Assessing variability of evapotranspiration over ... - Wiley Online Library

PUBLICATIONS Water Resources Research RESEARCH ARTICLE 10.1002/2013WR013518

Assessing variability of evapotranspiration over the Ganga river basin using water balance computations

Key Points: Estimating ET using water balance computations Characterizing the variability of ET in the Ganga river basin Identifying the hydroclimatic controls of ET variability in the region

Tajdarul H. Syed1, Peter J. Webster2, and James S. Famiglietti3,4,5

Supporting Information: Readme Figure S1

Abstract A thorough assessment of evapotranspiration (ET) pervades several important issues of the 21st

Correspondence to: T. H. Syed, [email protected] Citation: Syed, T. H., P. J. Webster, and J. S. Famiglietti (2014), Assessing variability of evapotranspiration over the Ganga river basin using water balance computations, Water Resour. Res., 50, 2551–2565, doi:10.1002/ 2013WR013518. Received 10 JAN 2013 Accepted 19 FEB 2014 Accepted article online 24 FEB 2014 Published online 21 MAR 2014

1

Indian School of Mines, Department of Applied Geology, Dhanbad, Jharkhand, India, 2School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia, USA, 3UC Center for Hydrologic Modeling, University of California, Irvine, California, USA, 4Department of Earth System Science, University of California, Irvine, California, USA, 5 Department of Civil and Environmental Engineering, University of California, Irvine, California, USA

century including climate change, food-security, land-management, flood and drought prediction, and water resources assessment and management. Such a proper assessment is of particular importance in the Ganga river basin (GRB) with its backdrop of a rapidly increasing population pressure and unregulated use of water resources. Spatially averaged ET over the GRB is computed as the residual of atmospheric and terrestrial water budget computations using a combination of model simulations and satellite and groundbased observations. The best estimate of monthly ET is obtained as the monthly mean of atmospheric and terrestrial water balance computations for the period 1980–2007. The mean monthly average of ET from these various estimates is 72.3 6 18.8 mm month21. Monthly variations of ET peak between July and August and reach a minimum in February. For the entire study period, the rate of change of ET across the GRB is ~o, results allude to the 211 mm yr22 (i.e., mm/yr/yr). Alongside a notable influence of the 1997–1998 El Nin existence of interim periods during which ET trends varied significantly. More specifically, during the period of 1998–2002, the rate of decline increased to 255.8 mm yr22, which is almost 5 times the overall trend. Based on the correlation between ET and independent estimates of near-surface temperature and soil moisture, we can infer that the ET over the GRB is primarily limited by moisture availability. The analysis has important potential for use in large-scale water budget assessments and intercomparison studies. The analysis also emphasizes the importance of synergistic use of mutliplatform hydrologic information.

1. Introduction The Ganga river basin (GRB) is one of the largest river basins on the planet and accounts for 26% of India’s landmass with 950,000 km2 and one of the most densely populated areas in the world with 530 persons per sq km [Thenkabail et al., 2005] and is one of the most extensively irrigated basins in the world [National Ganga River Basin Authority (NGRBA), 2011]. Tropical and subtropical climates predominate across the entire basin with temperatures varying from around 40 C during May to 10 C during January. The mean annual average rainfall in the GRB is about 2000 mm, the majority of which occurs during the Southwest Monsoon season from June to October. About 73% of the area of the GRB is extensively cultivated while 16% of the rest is covered by forest. However, the forest ecosystem in the GRB is under severe stress due to human intervention [NGRBA, 2011]. The GRB has experienced significant changes in the patterns of land use [Kothyari and Singh, 1996], precipitation [e.g., Webster et al., 1998; Hoyos and Webster, 2007], and surface temperature [Goswami et al., 2006] over the last couple of decades. During the late 20th century, rapid economic growth, increased population pressure, and industrialization of agriculture has led to a marked increase of water use [Mall et al., 2006] which is primarily the result of anarchic extraction of groundwater [Shah, 2008]. Further, Webster and Jian [2011] have calculated that during the remainder of this century, despite an expected increase in precipitation and river discharge, freshwater availability (total river discharge/population) would reduce to half of what is currently available. The demographics of the GRB and the possible impacts of climate change point toward the need for an accurate assessment of ET. Such an assessment is necessary to allow several management and operational tasks on local to regional scales to move ahead. These tasks include watershed management [Su, 2002] to provide a balance of streamflow for storage and to maintain availability of freshwater for the populace and

SYED ET AL.

C 2014. American Geophysical Union. All Rights Reserved. V

2551

Water Resources Research

10.1002/2013WR013518

industrial and agricultural use. In addition, the determination of ET is critical for irrigation design and scheduling [Bastiaanssen et al., 2005] and weather and streamflow/flood forecasting [e.g., Hopson and Webster, 2010; Webster et al., 2010]. However, despite its critical importance, ET remains as one of the least understood components of the hydrological cycle over the GRB resulting from a lack of reliable estimates [Jain, 2012]. The best, and perhaps the most recent available estimate of ET over the GRB, based on a report by National Commission for Integrated Water Resources Development Plan [MoWR, 1999], is considered to be grossly underestimated [e.g., Jain, 2012; Narasimhan, 2008]. Thus, there exists a need for an observationally driven, spatially and temporally continuous ET estimate over the GRB. With the water resources of India being predicted to come under increasing pressure due to climate change [e.g., Mall et al., 2006; Kumar et al., 2005] it is imperative to focus on the assessment of long-term variability of ET at river basin scales. ET has long been recognized as a determining factor controlling the exchange of energy and mass between different spheres of the Earth system. Questions about how and what controls long-term variations of ET on regional and global scales remain major issues and at best uncertain [Trenberth et al., 2007]. While there is a general agreement that there is an upward trend in monsoon precipitation and freshwater discharge in major South and East Asian monsoon rivers [Gedney et al., 2006; Syed et al., 2010; Webster and Jian, 2011], a trend in ET, though, is still a debated issue [Ryu et al., 2008; Teuling et al., 2009]. Furthermore, this uncertainty in a potential intensification of the hydrologic cycle complicates determining the driving forces of ET [Ohamura and Wild, 2002; Huntington, 2006]. Moreover, efforts to compute realistic estimates of ET are challenged further by infrequent ‘‘ground-truth’’ observations and large variations in land surface characteristics [Bateni et al., 2013]. To address these problems, a variety of ET retrieval methods have been proposed for a wide range of spatial scales. These approaches include the direct utilization of observations: in situ [Jung et al., 2010; Baldocchi et al., 2001] and satellite [e.g., Irmak et al., 2012; Zhang et al., 2010; Tang et al., 2009; Leuning et al., 2008; Fisher et al., 2008; Cleugh et al., 2007; Mu et al., 2007; McCabe and Wood, 2006]. The capabilities of the NASA’s Gravity Recovery and Climate Experiment (GRACE) satellites [Tapley et al., 2004] in monitoring basin scale land water storage at monthly intervals represent major progress toward understanding the contributions of land water storage in basin and larger-scale hydrologic budgets [Rodell et al., 2004]. However, most of these ET estimates extend over a short time period and are limited by the accuracy and availability of river discharge data [Rodell et al., 2011]. Comprehensive reviews of the various in situ and remote sensing-based ET estimation methods, together with assessments of their accuracies are available [e.g., Wang and Dickinson, 2012; Li et al., 2009; Kalma et al., 2008; Verstraeten et al., 2008; Shuttleworth, 2007]. Other approaches are based on terrestrial water balance methods [Rodell et al., 2011; Karam and Bras, 2008] and the use of reanalysis data products [Yeh and Famiglietti, 2008; Lenters et al., 2000]. The water balance method is described in greater detail below. A number of model-based approaches have also been developed. For example, land surface model (LSM) simulations driven by observed forcing [Jimenez et al., 2011; Mueller et al., 2011; Teuling et al., 2009] and through the assimilation of data into land surface models (LSM) [e.g., Peters-Lidard et al., 2011; Jang et al., 2010; Crow and Kustas, 2005] or the statistical downscaling of grid point data obtained from low resolution models to higher resolution [Sellers et al., 2007]. Most recently, the conjunctive use of point-observations and semiempirical models [Wang et al., 2010] has been invoked. A majority of previous studies have concentrated on ET estimated as a residual of the column-integrated atmospheric moisture budget [e.g., Betts et al., 2005, 1999]. But some of the most recent studies have utilized terrestrial water budget computations [Rodell et al., 2004, 2011; Ramillien et al., 2006] and this is the method we will follow in the present study. Without a contemporary assessment of ET over the GRB, our current understanding is inadequate to isolate the factors controlling the variability of GRB ET [Jain, 2012]. Here we explore the issue of estimating ET and its relationship with major hydro-climatic factors by integrating many different data sets and, at the same time, taking advantage of their respective merits while taking into account their deficiencies. Here we utilize various land and atmospheric water balance computations to yield a robust, long-term estimate of ET for the GRB. All estimates presented here are based either on observations or observationally constrained model simulations. Our objective is to present a comparable and viable alternative to observed and assimilated data products such as those discussed above. Such a data set will help to further understand the

