 is illustrated in Fig. 1. All of eastern Brazil and parts of midwestern Brazil experienced a substantial depletion of surface and groundwater with aggravated conditions across the southeast by 2014. The impacts of droughts on eastern Brazil resulted in an abrupt decay of TWSA in the last three years. As illustrated in Fig. 2a, based on a linear regression over the three most recent years of GRACE TWSA data, that is, from February 2012 to January 2015, water depletion rates are −6.1 and −3.2 cm yr−1 in the SE and NE regions, corresponding to −56 and −49 km3 yr−1.
GRACE-based TWSA subtracted from the 2002–14 average during austral falls (AMJ) in (a) 2014, (b) 2013, and (c) 2012. Black lines delineate Brazilian regions.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
GRACE-based TWSA subtracted from the 2002–14 average during austral falls (AMJ) in (a) 2014, (b) 2013, and (c) 2012. Black lines delineate Brazilian regions.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
GRACE-based TWSA subtracted from the 2002–14 average during austral falls (AMJ) in (a) 2014, (b) 2013, and (c) 2012. Black lines delineate Brazilian regions.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
Time series of (a) GRACE-based TWSA, (b) TWSA with the mean seasonal cycle removed, (c) S with the mean seasonal cycle removed, (d) P with the mean seasonal cycle removed, (e) R with the mean seasonal cycle removed, and (f) ET. TWSA time series starts in April 2002. The S, P, R, and ET values are derived from the 1° monthly averaged GLDAS products and also had their mean seasonal cycles removed. Colors represent the SE (red) and NE (blue) regions. Dashed lines in (b) correspond to averages of three subseries defined by Hubert’s segmentation method. Average values are also provided.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
Time series of (a) GRACE-based TWSA, (b) TWSA with the mean seasonal cycle removed, (c) S with the mean seasonal cycle removed, (d) P with the mean seasonal cycle removed, (e) R with the mean seasonal cycle removed, and (f) ET. TWSA time series starts in April 2002. The S, P, R, and ET values are derived from the 1° monthly averaged GLDAS products and also had their mean seasonal cycles removed. Colors represent the SE (red) and NE (blue) regions. Dashed lines in (b) correspond to averages of three subseries defined by Hubert’s segmentation method. Average values are also provided.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
Time series of (a) GRACE-based TWSA, (b) TWSA with the mean seasonal cycle removed, (c) S with the mean seasonal cycle removed, (d) P with the mean seasonal cycle removed, (e) R with the mean seasonal cycle removed, and (f) ET. TWSA time series starts in April 2002. The S, P, R, and ET values are derived from the 1° monthly averaged GLDAS products and also had their mean seasonal cycles removed. Colors represent the SE (red) and NE (blue) regions. Dashed lines in (b) correspond to averages of three subseries defined by Hubert’s segmentation method. Average values are also provided.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
a. Break tests
To determine whether a recent break has occurred in the time series, two break tests were performed using MSC-detrended TWSA time series over eastern Brazil. The application of Hubert’s segmentation points to a few breaks in the time series, but two dates seem to be especially significant: the transitions January–February 2004 and February–March 2012, both occurring during wet seasons (austral summer). As one can see in Fig. 2b, the first one defines the end of an extended and severe dry season that began in the early 2000s, causing extreme reservoir depletions and subsequent nationwide electricity crisis and blackouts in 2001. The second one is identified in the time series as the transition between an unusually wet season (with a few dry months) and an extended more-than-usual dry season. Contrasting MSC-detrended TWSA averages for the SE and NE regions over the three periods delimited by these two transitions are −3.5 and −4.3 cm (or −32 and −67 km3, respectively) in 2002–04, 2.3 cm for both regions (−21 and −36 km3) in 2004–12, and −4.1 and −3.7 cm (−38 and −58 km3) in 2012–15, demonstrating clear breaks in the time series. The Pettitt test indicates that, for all the significance levels (0.01, 0.05, and 0.1), a breakpoint occurred in February 2012. This result corroborates with those obtained with Hubert’s segmentation, evidencing that the latter was the most significant break occurring in the 13-yr monthly time series.
