Content uploaded by Gang Zhao
Author content
All content in this area was uploaded by Gang Zhao on Nov 25, 2019
Content may be subject to copyright.
Towards Global Hydrological Drought Monitoring Using
Remotely Sensed Reservoir Surface Area
Gang Zhao
1
and Huilin Gao
1
1
Zachry Department of Civil Engineering, Texas A&M University, College Station, TX, USA
Abstract Hydrological drought quantification using surface water information has been lacking at a large
scale due to limited in situ data. We introduce a framework for monitoring hydrological droughts using a
global, long‐term, monthly, remotely sensed reservoir surface area dataset. At regional scales, a new index—
the reservoir area drought index (RADI)—is defined as the monthly normalized reservoir area time series.
RADI was validated using an in situ reservoir storage based index (average R
2
= 0.83). RADI not only helps to
characterize drought propagation from meteorological and agricultural droughts to hydrological droughts
but also fills the information gap between streamflow/runoff based and groundwater based drought indices.
The surface area dataset was further used to characterize the recovery rate (and to estimate the recovery
time) at the individual reservoir scale during droughts. This across‐scale drought monitoring framework can
help mitigate drought impacts and increase water use efficiency among multi‐reservoir systems.
Plain language summary Drought is a global phenomenon that can have substantial impacts on
regional economies and ecosystems. Quantification of droughts is of great importance for mitigating losses.
Leveraging an existing remotely sensed global reservoir area dataset, we introduce a new hydrological
drought index–the reservoir area drought index, RADI—that can serve as an indicator of surface water
availability at a regional scale. Nine drought‐prone regions were chosen to evaluate the performance of
RADI. Results suggest that RADI can contribute new information to hydrological drought monitoring and
can complement other indices to fully depict the propagation of different kinds of droughts. In addition,
reservoir area time series can also be used for quantifying the responses of individual reservoirs to droughts.
1. Introduction
Drought is a recurrent global phenomenon that hampers water‐food security and economic stability (Griggs
et al., 2013). Based on their characteristics and propagation mechanisms, droughts are generally classified
into four categories: meteorological, agricultural, hydrological, and economical droughts (Wang et al.,
2016; Wilhite & Glantz, 1985). Initiated by meteorological droughts, hydrological droughts manifest the
negative anomalies of freshwater in streams, lakes, reservoirs, and underground aquifers (Tallaksen &
Van Lanen, 2004; Van Loon, 2015). Directly affecting regional water availability and local ecosystem health,
hydrological droughts warrant better monitoring and understanding, especially in light of global environ-
mental changes (Bond et al., 2008; Humphries & Baldwin, 2003; Zhao et al., 2018).
A better understanding of hydrological droughts—with a focus towards early preparedness and impact miti-
gation—requires reliable quantification methods up‐front. A number of drought indices based on various
hydrological terms have been developed, such as the streamflow drought index (SDI; Vicente‐Serrano
et al., 2011), the standardized runoff index (SRI; Shukla & Wood, 2008), the surface water supply index
(SWSI; Shafer, 1982), and the groundwater resource index (GRI; Mendicino et al., 2008). Comprehensive
reviews of these indices—as well as those of meteorological and agricultural drought indices—can be found
in Heim (2002), Dai (2011), and Sheffield and Wood (2012). However, the applications of these hydrological
indices have often been limited by the availability of observation data and/or by model uncertainties (Van
Loon, 2015). For instance, the SDI needs intensive in situ streamflow data at a regional scale, while the
model‐based SRI is susceptible to large uncertainties associated with the inputs and structure of the
given models.
Satellite remote sensing has notably advanced the capability of global drought monitoring at a low cost
(AghaKouchak et al., 2015). New drought indices using satellite observations—such as the empirical
©2019. American Geophysical Union.
All Rights Reserved.
RESEARCH LETTER
10.1029/2019GL085345
Key Points:
•A new hydrological drought index
was developed using remotely
sensed reservoir surface area time
series
•The drought evaluation framework
permits drought quantification at
both regional and local scales
•The new index is informative about
reservoir water availability and
contributes to the evaluation of
drought propagation.