SYED ET AL.


2552


10.1002/2013WR013518

large-scale land-atmosphere interactions related to ET and also advance our understanding of the mean state and spatiotemporal variability of this key component of the water cycle. Specifically we aim to (i) develop and intercompare different process-based estimates of ET over the GRB, (ii) quantify and analyze long-term variations in ET, and (iii) assess the factors controlling long-term variations in ET.

2. Data 2.1. Precipitation Here two different merged precipitation products are used to obtain monthly estimates of precipitation averaged over the GRB. The spatial extent of the basin, over which the gridded data sets are averaged, is obtained from Total Runoff Integrating Pathways (TRIP) data set [Oki and Sud, 1998]. This averaging technique has been used for all the data sets used in this study. The first is the Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (version 2) [Adler et al., 2003]. This monthly precipitation data set is a merged analysis of global precipitation available from 1979 to the present, at 2.5 3 2.5 latitudelongitude grids. It combines the strengths of various sources of precipitation information, such as microwave-based estimates from the Special Sensor Microwave Imager (SSM/I), infrared (IR) rainfall estimates from geostationary and polar-orbiting satellites, with quality-controlled surface rain gauge data. Also included are precipitation estimates obtained from Television Infrared Observation satellite’s (TIROS) Operational Vertical Sounder (TOVS) and Outgoing Longwave Radiation (OLR) measurements. The GPCP data set is characterized by sophisticated quality control, low bias errors from the input information, and superior sampling of monthly products obtained from multiple satellites [Adler et al., 2003]. The other precipitation data, obtained from the Global Precipitation Climatology Center (GPCC), utilized rain gauge-based gridded monthly precipitation data set available from 1951 to the present [Rudolf et al., 1994]. This gauge-based analysis uses a robust empirical spatial interpolation technique to construct, a qualitycontrolled data set. Over 6000 rainfall gauges over the global land surface are used. In spite of the differences in the intricacies of data processing there is a high degree of agreement between the two products (see supporting information Figure S1). 2.2. River Discharge Discharge data used in this study are derived from water levels measured at the gauging station near Hardinge Bridge on the river Ganga (Figure 1a). The station is located in Bangladesh and is very close to the India-Bangladesh border. Upstream data from India are not available. Daily discharge data acquired from the gauge are aggregated to yield monthly estimates of river discharge from 1980 to 2006 (Figure 1b). 2.3. Terrestrial Water Storage Since its inception in March 2002, the GRACE satellite set has provided fundamental quantification of storage changes in various components of the hydrosphere. It has provided ample evidence that, on time scales considered here, variations in the Earth’s gravity field are controlled primarily by terrestrial water mass changes. These data, then act as a surrogate of an integrated estimate of storage variations in the land and atmospheric column [Wahr et al., 1998; Tapley et al., 2004]. Terrestrial water storage represents an aggregate of all forms of water stored above and underneath the land surface, including snow, surface waters, soil moisture, and groundwater. However, since GRACE can only sense variations in columnintegrated water mass, it cannot differentiate between changes in the individual components of the storage system. Recent applications of GRACE-derived water storage variations include validation and improvement of LSMs [Lo et al., 2010; Syed et al., 2008; Niu and Yang, 2006], estimation of important terrestrial hydrologic components such as ET [Rodell et al., 2011; Ramillien et al., 2006] and discharge [Syed et al., 2010, 2009]. In addition GRACE has furnished estimates of groundwater storage [Famiglietti et al., 2011; Rodell et al., 2009] and has monitored changes in the mass of glaciers and ice sheets [Chen et al., 2006]. GRACE data processing techniques involve truncation, smoothing, and removal of correlated errors in the data set and are discussed extensively in Tapley et al. [2004] and Wahr et al. [1998]. In this study, we use the most recent release (RL04) of GRACE gravity data produced by the University of Texas, Center for Space Research, which have been optimized for applications on land (available from http://grace.jpl.nasa.gov). The data set used here has been truncated at degree 60 and smoothed with a 300 km Gaussian averaging kernel. All the latest data improvement techniques are utilized as described by

SYED ET AL.


2553


10.1002/2013WR013518

(a)

(b) 180

River Discharge (mm/month)

160 140 120 100 80 60 40 20 0 198002 198202

198402 198602 198802 199002 199202 199402 199602 199802 200002 200202 200402 200602

Time (YearMonth) Figure 1. (a) Map of the study domain. Ganga river basin (hatched area) within the political boundaries of India and Bangladesh (marked with a red line). Location of the gauging station at Hardinge Bridge is represented with a black dot (b) Monthly time series of Ganga river discharge observed at Hardinge Bridge.

Swenson and Wahr [2006a]. We use 59 monthly fields covering the period from August 2002 to July 2007, with June 2003 data missing due to technical failure. Additionally, long-term estimates of terrestrial water storage are derived from simulations using the Land Dynamics Model (LaD) [Milly and Shmakin, 2002a]. For detailed discussion on model characteristics and evaluation of its performance, see Milly and Shamkin [2002b]. The LaD model provides monthly gridded time series of water and energy fluxes over the globe. Water is stored in the form of snow depth, soil moisture,

SYED ET AL.


2554


10.1002/2013WR013518

and groundwater storage within each nonglaciated cell. However, there are certain limitations in this model. The groundwater component represents only the relatively shallow and dynamic unconfined aquifers, that typically contribute to base flow [Milly and Shamkin, 2002a], in spite of the fact that there may also be large contributions from deeper aquifers. Also absent from the data set are stores of surface water. Although, not particularly relevant in the context of the current study area, contributions to terrestrial water storage can be significant in areas dominated by large lakes and extensive wetlands. Since none of the LSMs produce terrestrial water storage estimates, here we use monthly aggregates of soil water, groundwater, and snow amount produced by the LaD model. 2.4. Evapotranspiration For the purpose of validation and comparison, we utilize three different ancillary estimates of ET. These estimates are obtained from a LSM (LaD), a General Circulation Model (GCM), National Center for Environmental Protection (NCEP)/Department of Energy (DOE) reanalysis (NCEPR2) [Kanamitsu et al., 2002], and a satellitebased data set using Moderate Resolution Imaging Spectroradiometer (MODIS). Philosophically, a reanalysis attempts to produce a dynamically consistent global analysis of the state of the atmosphere using observations to constrain a dynamical GCM and to compromise between the properties of complete data coverage and accuracy. NCEPR2 uses an updated model, additional parameterizations and corrections in the assimilation of observational data [Kanamitsu et al., 2002]. As discussed earlier, satellite-based observations of ET hold considerable potential for large-scale continuous monitoring. Herein, we use data computed using a modified Penman-Monteith model [Mu et al., 2007]. The algorithm used remotely sensed information on land-surface and vegetation characteristics obtained from the MODIS satellite onboard NASA’s Aqua and Terra platforms. The strength of the methodology, developed by Mu et al. [2007], lies in the use of a combination of daily meteorological data (temperature, actual vapor pressure, and incoming solar radiation) and remotely sensed vegetation condition parameters (e.g., Leaf Area Index, LAI, and Normalized Difference Vegetation Index, NDVI). This builds upon the methodology developed by Cleugh et al. [2007] that provide 8 day composites at 1 km spatial resolution. Major modifications in the methodology established by Mu et al. [2007] include accounting for stomatal response to temperature and atmospheric vapor pressure deficit variations and an explicit consideration of soil evaporation.