According to Fig. 2b, most months from 2012 to 2015 were drier than the temporal average of eastern Brazil. From February 2012 to January 2013, and from February 2013 to January 2014, SE had an average water deficit of −1.1 and −0.6 cm (or −10 and −5 km3, considering the surface area of the region), respectively. The situation deteriorated in the following 12 months, from February 2014 to January 2015, with an average water deficit of −10.3 cm (−95 km3). The driest month registered by GRACE over SE was January 2015 with a water deficit of −15.7 cm (−146 km3). The last 12 months of GRACE TWSA data can be considered as the driest period over SE ever observed by the satellite, overcoming the drought occurring in the early 2000s, when water scarcity averaged −3.4 cm (−31 km3). Similar conditions arose in NE, with water deficits of −2 cm (−32 km3) for 2012/13, −3.7 cm (−58 km3) for 2013/14, and −4.5 cm (−70 km3) for 2014/15. Note that these water deficits are average values across a given region and hotspots of especially aggravated drought may occur.
b. Vertical flux evaluation
Since GLDAS LSMs are composed of different parameterizations, a 1.90-m soil depth, as parameterized by VIC, was used as a reference for this study. In this sense, the equivalent S was determined for the same depth from other models by assuming that the moisture within soil layers is homogeneously distributed, converting it to equivalent water depth, and then summing the amount of available water within the predefined depth. A similar process of subtracting MSCs was performed to obtain the soil moisture deficit (Fig. 2c), revealing that both TWSA and S strongly agree over time, with correlation coefficients of 0.82 and 0.81 for SE and NE regions, respectively. MSC-detrended S time series confirm the severity of the 2012–15 drought in comparison to the early 2000s drought, with soil moisture deficits in November and December 2014 in SE and NE reaching −19 and −14 mm, respectively. According to the complete monthly S time series derived from average GLDAS outputs (not shown), the 2012–15 drought resulted in the driest soil in eastern Brazil over the last 35 years. Both break tests were applied to MSC-detrended S, resulting in similar breakpoints (December–January 2003 and March–April 2012). These results show consistency between the relative timing of drought derived from GLDAS and GRACE.
Recent precipitation rates are lower than the long-term average. During the 2012–15 drought, precipitation was 16% (SE) and 19% (NE) lower than the average over the GRACE-observed timespan (Fig. 2d). When compared against the average over the last 35 years, this value is slightly higher (20% and 23%, respectively). This means that those regions are experiencing a long-term dry period with extreme droughts occurring sporadically. The prolonged reduction of precipitation rate affects both surface runoff and base flow. Total runoff was highly impacted by the 2012–15 drought, dropping 44% (SE) and 49% (NE), in comparison to the 13-yr average (Fig. 2e).
The significantly reduced soil moisture has a major impact on vegetation, resulting in negative trends in ET over the last three years, with rates of −0.21 (SE) and −0.20 mm yr−1 (NE). In SE, averaged MSC-detrended ET over the 2012–15 drought does not show a significant change compared to the GRACE-observed timespan, while a difference of −0.23 mm day−1 is detected in NE. There is high correlation between MSC-detrended ET and GRACE-based TWSA over SE and NE regions (0.42 and 0.63, respectively). It is known that ET depends mainly on soil moisture in most situations. Since TWSA aggregates both surface water and groundwater storages, further evaluations are necessary to determine the dependence of evapotranspiration on TWSA.