Supporting Information:
•Supporting Information S1
Correspondence to:
H. Gao,
hgao@civil.tamu.edu
Citation:
Zhao, G., & Gao, H. (2019). Towards
global hydrological drought monitoring
using remotely sensed reservoir surface
area. Geophysical Research Letters,46.
https://doi.org/10.1029/2019GL085345
Received 9 SEP 2019
Accepted 1 NOV 2019
ZHAO AND GAO 1
standardized soil moisture index (ESSMI; Carrão et al., 2016), the evapotranspiration drought index (ETDI;
Mu et al., 2013), and the drought index based on the Gravity Recovery and Climate Experiment (GRACE;
Thomas et al., 2014)—have been employed to assess the severity, duration, and spatial extent of hydrological
droughts. Although reservoirs play a critical role in water supply and drought mitigation (Huang & Chou,
2005; Lehner et al., 2011; Zhao & Gao, 2019), there have been no indices that are based on remotely sensed
reservoir status for global drought monitoring. On the one hand, elevation and storage variations of large
reservoirs have been monitored from space for decades (Birkett, 1994; Busker et al., 2019; Crétaux et al.,
2011; Gao et al., 2012). On the other hand, only several hundred reservoirs can be monitored via remote sen-
sing (due to radar/lidar altimetry data availability).
Recent developments of global scale surface water mapping techniques have offered promise for closing
such gaps (Klein et al., 2017; Pekel et al., 2016; Zhao & Gao, 2018). For instance, Pekel et al. (2016) created
a global surface water dataset (GSWD) using 3 million Landsat images from 1984 to 2015. Nonetheless, the
reservoir area time series resulting from GSWD are biased due to image contamination from clouds, cloud
shadows, terrain shadows, and scan line corrector failure. Building upon GSWD, Zhao and Gao (2018) devel-
oped a global reservoir surface area dataset (GRSAD) by applying an image enhancement algorithm to repair
the contaminated images. This resulted in monthly area time series for 6,817 global reservoirs. Because
reservoir surface area is correlated to storage (based on the rating curve), it can be used as a surrogate of sto-
rage to support drought quantification (Ogilvie et al., 2016; Zou et al., 2017).
Thus, the objective of this study is to develop a hydrological drought monitoring framework using remotely
sensed reservoir area data in support of drought mitigation practices on both regional and local scales. A new
reservoir area drought index (RADI) was designed to characterize droughts at a regional scale. By comparing
with in situ reservoir storage data, we demonstrated that RADI can effectively represent the reservoir storage
deficit. Next, RADI was calculated for nine drought‐prone regions, and was compared with two other com-
monly used drought indices. We show that RADI can contribute additional information towards compre-
hensive hydrological drought monitoring. Furthermore, GRSAD was adopted to characterize the
vulnerability of individual reservoirs to droughts (with the California region used as an example). We expect
that this new drought monitoring framework can complement other indices for fully depicting the propaga-
tion of different types of droughts.
2. Data and Methodology
In this study, GRSAD data from Zhao and Gao (2018) were used to facilitate drought quantification. GRSAD
contains monthly area data for 6,817 global reservoirs from March 1984 to October 2015. With the image
contaminations from multiple sources (e.g., cloud, shadow, and Landsat 7 scan line corrector failure) cor-
rected, high‐quality Landsat water classifications were achieved for accurate reservoir area estimations.
Thus, missing data and/or underestimations due to contaminations were minimized in the GRSAD time ser-
ies—which is essential for quantifying droughts accurately and consistently. The benefits of GRSAD based
drought analysis are twofold. First, drought severity at a regional scale can be quantified via a new index
which represents the relative area variations of a group of reservoirs (Section 2.1). Second, the vulnerability
and resilience of each individual reservoir to drought can be evaluated using the area percentage of the given
reservoir (section 2.2).