3. Methods The terrestrial water budget can be formalized as: ET 5 P – R – @S=@t

(1)

where ET is evapotranspiration, @S/@t is the change in terrestrial water storage averaged over space, P is precipitation, and R is river basin discharge. Here @t is approximately 30 days, consistent with the temporal sampling of GRACE data. Despite its limitations, the solving of the terrestrial water balance, equation (1) is a relatively straightforward approach toward the estimation of ET at monthly intervals over large river basin scales. Equation (1), as utilized here, uses two different sets of @S/@t estimates. Those derived from GRACE are used for the purpose of comparison and validation, for the period of 2002–2007, and those derived from LaD simulations are used to compute long-term monthly estimates of ET. These long-term estimates are analyzed further for trends and to assess the hydroclimatic factors controlling the variations in ET. In order to produce a consistent estimate of @S/@t, storage components derived from LaD simulations are aggregated and processed in a manner similar to that of GRACE. Hence, the total storage from LaD model constitutes of total column soil moisture, snow water equivalent, and groundwater. Surface water storage and inland water bodies are not represented in the model simulations although it can be an important component of total terrestrial water storage in certain regions of the globe [Frappart et al., 2006]. Estimates of storage change from GRACE are best described as changes in storage averaged over a period of 30 days. These estimates require a careful consideration of the method to aggregate the hydrologic fluxes, P and R, on the right-hand side of (1), when used in combination with GRACE-derived or similarly assessed storage changes. Although alternative

SYED ET AL.


2555


10.1002/2013WR013518

schemes of aggregation have been published [Rodell et al., 2004; Swenson and Wahr, 2006b], here, the hydrologic fluxes in equation (1) are aggregated following the scheme described in Syed et al. [2005]. Comparisons of these methods of aggregation produced negligible difference in the water budget estimates at river basin scales [Syed et al., 2009]. Additionally, ET is estimated using the atmospheric water budget equation as follows: @W=@t 5 ET – P – div Q ð 1 Ps q dp W5 g PT Q5

1 g

(2) (3)

ð Ps qV dp

(4)

PT

P – ET 5 –@W=@t – div Q

(5)

ET 5 @W=@t 1 div Q 1 P

(6)

where W is column-integrated water vapor and div Q is the horizontal divergence of vapor flux vector integrated vertically over a number of pressure levels from the top of the atmosphere to the land surface, represented by PT and PS, respectively, q is specific humidity, g is the gravitational acceleration, and V is horizontal wind velocity. Here the divergence and precipitable water terms in equation (2) are obtained from the mass corrected, vertically integrated mass, moisture, heat, and energy budget products derived from NCEP/NCAR reanalysis data (NCEPR) [Kalnay et al., 1996] and from the European Centre for MediumRange Forecasts (ECMWF) operational forecast analysis (henceforth ECMWF) [Balsamo et al., 2009]. For details on the mass corrections applied to NCEPR, see Trenberth et al. [2002] and Trenberth [1997]. Notwithstanding their limitations [Cullather et al., 2000], reanalysis data sets combine the strengths of geophysical fluid-dynamical models of the atmosphere and high-quality observations to produce robust representations of water and energy fluxes that maintain continuity in time and space.

4. Results 4.1. Comparison With Ancillary Data We use several data sets to compute the variability of ET over the GRB. Here we compare simulated estimates of ET obtained from a GCM (NCEPR2-ET), a LSM (LaD-ET), and satellite-based observation (MODIS-ET) with spatially averaged estimates of ET computed as a residual of atmospheric water balance (AWB) (NCEPR-AWB and ECMWF-AWB) and terrestrial water balance (TWB) computations (GRC-TWB and LaD-TWB). The terrestrial water balance estimates are computed using P and R data with two different estimates of @S/ @t, one from GRACE and another from LaD and are referred to as GRC-TWB and LaD-TWB, respectively, in Figure 2. This comparison is restricted to the period common to all the different data sets (i.e., September 2002 to June 2007). The P data used here are computed as the monthly mean of those obtained from GPCP and GPCC. The mean values of P obtained from GPCP and GPCC over the study are 91.8 mm month21 and 94.4 mm month21, respectively. Figure 2 also reveals that ECMWF-AWB estimates are consistently higher in comparison with the rest of the estimates while MODIS-ET values are consistently lower. With the exception of ECMWF-AWB, most of the ET estimates for the GRB show a great deal of congruence and strong seasonality with a definite peak around July–August of every year. Similar observations of higher ET estimates from ECMWF interim analysis had been noted in several river basins around the globe [Vinukollu et al., 2011a]. In addition to being consistently greater than the rest of the estimates, the timing of the peaks in ECMWF-AWB estimates are distinctly different, particularly during the northern hemisphere winter when ET is generally low. The LaD-TWB time series has a somewhat larger mean, 67.2 mm month21, than the GRC-TWB time series at 57.9 mm month21 and that of MEAN-TWB (monthly average of LaD-TWB and GRC-TWB) at 61.6 mm month21. The gaps in the time series of TWB data sets, near the end of the study period, are due to the unavailability of river discharge during those months. In spite of being close to the time mean of NCEPR-AWB (67.2 mm month21), NCEPR2-ET (59.6 mm month21), and LaD-ET (58.4 mm month21), both these estimates show significant

SYED ET AL.


2556


10.1002/2013WR013518

200

NCEPR−AWB ECMWF−AWB NCEPR2−ET MODIS−ET LaD−ET GRC−TWB LaD−TWB MEAN-TWB

180

160

Evapotranspiration (mm/month)

140

120

100

80

60

40

20

0 200209

200309

200409

200509

200609

200706

Time(YearMonth)

Figure 2. Comparison of ET estimates over the Ganga river basin for the period of 2002–2007. Monthly time series of ET computed from AWB using NCEPR (NCEP-AWB; solid red line) and ECMWF (ECMWF-AWB; solid blue line) and from TWB computations using land water storage changes from GRC (GRC-TWB; black circle) and Lad (LaD-TWB; black triangle) and their monthly average (MEAN-TWB; solid black line). Also shown are direct estimates of ET obtained from NCEPR2 (NCEPR-ET; broken green line), MODIS (MODIS-ET; broken cyan line), and LaD (LaD-ET; broken magenta line).