c. Relating reservoir water storage change and GRACE TWSA
As shown above, total runoff was highly impacted by the drought, significantly decreasing surface water storage across eastern Brazil. This decline may alter water availability in reservoirs in the region. Regardless of their small extent relative to the satellite gridcell size, it is a plausible hypothesis that the surface water storage change in large reservoirs might be detected by GRACE signals. Indeed, Wang et al. (2011) have found that RWSC measurements and GRACE-derived TWSA at the Three Gorges reservoir presented an optimal correlation value r of 0.87. According to ground-based measurements at Brazilian reservoirs, maximal amplitudes of RWSC vary from 1.3 (at reservoir 16) to 15.2 km3 (reservoir 2). Maximal amplitudes were computed as the difference between the highest and lowest values in the monthly time series. As shown in Table 1, the ratio between RWSC and TWSA (converted to km3 over 3 × 3 gridcell boxes) amplitudes vary from 0.05 to 0.36. This means that RWSC can be nonnegligible, corresponding to up to 36% of TWSA amplitudes.
The water deficit’s spatial expansion shown in Fig. 1 corroborates with water storage change at reservoirs located within southeastern Brazil. According to results shown in Fig. 3, sudden water depletion is evident in RWSC. All reservoirs have registered minimum water storage in the past 12 months, and most of them show a separate cluster of dots for that period (black dots shown in the scatterplots). Indeed, the past 12 months present the lowest water storage registered in these reservoirs since 2005. In particular, reservoirs 1, 6, and 12 present a very irregular pattern during the last 12 months, indicating the presence of a different relation between RWSC and TWSA. In such extreme situations, an increase of TWSA is observed when reservoir water storages reach their minimum values.
Scatterplots of monthly GRACE TWSA (cm; ordinate) and water storage (km3; abscissa) at 16 large reservoirs located across southeastern Brazil. The seasonal cycle was removed for both variables. Black dots represent months from February 2014 to January 2015. Variables Vmax and r stand for the max water storage capacity and correlation, respectively. The map indicates locations of reservoirs.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
Scatterplots of monthly GRACE TWSA (cm; ordinate) and water storage (km3; abscissa) at 16 large reservoirs located across southeastern Brazil. The seasonal cycle was removed for both variables. Black dots represent months from February 2014 to January 2015. Variables Vmax and r stand for the max water storage capacity and correlation, respectively. The map indicates locations of reservoirs.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
Scatterplots of monthly GRACE TWSA (cm; ordinate) and water storage (km3; abscissa) at 16 large reservoirs located across southeastern Brazil. The seasonal cycle was removed for both variables. Black dots represent months from February 2014 to January 2015. Variables Vmax and r stand for the max water storage capacity and correlation, respectively. The map indicates locations of reservoirs.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0096.1
Even though reservoirs correspond to a small fraction of a GRACE grid cell [percentages vary from 2.6% (reservoir 15) to 14.7% (reservoir 2)], r values between MSC-detrended GRACE TWSA and RWSC vary between 0.42 and 0.82, with high statistical significance (p value < 10−4 for all reservoirs), indicating an overall agreement between satellite and ground-based measurements. The highest r values were obtained for two of the largest reservoirs in terms of both water storage capacity (reservoirs 2 and 3) and RWSC amplitudes. They are located within the driest region, as informed by GRACE. Small reservoirs located in the proximities of the driest region (reservoirs 14 and 15) also had significant water storage depletion and high r values. The lowest r values were obtained for small reservoirs (11, 13, and 16) located south of the driest spot. Dots representing the last 12 months in the respective scatterplots are within the main cluster of observations, indicating that, although dry, these reservoirs have not been through an extreme event. Though this comparison reveals correlation between GRACE TWSA and RWSC, estimates of total surface water (including lakes and rivers) and groundwater storage change within the selected areas are still needed to perform a conclusive evaluation of impacts of reservoirs on GRACE signals.