2.1. Drought Quantification at a Regional Scale
To quantify regional drought severity, we defined RADI using the Landsat based GRSAD. RADI was calcu-
lated after the following three steps:
1) Calculation of monthly time series of the total reservoir surface area within the region of interest.
2) Calculation of the cumulative distribution function of the regional monthly reservoir area from Step 1.
3) Conversion of a given area value from the cumulative distribution function to a RADI value that is nor-
mally distributed using the Probit Function (supporting information, Figure S1).
The RADI was calculated over nine drought‐prone regions (Figure 1). These regions were selected based on
the water stress level reported by the World Resources Institute (Gassert et al., 2013): (1) California, (2) the
Colorado River, (3) the Indus River, (4) Krishna‐Godavari, (5) the Murray‐Darling River, (6) Spain, (7) Texas
Gulf, (8) the Tigris‐Euphrates River, and (9) the Yellow River. While regions 2, 3, 5, 8, and 9 are defined after
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 2
major river basins, the other four regions are combinations of multiple small river basins. Inside of each
region, only the reservoirs with consistent records over the 32 years were selected. This was done to
exclude the impacts of new reservoir construction and of data sparsity before 1999 (when Landsat 7 was
launched). Most of these regions are located in the mid‐latitude area, which has a large population and is
vulnerable to droughts (Screen & Simmonds, 2014).
The RADI was validated by comparing it with in situ observed reservoir storage data. Similar to RADI, the
Reservoir Storage Drought Index (RSDI), which serves as the ground truth, is calculated as the normalized
monthly time series of total observed storage in a region. The RSDI was calculated for four regions that have
in situ observed storage data: California, Colorado, Texas Gulf, and the Murray‐Darling regions (Figure S2).
The data sources are summarized in the supporting information, Table S1. Furthermore, RADI was com-
pared to two commonly used drought indices. The first is the Palmer Drought Severity Index (PDSI; Alley,
1984). Even though it is a meteorological drought index, it also has been widely used to represent hydrolo-
gical conditions (Dai, 2011). Specifically, we used the self‐calibrating PDSI (sc‐PDSI), which replaces the
empirical constants in PDSI with dynamically calculated ones, such that the index values are consistent
and comparable among different locations (Wells et al., 2004). The global sc‐PDSI data used in this study
were collected from van der Schrier et al. (2013). The second index is the GRACE based Drought Severity
Index (GDSI; Thomas et al., 2014; Zhao et al., 2017). Specifically, we adopted the GDSI after Thomas et al.
(2014), which preserves the seasonality of the terrestrial water storage anomalies (TWSA). The TWSA values
were averaged from three representative Release‐05 datasets (by JPL, GFZ, and CSR). In order to be consis-
tent with RSDI and RADI, both the sc‐PDSI and GDSI time series were normalized after areal averaging for
each month.
2.2. Drought Quantification for Individual Reservoirs
At the local scale, we used the surface area time series to characterize the drought resilience of individual
reservoirs. First, the area at the top of the conservation pool of a given reservoir was selected as the reference
area (A
R
). If the conservation pool area information was unavailable, a percentile (e.g., 75%) from the 32‐year
time series was used for representing the normal status. Then, the relative area (in percentage) to the normal
Figure 1. The nine drought‐prone regions selected for this study, with the number of reservoirs and total storage value (in km
3
) for each region indicated in the
legend.
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 3
condition was calculated by dividing the current area value with the reference area. Such relative area values
are indicators of the fullness of a reservoir with regard to storage, as shown in Figure S5.
The GRSAD dataset can also be used for characterizing the capability of a reservoir to recover from drought.