differences with the ECMWF-ET and MODIS-ET estimates, the time mean of which are 95.2 mm month21 and 43.7 mm month21, respectively. Furthermore, the LaD-TWB time series correspond fairly well with that of GRC-TWB. Computations of Willmott’s refined index of agreement [Willmott et al., 2011] and correlation coefficient between the time series of LaD-TWB and GRC-TWB yield values of 0.56 and 0.78, respectively. Each of these data sets has its own set of limitations. For example, continuous changes over time in the observed states of operational analyses may lead to spurious apparent changes in the fields that affect the hydrological cycle in the assimilating models. Moreover, sometimes these models may not conserve quantities that should have been conserved based on physical principles due to the use of analysis increments (e.g., nudging) [Trenberth et al., 2011]. Thus, in spite of the major advances in modern reanalyses, substantial changes have been noted that affects the land surface water and energy fluxes [Dee and Uppala, 2009]. Some of the noted discrepancy can also be attributed to the scale differences in the individual variables, and merged products, used in the water budget computations [McCabe and Wood, 2006]. Mismatches in spatial and temporal scales at which each of the individual variables are generated can affect the magnitude of the ET estimate, obtained as a residual of the basin scale water budget. For example, GRACE data have nominal resolution of several thousand sq km at monthly intervals whereas MODIS-ET retrievals are made at 5–10 km pixels at subdaily resolutions. In particular, scaling issues have been noted earlier in the MODIS-ET values, because of its dependence on NDVI estimates and LAI models [Mu et al., 2007]. Vinukollu et al. [2011a] specifically noted that the algorithm proposed by Mu et al. [2007], also used in this study, underestimates ET at different test sites. This may be one of the reasons why the MODIS-ET estimates are consistently lower in the comparison shown in Figure 2. Improvement of retrieval algorithms, land surface model parameterizations, data assimilation, and bias correction schemes is a continually evolving active area of research. These changes can be episodic when there is a change in the satellite instrument [e.g., Schlosser and Houser, 2007]. Consequently, there is a dynamic range and additional uncertainty associated with the retrieval mechanisms of each of the variables. The inconsistencies noted in Figure 2 are perhaps best explained by the fundamental differences in the computation and acquisition of required input data sets used for each of the estimates and also because of the fact that these are only short-term averages over the period of 2002–2007. For greater details, please refer to studies by Ferguson et al. [2010] and Vinukollu et al. [2011b, 2012], which are exclusively dedicated to investigate the differences in ET estimates obtained from a variety of sources, as discussed earlier.

SYED ET AL.


2557


10.1002/2013WR013518

250 12−Month Avg ET−AVG AWB−E TWB−E


200

150

100

50

0 198002 198202 198402 198602 198802 199002 199202 199402 199602 199802 200002 200202 200402 200602

Time (YearMonth) Figure 3. Time series of ET over the Ganga river basin obtained, from terrestrial water balance (LaD-TWB), atmospheric water balance (NCEPR-AWB) and the average of the two (ET-AVG), for the period of 1980–2007. The shades of orange represent the uncertainty in the averaged monthly estimates. Also shown here as black solid line is average ET estimate (ET-AVG) smoothed with a 12 month moving average filter.

The primary motivation behind this comparison is to establish the efficacy of the water balance-based approach to estimate ET over the GRB. In turn, this tests the effectiveness of the use of LaD-based @S/@t and NCEPR-based W and div Q estimates to compute long-term variations of ET over the study area. While it is acknowledged that there are fundamental differences in the estimates of @S/@t obtained from LaD and GRACE, the comparison (Figure 2) demonstrates a high degree of compatibility. This confirms the credibility of @S/@t estimates obtained from LaD that can be used in long-term water balance computations. This is an important conclusion as it means that we now have, from the LaD, a reliable source of estimates of @S/@t for the GRB prior to the launch of GRACE allowing the determination of ET over longer periods. 4.2. Assessment of Long-Term ET Estimates Here two different estimates of ET are computed over the GRB using data from LaD-TWB and NCEPR-AWB, to extend the period to 1980–2007. Figure 3 shows time variations of LaD-TWB, NCEPR-AWB determinations, and their monthly average (ET-AVG). The seasonality in ET over the GRB, observed in both the estimates, shows a distinct peak during the months of July–August and a minimum in the January–February period. In addition to the strong phasal correspondence between the LaD-TWB and NCEPR-AWB estimates (R 5 0.86), there is also consistent difference in amplitudes particularly in their minimum values. The mean of LaD-TWB and NCEPR-AWB, for the entire study period, is estimated to be 63.2 6 24.8 mm month21 and 83.2 6 9.4 mm month21, respectively. Herein, the monthly average of these two estimates (ET-AVG) is considered as the best available long-term estimate of ET obtained from this study and the average of the two estimates for the entire study period is estimated to be 72.3 6 18.8 mm month21. In comparison, the most recent assessment [MoWR, 1999] suggested a much lower estimate of 20 mm month21 as reported by Jain [2012]. Figure 3 also displays the monthly values of uncertainty in the ET-AVG. These are computed by the statistical propagation of errors in each of the components of equations (1) and (6) at 95% confidence level [Syed et al., 2005]. The magnitude of errors in ET-AVG is quantified as the square root of the sum of squared errors in LaD-TWB and NCEPR-AWB. Errors in LaD-derived storage changes are estimated, in a way commonly used for GRACE [Wahr et al., 2006; Syed et al., 2009], using a least squares fit, consisting of annual,

SYED ET AL.


2558


10.1002/2013WR013518

semiannual, and linear terms, to monthly estimates of basin-averaged water storage. The root mean square error (RMS) of the residuals from the least squares fit is then Estimates of ET Mean Error dT used as a conservative estimate of the upper bound of error in water storage change estimates [Wahr et al., 2006]. LaD-TWB 63.2 24.8 0.82 NCEPR-AWB 83.2 9.4 0.82 For the computation of uncertainties in NCEPR-AWB, we ET-AVG 72.3 18.8 1.0 assumed a value of 10% for the relative error of div Q and a Also shown are the estimates of error and values @W/@t, primarily because no published estimates are availof the refined index of agreement (dT) with respect to able, and that for P and Q as 11% and 10%, respectively ET-AVG (average of LaD-TWB and NCEPR-AWB). [Rodell et al., 2011]. Larger errors in these terms would, of course, result in larger uncertainties in our ET estimates. Table 1 lists the monthly mean estimates of ET, average uncertainty and values of the refined index of agreement (dT) with respect to ET-AVG. Importantly, most of the monthly values of LaD-TWB and NCEPR-AWB fall within the range of error in ET-AVG. In order to analyze the trends and isolate possible factors controlling the interannual variability of ET the monthly estimates are smoothed with a centered 12 month moving average. This is done to reduce the dominant influence of seasonal variability over the long-term variations (Figure 3). Table 1. Monthly Mean Estimates of ET, Given in mm Month21, Over the Ganga River Basin for the Period of 1980–2007a

100

5

95

4

90

3

85

2

80

1

75

0

70

−1

65

−2

60

−3

55

−4

ENSO Indices


Figure 4 illustrates the interannual variations in ET (the average of LaD-TWB and NCEPR-AWB) obtained over the GRB between 1980 and 2006. Here the magnitude of trend is estimated as the slope of the least squares estimate of the best-fit line to the smoothed (12 month moving average) ET estimates. This linear filtering technique smooths out higher frequency variations and emphasize the interannual variations, which is one of the primary objectives of this study. Note that by using a 12 month moving average, we automatically introduce serial correlation to the original time series. While the use of moving average filter is imperative to highlight long-term trends, the significance of trends obtained from it is always overestimated. Given that the data set is smoothed with a 12 month moving average filter, the uncertainty analysis, in terms of p values and confidence intervals, of the estimated trends is misleading because they are based on the assumption that the fluctuations away from the trend line are independently distributed normal values. This is clearly not the case. Hence, the estimated trends are reported here without any uncertainty analysis.

50 −5 198008 198208 198408 198608 198808 199008 199208 199408 199608 199808 200008 200208 200408 200608

TIME(YearMonth)

Figure 4. Monthly time series of Ganga basin evapotranspiration after smoothing with a 12 month moving average filter. Least square estimate of best-fit trend line for the entire study period 198008–200612 is represented as a broken red line. The light green line shows a time series of the Ni~ no 3.4 ENSO index smoothed with a 12 month moving average filter.

SYED ET AL.