4. Summary
Three climate-sensitive sectors, water supply, agriculture, and energy generation, have been severely affected by the recent and prolonged drought in eastern Brazil. That region is highly populated and accommodates the major hydropower plants in the country, numerous industrial centers, and irrigation fields, which together require a continuous and large amount of water. Results presented in this paper demonstrate that large-scale extreme droughts during three subsequent austral falls in Brazil could be detected by GRACE. Two break tests [the method from Pettitt (1979) and Hubert’s segmentation procedure (Hubert et al. 1989)] were performed, pointing out two main breaks in MSC-detrended GRACE TWSA time series, determining the end and beginning of two dry seasons along 13 years of data. Comparisons between MSC-detrended GRACE TWSA and RWSC resulted in high correlations, demonstrating the ability of GRACE in detecting water volume change in reservoirs covering small fractions of TWSA product’s spatial resolution.
Despite the small reservoir surface area relative to the spatial resolution of GRACE, statistically significant correlations were found between RWSC and TWSA averaged over 3 × 3 gridcell boxes. These results could be explained by RWSC amplitude, reservoir size, and their proximity to the drought nucleus. However, the actual impact of reservoirs on GRACE TWSA is inconclusive since other factors can influence the observed correlations. These factors include the natural hydrological seasonality observed in the surrounding area (other water bodies, such as rivers and other lakes), soil moisture, and aquifers. Additionally, GRACE TWSA amplitude errors over those boxes should be on the order of 2 km3, indicating that the RWSC of small reservoirs might not significantly impact GRACE TWSA. An effective way to demonstrate such dependencies would be to perform model runs (LSMs coupled with river routing schemes capable of simulating reservoir operation rules), considering multiple scenarios, and quantitatively determine a correlation between variables. However, such an analysis is beyond the scope of this paper.
Based on complementary data from model outputs derived from GLDAS, it was demonstrated that the recent water deficit observed over eastern Brazil is mostly due to lower-than-usual precipitation rates. Though the reasons for the recent low precipitation are still unknown, there is increasing speculation that human-induced climate change (Escobar 2015) and deforestation in Amazonia (Nazareno and Laurance 2015) may be altering the moisture transport from Amazonia to southeastern South America (SESA). Indeed, recent studies have demonstrated the existence of a dipole-like structure between SESA and the South American convective zone (e.g., Junquas et al. 2012) and moisture transport across the continent through low-level jets and aerial rivers from tropical to subtropical regions of South America (e.g., Marengo et al. 2004; Poveda et al. 2014). Future studies investigating possible causes for the reduced precipitation are recommended. These studies should consider, among other factors, the simultaneous drought over eastern Brazil and floods over the Amazon (Marengo et al. 2013), potential relationships with low-level jets and aerial rivers carrying moisture from tropical to subtropical regions in South America (Marengo et al. 2004; Poveda et al. 2014), and impacts of deforestation on these fluxes.
Last, the use of remote sensing data in environmental studies can empower policy and decision-makers to develop social, economic, and environmental policies better adapted to extreme events and capable of preventing major socioeconomic losses due to extreme droughts. Because of the short time series and a nonnegligible time shift between data acquisition, processing, and availability to the scientific community, GRACE is not yet ready for use in seasonal predictions or near-real-time drought forecasts (Thomas et al. 2014). However, such data, combined with additional hydrological information from current and next-generation satellite missions (Alsdorf et al. 2007; Getirana and Peters-Lidard 2013; Papa et al. 2013; Kumar et al. 2014), as well as model outputs, can be used to diagnose recent extreme water-related events and can contribute to water scarcity predictions for upcoming dry seasons (Landerer and Swenson 2012) and be integrated into model calibration frameworks and decision support systems. Combining such techniques with a GRACE data assimilation framework (Li and Rodell 2014) and model forecasts could produce useful decision-making tools for minimizing the impacts of future extreme and prolonged drought events on water supply and energy generation.
Acknowledgments
The author would like to thank ONS and Sabesp for providing water storage at reservoirs. He is also thankful to Matthew Rodell, Jamon Van den Hoek, Jan-Carlo Espinoza, and three anonymous reviewers for their valuable comments and revision. GRACE land data (available at http://grace.jpl.nasa.gov) processing algorithms were provided by Sean Swenson and supported by the NASA MEaSUREs Program. Landsat-based water extent was provided by J. Van den Hoek. GLDAS data are available through the Goddard Earth Sciences Data and Information Services Center (GES DISC; http://disc.sci.gsfc.nasa.gov/hydrology/data-holdings).
REFERENCES
Adler, R. F., and Coauthors, 2003: The version 2 Global Precipitation Climatology Project (GPCP) Monthly Precipitation Analysis (1979–present). J. Hydrometeor., 4, 1147–1167, doi:10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Alsdorf, D., Bates P. , Melack J. , Wilson M. , and Dunne T. , 2007: Spatial and temporal complexity of the Amazon flood measured from space. Geophys. Res. Lett., 34, L08402, doi:10.1029/2007GL029447.
Bosilovich, M. G., Chen J. , Robertson F. R. , and Adler R. F. , 2008: Evaluation of global precipitation in reanalyses. J. Appl. Meteor. Climatol., 47, 2279–2299, doi:10.1175/2008JAMC1921.1.
Dai, Y., and Coauthors, 2003: The Common Land Model. Bull. Amer. Meteor. Soc., 84, 1013–1023, doi:10.1175/BAMS-84-8-1013.
Derber, J. C., Parrish D. F. , and Lord S. J. , 1991: The new global operational analysis system at the National Meteorological Center. Wea. Forecasting, 6, 538–547, doi:10.1175/1520-0434(1991)006<0538:TNGOAS>2.0.CO;2.
Ek, M., Mitchell K. , Yin L. , Rogers P. , Grunmann P. , Koren V. , Gayno G. , and Tarpley J. , 2003: Implementation of Noah land-surface model advances in the NCEP operational mesoscale Eta model. J. Geophys. Res., 108, 8851, doi:10.1029/2002JD003296.
Escobar, H., 2015: Drought triggers alarms in Brazil’s biggest metropolis. Science, 347, 812, doi:10.1126/science.347.6224.812.
Espinoza, J. C., and Coauthors, 2009: Contrasting regional discharge evolutions in the Amazon basin (1974–2004). J. Hydrol., 375, 297–311, doi:10.1016/j.jhydrol.2009.03.004.
Espinoza, J. C., Ronchail J. , Frappart F. , Lavado W. , Santini W. , and Guyot J. L. , 2013: The major floods in the Amazonas River and tributaries (western Amazon basin) during the 1970–2012 period: A focus on the 2012 flood. J. Hydrometeor., 14, 1000–1008, doi:10.1175/JHM-D-12-0100.1.
Famiglietti, J. S., and Rodell M. , 2013: Water in the balance. Science, 340, 1300–1301, doi:10.1126/science.1236460.
Frappart, F., and Coauthors, 2011: Satellite-based estimates of groundwater storage variations in large drainage basins with extensive floodplains. Remote Sens. Environ., 115, 1588–1594, doi:10.1016/j.rse.2011.02.003.
Getirana, A. C. V., and Peters-Lidard C. , 2013: Estimating water discharge from large radar altimetry datasets. Hydrol. Earth Syst. Sci., 17, 923–933, doi:10.5194/hess-17-923-2013.
Getirana, A. C. V., and Coauthors, 2014: Water balance in the Amazon basin from a land surface model ensemble. J. Hydrometeor., 15, 2586–2614, doi:10.1175/JHM-D-14-0068.1.
Hubert, P., Carbonnel J. P. , and Chaouche A. , 1989: Segmentation des séries hydrométéorologiques—Application à des séries de précipitations et de débits de l’afrique de l’ouest. J. Hydrol., 110, 349–367, doi:10.1016/0022-1694(89)90197-2.
Junquas, C., Vera C. , Li L. , and Le Treut H. , 2012: Summer precipitation variability over southeastern South America in a global warming scenario. Climate Dyn., 38, 1867–1883, doi:10.1007/s00382-011-1141-y.