2559


10.1002/2013WR013518

The trend for the entire study period (198008–200612) is estimated to be 211 mm yr22. However, there are indications that the overall variability of ET may be influenced by ENSO events, as a result of which sharp changes in the ET variability are noted during certain portions (e.g., 199801–200206) of the study period (refer to section 5 for specific details). 4.3. ENSO-ET Relationship We compare the GRB ET time series, as derived above, for the period 1980–2006, with the commonly used ~o 3.4 index, which measures the sea-surface tempermeasure of El Nino-Southern Oscillation (ENSO), the Nin ature variability in the central-eastern equatorial Pacific Ocean (Figure 4). The aim is to determine if there is ~o planetary scale dynamical influence in the GRB ET, here filtered with a 12 month moving average. The Nin ~o and negative values La Nin ~a. 3.4 index varies with the phase of ENSO with positive values indicating El Nin Both the ENSO index and the ET are filtered using a 12 month moving average as was used in Figure 3. ~o 3.4 index. Results reveal no statistically significant relationship in the temporal variability of ET and Nin However, despite the lack of overall statistical confirmation, the largest break in the temporal variability of ~o 3.4 index around December 1997. One may surmise that ET over the ET occurs with the largest peak in Nin ~o event of 1997–1998, which is the strongest El Nin ~o event on GRB is significantly affected by the El Nin record [Dai et al., 2009]. Beyond 1997, ET values demonstrate a drastic decrease during the post 1997 La ~a event and continue till the development of another El Nin ~o event in the middle of 2002, which was a Nin major drought year in the GRB [Webster and Hoyos, 2004; Hoyos and Webster, 2007]. This decline in ET, sub~o event, is also consistent with the decline noted in global terrestrial ET after the sequent to the 1997 El Nin ~o event [Jung et al., 2010; Vinukollu et al., 2011b]. In addition, the decline in ET occurs at last major El Nin roughly the same time as the flattening of the global surface temperature record (http://www.cru.uea.ac.uk/ documents/421974/1295957/Info1sheet1%231.pdf/c612fc7e-babb-463c-b5e3–124ac76680c5). Since 1998, the average summer rainfall over India has also shown a steady decrease (http://www.tropmet.res.in/kolli/ mol/Monsoon/frameindex.html). ~o events It is important to reiterate that similar decreases in ET have not been observed during any of El Nin prior to 1997. Even though there seems to be a positive emerging trend in ET beyond 2002, nothing conclusive can be stated regarding the trend estimated for this latest period due to the short time span and weak statistical significance obtained. 4.4. Factors Controlling ET Variations The sensitivity of ET to soil moisture and precipitation is central to understand coupled land-atmosphere systems. A key eco-hydrological problem is uncoupling the effect of moisture supply and atmospheric demand on ET at local and regional scales. The response of ET to atmospheric demand, often represented by temperature or moisture supply, varies considerably with surface and subsurface characteristics, such as soil type, vegetation cover, topography, land use, and land cover. In order to decipher the relative importance between factors controlling the variability of ET over the GRB, we analyze independent estimates soil moisture and surface temperature averaged over the study area. Here near-surface atmospheric temperature is used as a surrogate of atmospheric demand since the meteorological factors controlling ET, such as incoming solar radiation and vapor pressure deficit, are strongly correlated with temperature. Similarly, changes in soil moisture content are directly linked to the amount of water accessible for plant uptake and thus limit the overall rate of evapotranspiration. Here we used the Climate Prediction Center (CPC) global soil moisture data set [Fan and van den Dool, 2004]. This data set is computed using a simple one-layer bucket water balance model driven by monthly inputs of gauge-based CPC global land precipitation and global reanalysis of 2 m air temperature. These modeled estimates have been validated with in situ observations and satellite-based estimates for near-real time hydrological applications. For monthly near-surface temperature time series, the average of monthly gridded data sets from the Center for Climate Research, University of Delaware (http://climate.geog.udel.edu/climate/) and the Climate Research Unit, University of East Anglia temperature data set, version 3.1 [Mitchell and Jones, 2005] is used. These data sets were primarily produced by the compilation of station data from several networks all over the globe, implementing a combination of spatial interpolation techniques that are embellished by climatological information and terrain characteristics. Figure 5 shows the temporal variations of ET, temperature and soil moisture estimates, averaged over the GRB, and smoothed with a 12 month moving average filter. Overall, results show that while ET estimates

SYED ET AL.


2560


10.1002/2013WR013518

25

95 90

24

85 80

23

75 70

22

65 60

21

55 50 20 198008 198208 198408 198608 198808 199008 199208 199408 199608 199808 200008 200208 200408 200608

Near-Surface Temperature(ºC)

100

TIME(YearMonth)

(b) 100

320

95 90

300

85 280

80 75

260

70 240

65 60

220

55 50

Soil Moisture (mm)



(a)

200

198008 198208 198408 198608 198808 199008 199208 199408 199608 199808 200008 200208 200408 200608

TIME(YearMonth) Figure 5. Coherence in temporal variability of ET compared to that of (a) near-surface temperature and (b) soil moisture. All the estimates are spatial averages over the domain of study and smoothed with a 12 month moving average filter.

are positively correlated with soil moisture (Pearson’s correlation coefficient R 5 0.35) it is negatively correlated with that of temperature (R 5 20.24). It is also important to note that, since 2002 the level of compatibility between ET and soil moisture decreased significantly. Perhaps, this in part, can be attributed to changes in land cover or vegetation properties within the basin [Douglas et al., 2006]. Based on the results of the analysis presented in Figure 5, we infer that ET changes over the GRB are dominantly controlled by variations in moisture supply, rather than atmospheric demand (as indicated by temperature). Further confirmations of this relationship are discussed in section 5. However, the dominant controls of ET can vary depending on the geographic region of interest. For example, Jung et al. [2010] presented results showing moisture-supply as the dominant controlling factor for global ET whereas Karam and Bras [2008] noted that energy limitation is the dominant control on ET over the Amazon River Basin.

5. Discussion and Conclusion Spatially averaged ET estimates are computed over the GRB, for the period 1980–2006, as a residual of basin-wide atmospheric and terrestrial water balance. Input data sets included domain averaged estimates of column-integrated atmospheric moisture flux divergence and column-integrated precipitable water tendency derived from NCEPR, precipitation data from the combination of GPCP and GPCC, land water storage changes from LaD model simulations and measurements of river discharge from the gauge located at Hardinge Bridge in Bangladesh. The computed estimates are initially validated over a short period of 2002– 2007 using ancillary estimates of ET obtained from computations of atmospheric water balance using NCEPR and ECMWF reanalysis data and terrestrial water balance using satellite observations of land water storage change from GRACE, a general circulation model (NCEPR2), land surface model (LaD), and those based on observations of MODIS satellite. Over the period of 2002–2007, ET estimates from atmospheric water balance computations (NCEPR-AWB) compared well with those estimated from terrestrial water balance (LaD-TWB), with a joint mean of 67.2

SYED ET AL.


2561


10.1002/2013WR013518

mm month21. Both NCEPR-AWB and LaD-TWB estimates agreed reasonably well and is placed in the middle of the range established by the other ancillary estimates. The basin-averaged ET, for the period of 1980– 2007, using the average of LaD-TWB and NCEPR-AWB estimates, is estimated to be 72.3 6 18.8 mm month21 in our estimation, the most robust representation of ET currently available for the study area. Monthly variations of ET peaked between July and August, the period coinciding with the maximum in soil water supply due to the predominant summer monsoon and reach a minimum around February. While the peaks in ET differed little in magnitude, the minima in LaD-TWB estimates are smaller than that of NCEPRAWB. Errors in these estimates are computed using the statistical propagation of errors in each of the water budget components. However, spatial averaging over the large area of the GRB (950,000 km2) reduced the uncertainty in our results. Analysis of trend in the ET estimate over GRB illustrates an overall decline at the rate of 11 mm yr22. Note that, prior to trend estimation, high frequency seasonal variations are removed from ET-AVG time series using a 12 month moving average filter, in order to emphasize the interannual variations. However, a detailed scrutiny of the temporal behavior of ET reveals possible existence of change points, beyond which substantial changes in ET variability are noted. In order to quantify this change in ET variability, trends in ET are also characterized for three subperiods of 198008–199712, 199801–200206, and 200207–200612. In detail, a gradual decline in ET for 198008–199712 (24.9 mm yr22) is followed by a significant decreasing trend for the period of 199801–200206 (255.8 mm yr22) and an increasing trend for the period of 200207– 200612 (9.6 mm yr22). However, some these trends should be interpreted with caution, given the short time period over which it is computed. We also performed a first-order attribution of the interannual variations in ET by assessing the contemporaneity of independent estimates of soil moisture and near-surface atmospheric temperature, as representatives of moisture-supply and atmospheric-demand, respectively. Strong temporal consistency patterns in the spatially averaged ET and soil-moisture estimates suggested that moisture supply is the dominant control of ET variability over the Ganga river basin. The correlation between soil moisture and ET are found to be even stronger when analyzed for each of the three subperiods mentioned above. Most importantly, the prominent decrease in ET, observed over the period of 199801–200206, is distinctly more coherent with soil moisture (R 5 0.73) than that of near-surface temperature (R 5 20.08). Correlations between ET and soil moisture for the other subperiods (198008–199712, R 5 0.51 and 200207–200612, R 5 0.61) are also significantly greater than those with near-surface temperature (198008–199712, R 5 0.23 and 200207–200612, R 5 0.18).