Koster, R. D., and Suarez M. J. , 1996: Energy and water balance calculations in the Mosaic LSM. NASA Tech. Memo. 104606, Vol. 9, 60 pp. [Available online at http://gmao.gsfc.nasa.gov/pubs/docs/Koster130.pdf.]
Kumar, S. V., and Coauthors, 2014: Assimilation of remotely sensed soil moisture and snow depth retrievals for drought estimation. J. Hydrometeor., 15, 2446–2469, doi:10.1175/JHM-D-13-0132.1.
Landerer, F. W., and Swenson S. C. , 2012: Accuracy of scaled GRACE terrestrial water storage estimates. Water Resour. Res., 48, W04531, doi:10.1029/2011WR011453.
Li, B., and Rodell M. , 2014: Evaluation of a model-based groundwater drought indicator in the conterminous U.S. J. Hydrol., 526, 78–88, doi:10.1016/j.jhydrol.2014.09.027.
Li, B., Rodell M. , Zaitchik B. , Reichle R. , and Koster R. , 2012: Assimilation of GRACE terrestrial water storage into a land surface model: Evaluation and potential value for drought monitoring in western and central Europe. J. Hydrol., 446–447, 103–115, doi:10.1016/j.jhydrol.2012.04.035.
Liang, X., Lettenmaier D. P. , Wood E. F. , and Burges S. J. , 1994: A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophys. Res., 99, 14 415–14 428, doi:10.1029/94JD00483.
Marengo, J. A., Soares W. R. , Saulo C. , and Nicolini M. , 2004: Climatology of the low-level jet east of the Andes as derived from the NCEP–NCAR reanalyses: Characteristics and temporal variability. J. Climate, 17, 2261–2280, doi:10.1175/1520-0442(2004)017<2261:COTLJE>2.0.CO;2.
Marengo, J. A., Alves L. M. , Soares W. R. , and Rodriguez D. A. , 2013: Two contrasting severe seasonal extremes in tropical South America in 2012: Flood in Amazonia and drought in Northeast Brazil. J. Climate, 26, 9137–9154, doi:10.1175/JCLI-D-12-00642.1.
Nazareno, A. G., and Laurance W. F. , 2015: Brazil’s drought: Beware deforestation. Science, 347, 1427, doi:10.1126/science.347.6229.1427-a.
Papa, F., Frappart F. , Güntner A. , Prigent C. , Aires F. , Getirana A. C. V. , and Maurer R. , 2013: Surface freshwater storage and variability in the Amazon basin from multi-satellite observations, 1993–2007. J. Geophys. Res. Atmos., 118, 11 951–11 965, doi:10.1002/2013JD020500.
Pettitt, A. N., 1979: A non-parametric approach to the change-point problem. Appl. Stat., 28, 126–135, doi:10.2307/2346729.
Poveda, G., Jaramillo L. , and Vallejo L. F. , 2014: Seasonal precipitation patterns along pathways of South American low-level jets and aerial rivers. Water Resour. Res., 50, 98–118, doi:10.1002/2013WR014087.
Raleigh, C., and Urdal H. , 2007: Climate change, environmental degradation and armed conflict. Polit. Geogr., 26, 674–694, doi:10.1016/j.polgeo.2007.06.005.
Ramillien, G., Frappart F. , Güntner A. , Ngo-Duc T. , Cazenave A. , and Laval K. , 2006: Time variations of the regional evapotranspiration rate from Gravity Recovery and Climate Experiment (GRACE) satellite gravimetry. Water Resour. Res., 42, W10403, doi:10.1029/2005WR004331.
Rapoza, K., 2014: Brazil loses billions as crops reduced by wacky weather. Forbes, accessed 5 May 2015. [Available online at http://www.forbes.com/sites/kenrapoza/2014/03/03/brazil-loses-billions-as-crop-losses-mount-from-wacky-weather/.]
Reuters, 2014: Brazil’s largest city may ration water this year, utility says. Reuters, accessed 5 May 2015. [Available online at http://www.reuters.com/article/2014/04/09/us-brazil-water-idUSBREA380YP20140409.]
Rodell, M., and Coauthors, 2004: The Global Land Data Assimilation System. Bull. Amer. Meteor. Soc., 85, 381–394, doi:10.1175/BAMS-85-3-381.
Rodell, M., Velicogna I. , and Famiglietti J. S. , 2009: Satellite-based estimates of groundwater depletion in India. Nature, 460, 999–1002, doi:10.1038/nature08238.
Rodell, M., McWilliams E. , Famiglietti J. , Beudoing H. , and Nigro J. , 2011: Estimating evapotranspiration using an observation based terrestrial water budget. Hydrol. Processes, 25, 4082–4092, doi:10.1002/hyp.8369.
Rowlands, D. D., Luthcke S. B. , Klosko S. M. , Lemoine F. G. R. , Chinn D. S. , McCarthy J. J. , Cox C. M. , and Anderson O. B. , 2005: Resolving mass flux at high spatial and temporal resolution using GRACE intersatellite measurements. Geophys. Res. Lett., 32, L04310, doi:10.1029/2004GL021908.
Swenson, S., Wahr J. , and Milly P. C. D. , 2003: Estimated accuracies of regional water storage variations inferred from the Gravity Recovery and Climate Experiment (GRACE). Water Resour. Res., 39, 1223, doi:10.1029/2002WR001808.
Swenson, S., Yeh P. J.-F. , Wahr J. , and Famiglietti J. S. , 2006: A comparison of terrestrial water storage variations from GRACE with in situ measurements from Illinois. Geophys. Res. Lett., 33, L16401, doi:10.1029/2006GL026962.
Tapley, B. D., Bettadpur S. , Ries J. C. , Thompson P. F. , and Watkins M. M. , 2004: GRACE measurements of mass variability in the Earth system. Science, 305, 503–505, doi:10.1126/science.1099192.
Thomas, A. C., Reager J. T. , Famiglietti J. S. , and Rodell M. , 2014: A GRACE-based water storage deficit approach for hydrological drought characterization. Geophys. Res. Lett., 41, 1537–1545, doi:10.1002/2014GL059323.
Voss, K. A., Famiglietti J. S. , Lo M. , de Linage C. , Rodell M. , and Swenson S. C. , 2013: Groundwater depletion in the Middle East from GRACE with implications for transboundary water management in the Tigris–Euphrates–Western Iran region. Water Resour. Res., 49, 904–914, doi:10.1002/wrcr.20078.
Wang, X., de Linage C. , Famiglietti J. , and Zender C. S. , 2011: Gravity Recovery and Climate Experiment (GRACE) detection of water storage changes in the Three Gorges Reservoir of China and comparison with in situ measurements. Water Resour. Res., 47, W12502, doi:10.1029/2011WR010534.
Werth, S., Güntner A. , Petrovic S. , and Schmidt R. , 2009: Integration of GRACE mass variations into a global hydrological model. Earth Planet. Sci. Lett., 277, 166–173, doi:10.1016/j.epsl.2008.10.021.
Xavier, L., Becker M. , Cazenave A. , Longuevergne L. , Llovel W. , and Rotunno Filho O. C. , 2010: Interannual variability in water storage over 2003–2008 in the Amazon basin from GRACE space gravimetry, in situ river level and precipitation data. Remote Sens. Environ., 114, 1629–1637, doi:10.1016/j.rse.2010.02.005.
Xie, P., and Arkin P. , 1997: Global precipitation: A 17-yr monthly analysis based on gauged observations, satellite estimates and numerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558, doi:10.1175/1520-0477(1997)078<2539:GPAYMA>2.0.CO;2.
Zaitchik, B. F., Rodell M. , and Reichle R. H. , 2008: Assimilation of GRACE terrestrial water storage data into a land surface model: Results for the Mississippi River basin. J. Hydrometeor., 9, 535–548, doi:10.1175/2007JHM951.1.