Acknowledgments Minor Research Project Grant of the Indian School of Mines sponsored this research. The MODIS data used in this work were provided by Qiaozhen Mu, University of Montana, Missoula, MT, USA. The Climate Dynamics Division of the National Science Foundation under grant NSF-AGS 0965610 provided funding support for this research supported P. J. Webster. We thank the Bangladesh Flood Forecasting and Warning Centre in Dhaka, Bangladesh for provided the daily discharge data. We also thank Francois Primeau, University of California-Irvine, for his comments and feedbacks on the statistical analysis. We are also very thankful to the anonymous reviewers who helped improve the content and presentation of this study. This support is gratefully acknowledged.

SYED ET AL.

The current study presents a detailed analysis on the magnitude and variability of ET over the GRB, which is particularly scarce in the scientific literature. Considering the spatial variability of ET under the influence of changing land cover scenarios can further extend this study. Besides, as the acquisition and precision of geophysical data continues to improve, the methodology presented here can be expected to produce improved results. Finally, our study complements several others emphasizing the importance of assessing and merging hydrologic information from several different platforms to compensate for their individual deficiencies.

References Adler, R. F., et al. (2003), The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979-present), J. Hydrometeorol., 4(6), 1147–1167. Baldocchi, D., et al. (2001), FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities, Bull. Am. Meteorol. Soc., 82(11), 2415–2434. Balsamo, G., A. Beljaars, K. Scipal, P. Viterbo, B. van den Hurk, M. Hirschi, and A. K. Betts (2009), A revised hydrology for the ECMWF model: Verification from field site to terrestrial water storage and impact in the integrated forecast system, J. Hydrometeorol., 10(3), 623–643, doi:10.1175/2008JHM1068.1. Bastiaanssen, W. G. M., E. J. M. Noordman, H. Pelgrum, G. Davids, B. P. Thoreson, and R. G. Allen (2005), SEBAL model with remotely sensed data to improve water resources management under actual field conditions, ASCE J. Irrig. Drain. E, 131, 85–93. Bateni, S. M., D. Entekhabi, and F. Castelli (2013), Mapping evaporation and estimation of surface control of evaporation using remotely sensed land surface temperature from a constellation of satellites, Water Resour. Res., 49, 1–19, doi:10.1002/wrcr.2007.1. Betts, A. K., J. H. Ball, and P. Viterbo (1999), Basin-scale surface water and energy budgets for the Mississippi from the ECMWF reanalysis, J. Geophys. Res., 104, 19,293–19,306. Betts, A. K., J. H. Ball, P. Viterbo, A. Dai, and J. Marengo (2005), Hydrometeorology of the Amazon in ERA40, J. Hydrometeorol., 6, 764–773. Chen, J. L., C. R. Wilson, and B. D. Tapley (2006), Satellite gravity measurements of confirm accelerated melting of Greenland ice sheet, Science, 313, 1958–1960.


2562


10.1002/2013WR013518

Cleugh, H. A., R. Leuning, Q. Mu, and S. W. Running (2007), Regional evaporation estimates from flux tower and MODIS satellite data, Remote Sens. Environ., 106(3), 285–304, doi:10.1016/j.rse.2006.07.007. Crow, W., and W. Kustas (2005), Utility of assimilating surface radiometric temperature observations for evaporative fraction and heat transfer coefficient retrieval, Boundary Layer Meteorol., 115(1), 105–130, doi:10.1007/s10546-004-2121-0. Cullather, R. I., D. H. Bromwich, and M. C. Serreze (2000), The atmospheric hydrologic cycle over the Arctic basin from reanalyses. Part I: Comparison with observations and previous studies, J. Clim., 13, 923–937. Dai, A., T. Qian, K. E. Trenberth, and J. D. Milliman (2009), Changes in continental freshwater discharge from 1948 to 2004, J. Clim., 22, 2773–2792, doi:10.1175/2008JCLI2592.1. Dee, D., and S. Uppala (2009), Variational bias correction of satellite radiance data in the ERA-Interim reanalysis, Q. J. R. Meteorol. Soc., 135, 1830–1841. Douglas, E. M., D. Niyogi, S. Frolking, J. B. Yeluripati, R. A. Pielke Sr., N. Niyogi, C. J. V€ or€ osmarty, and U. C. Mohanty (2006), Changes in moisture and energy fluxes due to agricultural land use and irrigation in the Indian monsoon belt, Geophys. Res. Lett., 33, L14403, doi: 10.1029/2006GL026550. Famiglietti, J. S., M. Lo, J. Bethune, K. J. Anderson, T. H. Syed, S. Swenson, C. R. de Linage, and M. Rodell (2011), Satellite measure recent rates of groundwater depletion in California’s Central Valley, Geophys. Res. Lett., 38, L03403, doi:10.1029/2010GL046442. Fan, Y., and H. van den Dool (2004), Climate prediction center global monthly soil moisture data set at 0.5 resolution for 1948 to present, J. Geophys. Res., 109, D10102, doi:10.1029/2003JD004345. Ferguson, C. R., J. Sheffield, E. F. Wood, and H. Gao (2010), Quantifying uncertainty in a remote sensing-based estimate of evapotranspiration over the continental United States, Int. J. Remote Sens., 31, 3821–3865, doi:10.1080/01431161.2010.483490. Fisher, J. B., K. P. Tu, and D. D. Baldocchi (2008), Global estimates of the land-atmosphere water flux based on monthly AVHRR and ISLSCPII data, validated at 16 FLUXNET sites, Remote Sens. Environ., 112(3), 901–919, doi:10.1016/j.rse.2007.06.025. Frappart, F., K. D. Minh, J. L’Hermitte, A. Cazenave, G. Ramillien, T. L. Toan, and N. Mognard (2006), Water volume change in the lower Mekong from satellite altimetry and imagery data, Geophys. J. Int., 67(2), 570–584. Gedney, N., P. M. Cox, R. A. Betts, O. Boucher, C. Huntingford, and P. A. Stott (2006), Detection of a direct carbon dioxide effect in continental river runoff records, Nature, 439, 835–838, doi:10.1038/nature04504. Goswami, B. N., V. Venugopal, D. Sengupta, M. S. Mahusoodanan, and P. K. Xavier (2006), Increasing trends of extreme rain events over India in a warming environment, Science, 314, 1442–1445. Hopson, T. M., and P. J. Webster (2010), A 1–10 day ensemble forecasting scheme for the major river basins of Bangladesh: Forecasting severe floods of 2003–2007, J. Hydrometeorol., 11, 618–641, doi:10.1175/2009JHM1006.1. Hoyos, C. D., and P. J. Webster (2007), The role of intraseasonal variability in the nature of Asian monsoon precipitation, J. Clim., 20(17), 4402–4424, doi:10.1175/JCLI4252.1. Huntington, T. G. (2006), Evidence for the intensification of the hydrologic cycle: Review and synthesis, J. Hydrol., 319, 83–95, doi:10.1016/ j.jhydrol.2005.07.003. Irmak, A., R. G. Allen, J. Kjaersgaard, J. Huntington, B. Kamble, R. Trezza, and I. Ratcliffe (2012), Operational remote sensing of ET and challenges, in Evapotranspiration—Remote Sensing and Modeling, edited by A. Irmak, InTech, Croatia, pp. 467–492, doi:10.5772/25174. Jain, S. K. (2012), India’s water balance and evapotranspiration, Curr. Sci., 102(7), 964–967. Jang, K., S. Kang, J. Kim, C. B. Lee, T. Kim, J. Kim, R. Hirata, and N. Saigusa (2010), Mapping evapotranspiration using MODIS and MM5 fourdimensional data assimilation, Remote Sens. Environ., 114(3), 657–673, doi:10.1016/j.rse.2009.11.010. Jim enez, C., et al. (2011), Global intercomparison of 12 land surface heat flux estimates, J. Geophys. Res., 116, D02102, doi:10.1029/ 2010JD014545. Jung, M., et al. (2010), Recent decline in the global land evapotranspiration trend due to limited moisture supply, Nature, 467, 951–954, doi:10.1038/nature09396. Kalma, J. D., T. R. McVicar, and M. F. McCabe (2008), Estimating land surface evaporation: A review of methods using remotely sensed surface temperature data, Surv. Geophys., 29(4–5), 421–469, doi:10.1007/s10712-008-9037-z. Kalnay, E., et al. (1996), The NCEP/NCAR 40-year reanalysis project, Bull. Am. Meteorol. Soc., 77(3), 437–470. Kanamitsu, M., W. Ebisuzaki, J. Woollen, S. K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter (2002), NCEP/DOE AMIP-II reanalysis (R-2), Bull. Am. Meteorol. Soc., 83, 1631–1643. Karam, H. N., and R. L. Bras (2008), Climatological basin-scale Amazonian evapotranspiration estimated through a water budget analysis, J. Hydrometeorol., 9, 1048–1060, doi:10.1175/2008JHM888.1. Kothyari, U. C., and V. P. Singh (1996), Rainfall and temperature trends in India, Hydrol. Processes, 10, 357–372. Kumar, R., R. D. Singh, and K. D. Sharma (2005), Water resources of India, Curr. Sci., 89, 794–811. Lenters, J. D., M. T. Coe, and J. A. Foley (2000), Surface water balance of the continental United States, 1963–1995: Regional evaluation of a terrestrial biosphere model and the NCEP/NCAR reanalysis, J. Geophys. Res., 105, 22,393–22,425, doi:10.1029/2000JD900277. Leuning, R., Y. Q. Zhang, R., A. Rajaud, H. A. Cleugh, and K. Tu (2008), A simple surface conductance model to estimate regional evaporation using modis leaf area index and the Penman-Monteith equation, Water Resour. Res., 44, W10419, doi:10.1029/2007WR006562. Li, Z. L., R. Tang, Z. M. Wan, Y. Y. Bi, C. H. Zhou, B. H. Tang, G. J. Yan, and X. Y. Zhang (2009), A review of current methodologies for regional evapotranspiration estimation from remotely sensed data, Sensors, 9(5), 3801–3853, doi:10.3390/s90503801. Lo, M. H., J. S. Famiglietti, P. J. F. Yeh, and T. H. Syed (2010), Improving parameter estimation and water table depth simulation in a land surface model using GRACE water storage and estimated base flow data, Water Resour. Res., 46, W05517, doi:10.1029/2009WR007855. Mall, R. K., A. Gupta, R. Singh, R. S. Singh, and L. S. Rathore (2006), Water resources and climate change: An Indian perspective, Curr. Sci., 90(12), 1610–1626. McCabe, M. F., and E. F. Wood (2006), Scale influences on the remote estimation of evapotranspiration using multiple satellite sensors, Remote Sens. Environ., 105(4), 271–285, doi:10.1016/j.rse.2006.07.006. Milly, P. C. D., and A. B. Shmakin (2002a), Global modeling of land water and energy balances. Part I: The Land Dynamics (LaD) model, J. Hydrometeorol., 3, 283–299. Milly, P. C. D., and A. B. Shmakin (2002b), Global modeling of land water and energy balances. Part II: Land-characteristic contributions to spatial variability, J. Hydrometeorol., 3, 301–310. Mitchell, P. D., and T. D. Jones (2005), An improved method of constructing a database of monthly climate observations and associated high-resolution grids, Int. J. Climatol., 25, 693–712. MoWR (1999), Integrated Water Resources Development—A Plan for Action: Report for the National Commission for Integrated Water Resources Development, vol. 1, pp. 515, Gov. of India, New Delhi.

SYED ET AL.


2563


10.1002/2013WR013518

Mu, Q., F. A. Heinsch, M. Zhao, and S. W. Running (2007), Development of a global evapotranspiration algorithm based on MODIS and global meteorology data, Remote Sens. Environ., 111(4), 519–536, doi:10.1016/j.rse.2007.04.015. Mueller, B., et al. (2011), Evaluation of global observations-based evapotranspiration datasets and IPCC AR4 simulation, Geophys. Res. Lett., 38, L06402, doi:10.1029/2010GL046230. Narasimhan, T. N. (2008), A note on India’s water budget and evapotranspiration, J. Earth Syst. Sci., 117(3), 237–240. National Ganga River Basin Authority Report (NGRBA) (2011), Environmental and Social Analysis, vol. 1, Minist. of Environ. and For., Gov. of India, New Delhi. Niu, G.-Y., and Z.-L. Yang (2006), Assessing a land surface model’s improvements with GRACE estimates, Geophys. Res. Lett., 33, L07401, doi: 10.1029/2005GL025555. Ohamura, A., and M. Wild (2002), Is the hydrologic cycle accelerating?, Science, 293, 1345–1346. Oki, T., and Y. C. Sud (1998), Design of Total Runoff Integrating Pathways (TRIP)—A global river channel network, Earth Interact., 2(1), 1–36. Peters-Lidard, C. D., S. V. Kumar, D. M. Mocko, and Y. Tian, (2011) Estimating evapotranspiration with land data assimilation systems, Hydrol. Processes, 25, 3979–3992, doi:10.1002/hyp.8387. Ramillien, G., F. Frappart, A. G€ untner, T. Ngo-Duc, A. Cazenave, and K. Laval (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. Rodell, M., J. S. Famiglietti, J. Chen, S. I. Seneviratne, P. Viterbo, S. Holl, and C. R. Wilson (2004), Basin scale estimates of evapotranspiration using GRACE and other observations, Geophys. Res. Lett., 31, L20504, doi:10.1029/2004GL020873. Rodell, M., I. Velicogna, and J. S. Famiglietti (2009), Satellite-based estimates of groundwater depletion in India, Nature, 460(7258), 999– 1002, doi:10.1038/nature08238. Rodell, M., E. B. McWilliams, J. S. Famiglietti, H. K. Beaudoing, and J. Nigro (2011), Estimating evapotranspiration using and observation based terrestrial water budget, Hydrol. Processes, 25, 4082–4092, doi:10.1002/hyp.8369. Rudolf, B., H. Hauschild, W. Rueth, and U. Schneider (1994), Terrestrial precipitation analysis: Operational method and required density of point measurements, in Global Precipitations and Climate Change, NATO ASI Ser., vol. 126, edited by M. Desbois and F. Desalmond, pp. 173–186, Springer, New York. Ryu, Y., D. D. Baldocchi, S. Y. Ma, and T. Hehn (2008), Interannual variability of evapotranspiration and energy exchange over an annual grassland in California, J. Geophys. Res., 13, D09104, doi:10.1029/2007JD009263. Schlosser, C. A., and P. R. Houser (2007), Assessing a satellite-era perspective of the global water cycle, J. Clim., 20, 1316–1338. Sellers, P. J., M. J. Fennessy, and R. E. Dickinson (2007), A numerical approach to calculating soil wetness and evapotranspiration over large grid areas, J. Geophys. Res., 112, D18106, doi:10.1029/2007JD008781. Shah, T. (2008), Taming the Anarchy: Groundwater Governance in South Asia, p. 324, Routledge, Washington, D. C. Shuttleworth, W. J. (2007), Putting the ‘‘vap’’ into evaporation, Hydrol. Earth Syst. Sci., 11(1), 210–244, doi:10.5194/hess-11-210-2007. Su, Z. (2002), The surface energy balance system (SEBS) for estimation of turbulent heat fluxes, Hydrol. Earth Syst. Sci., 6, 85–99. Swenson, S., and J. Wahr (2006a), Post-processing removal of correlated errors in GRACE data, Geophys. Res. Lett., 33, L08402, doi:10.1029/ 2005GL025285. Swenson, S., and J. Wahr (2006b), Estimating large-scale precipitation minus evaporation from GRACE satellite gravity measurements, J. Hydrometeorol., 7(2), 252–270. Syed, T. H., J. S. Famiglietti, J. Chen, M. Rodell, S. I. Seneviratne, P. Viterbo, and C. R. Wilson (2005), Total basin discharge for the Amazon and Mississippi river basins from GRACE and a land-atmosphere water balance, Geophys. Res. Lett., 32, L24404, doi:10.1029/ 2005GL024851. Syed, T. H., J. S. Famiglietti, M. Rodell, and J. Chen (2008), Analysis of terrestrial water storage changes from GRACE and GLDAS, Water Resour. Res., 44, W02433, doi:10.1029/2006WR005779. Syed, T. H., J. S. Famiglietti, and D. P. Chambers (2009), GRACE-based estimates of terrestrial freshwater discharge from basin to continental scales, J. Hydrometeorol., 10, 22–40, doi:10.1175/2008JHM993.1. Syed, T. H., J. S. Famiglietti, D. P. Chambers, J. K. Willis, and K. Hilburn (2010), Satellite-based global ocean mass balance reveals water cycle acceleration and increasing continental freshwater discharge, 1994–2006, Proc. Natl. Acad. Sci. U. S. A., 107(42), 17,916–17,921, doi: 10.1073/pnas.1003292107. Tang, Q., S. Peterson, R. H. Cuenca, Y. Hagimoto, and D. P. Lettenmaier (2009), Satellite-based near-real-time estimation of irrigated crop water consumption, J. Geophys. Res., 114, D05114, doi:10.1029/2008JD010854. Tapley, B. D., S. Bettadpur, J. C. Ries, P. F. Thompson, and M. M. Watkins (2004), GRACE measurements of mass variability in the Earth system, Science, 305, 503–505. Teuling, A. J., et al. (2009), A regional perspective on trends in continental evaporation, Geophys. Res. Lett., 36, L02404, doi:10.1029/ 2008GL036584. Thenkabail, P. S., M. Schull, and H. Turral (2005), Ganges and Indus river basin land use/land cover (LULC) and irrigated area mapping using continuous streams of MODIS data, Remote Sens. Environ., 95, 317–341. Trenberth, K. E. (1997), Using atmospheric budgets as a constraint on surface fluxes, J. Clim., 10, 2796–2809. Trenberth, K. E., D. P. Stepaniak, and J. M. Caron (2002), Accuracy of atmospheric energy budgets from analyses, J. Clim., 15, 3343–3360. Trenberth, K. E., et al. (2007), Observations: Surface and atmospheric climate change, in Climate Change 2007: The Scientific Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, edited by S. Solomon et al., pp. 235–336, Cambridge Univ. Press, Cambridge, U. K. Trenberth, K. E., J. T. Fasullo, and J. Mackaro (2011), Atmospheric moisture transports from ocean to land and global energy flows in reanalyses, J. Clim., 24, 4907–4924, doi:10.1175/2011JCLI4171.1. Verstraeten, W. W., F. Veroustraete, and J. Feyen (2008), Assessment of evapotranspiration and soil moisture content across different scales of observation, Sensors, 8(1), 70–117, doi:10.3390/s8010070. Vinukollu, R. K., E. F. Wood, C. R. Ferguson, and J. B. Fisher (2011a), Global estimates of evapotranspiration for climate studies using multisensor remote sensing data: Evaluation of three process-based approaches, Remote Sens. Environ., 115, 801–823. Vinukollu, R. K., R. Meynadier, J. Sheffield, and E. F. Wood (2011b), Multi-model, multi-sensor estimates of global evapotranspiration: Climatology, uncertainties and trends, Hydrol. Processes, 25, 3993–4010, doi:10.1002/hyp.8393. Vinukollu, R. K., J. Sheffield, E. F. Wood, M. G. Bosilovich, and D. Mocko (2012), Multimodel analysis of energy and water fluxes: Intercomparisons between operational analyses, a land surface model, and remote sensing, J. Hydrometeorol., 13, 3–26, doi:10.1175/ 2011JHM1372.1.

SYED ET AL.


2564


10.1002/2013WR013518

Wahr, J., M. Molenaar, and F. Bryan (1998), Time variability of the Earth’s gravity field: Hydrological and oceanic effects and their possible detection using GRACE, J. Geophys. Res., 103, 30,205–30,229. Wahr, J., S. Swenson, and I. Velicogna (2006), Accuracy of GRACE mass estimates, Geophys. Res. Lett., 33, L06401, doi:10.1029/ 2005GL025305. Wang, K., and R. E. Dickinson (2012), A review of global terrestrial evapotranspiration: Observation, modeling, climatology and climatic variability, Rev. Geophys., 50, RG2005, doi:10.1029/2011RG000373. Wang, K., R. E. Dickinson, M. Wild, and S. Liang (2010), Evidence for decadal variation in global terrestrial evapotranspiration between 1982 and 2002. 1: Model development, J. Geophys. Res., 115, D20112, doi:10.1029/2009JD013671. Webster, P. J., and C. Hoyos (2004), Prediction of monsoon rainfall and river discharge on 15–30 day time scales, Bull. Am. Meterol. Soc., 85(11), 1745–1765. Webster, P. J., and J. Jian (2011), Probability, uncertainty and prediction: A pathway towards the alleviation of poverty in the developing world, Philos. Trans. R. Soc. A, 369(1956), 1–30, doi:10.1098/rsta.2011.0160. Webster, P. J., V. O. Magana, T. N. Palmer, J. Shukla, R. A. Tomas, M. Yanai, and T. Yasunari (1998), Monsoons: Processes, predictability, and the prospects for prediction, J. Geophys. Res., 103, 14,451–14,510, doi:10.1029/97JC02719. Webster, P. J., et al. (2010), Extended-range probabilistic forecasts of Ganges and Brahmaputra floods in Bangladesh, Bull. Am. Meteorol. Soc., 91(11), 1493–1514, doi:10.1175/2010BAMS2911.1. Willmott, C. J., S. M. Robeson, and K. Matsuura (2011), A refined index of model performance, Int. J. Climatol., 33, 1053–1056, doi:10.1002/ joc.2419. Yeh, P. J. F., and J. S. Famiglietti (2008), Regional terrestrial water storage change and evapotranspiration from terrestrial and atmospheric water balance computations, J. Geophys. Res., 113, D09108, doi:10.1029/2007JD009045. Zhang, W., J. S. Kimball, R. R. Nemani, and S. W. Running (2010), A continuous satellite-derived global record of land surface evapotranspiration from 1983 to 2006, Water Resour. Res., 46, W09522, doi:10.1029/2009WR008800.

SYED ET AL.


2565