Here, we define the average recovery rate by modifying the one provided by Thomas et al. (2014):
Rrc ¼P68 A1−A0;…;Ai−Ai−1;…;An−An−1
½
AR
·100%;(1)
where R
rc
is the average recovery rate (%); A
i
is the surface area in the ith month that satisfies the constraint
A
i
≤A
R
and A
i−1
≤A
R
;nis the number of months that satisfy the constraint; and P
68
indicates the 68th
percentile (1 standard deviation) of an array. Thus, when the current surface area of a given reservoir is A
and that reservoir is under drought conditions (A≤A
R
), the anticipated recovery time will be (A
R
−A)/
A
R
/R
rc
months (for this reservoir). It is worth noting that R
rc
is a statistical metric rather than a physical
property of the reservoirs. Thus, the calculated recovery time is a mathematical expectation, but the real
value can be affected by real‐time hydrology and reservoir operations.
3. Results
3.1. Validating RADI Using in situ Data
The performance of RADI was validated against RSDI in four regions that have in situ observed storage data
(Figure 2). The number of reservoirs with in situ data in each of these four regions is 137 (out of 186) for
California, 32 (81) for Colorado, 49 (118) for Texas Gulf, and 34 (65) for Murray‐Darling (Figure S2). To
be consistent with RSDI, RADI values were calculated only based on these reservoirs with in situ
observations available.
Figure 2 shows that RADI and RSDI share the same interannual variability and seasonal fluctuations, result-
ing in high coefficient of determination (R
2
) values and low root mean square error (RMSE) values. Both
indices have captured several notable regional droughts including the 2012–2016 California drought
(Griffin & Anchukaitis, 2014), the 2011 Texas drought (Long et al., 2013), and the Australia Millennium
drought (from 2001 to 2009; Pittock & Finlayson, 2011; van Dijk et al., 2013). In the Colorado River basin,
RADI has declined continuously since 2000, due to the depletion of several large reservoirs from both drought
and excessive water withdrawal. The lower R
2
value in the Texas Gulf region is caused by reservoir flood
Figure 2. Comparison of reservoir area drought index with Reservoir Storage Drought Index for the following regions: (a)
California, (b) Colorado, (c) Texas‐Gulf, and (d) Murray‐Darling.
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 4
control operations, which is a rapid process and usually cannot be captured by the GSWD/GRSAD monthly
time step.
3.2. Comparisons with sc‐PDSI and GDSI
The RADI time series were compared with sc‐PDSI and GDSI for all of the nine drought prone regions
(Figure 3). Although sc‐PDSI is able to capture multi‐year droughts for most regions, it does not show the sea-
sonal variability—and sometimes overestimates/underestimates the drought severity. For example, the mon-
soon induced seasonality cannot be reflected in the Krishna‐Godavari region by sc‐PDSI. This is because sc‐
PDSI is an accumulative index that hides seasonal variability. In the Colorado River basin, the sc‐PDSI
resumed to normal after 2014, while the reservoir storage continued declining according to RADI. This per-
sistent decrement of RADI is mainly caused by the depletion of several large reservoirs—especially Lake
Mead and Lake Powell, which account for 56% of the total reservoir area in the region. In the Tigris‐
Euphrates region, there is a clear disagreement between RADI and sc‐PDSI. The decrease of RADI is mainly
caused by the depletion of Razazza Lake. Sustained by floodwater diversion from the upstream Habbaniyah
Lake, Razazza Lake represents 20% of the region's reservoir area and 11% of its storage at capacity (Lehner
et al., 2011). From 2000 to 2014, its surface area decreased by 66.9 km
2
/year due to increasing irrigation water
use from upstream areas. Because this anthropogenic effect is independent of meteorological drought, the R
2
value between RADI and sc‐PDSI is only 0.09. This suggests that it is necessary to compare the meteorological
drought index (e.g., sc‐PDSI) with RADI for accurately interpreting the causes of reservoir water deficit.
Comparisons between RADI and GDSI can assist with decomposing the mixed signals of GDSI from surface
water, groundwater, and soil moisture, when reservoir storage change is the primary component of the sur-
face water anomaly; although GDSI generally correlates well with RADI both interannually and seasonally.
For instance, the R
2
values between RADI and GDSI in both the Indus and Yellow River basins are close to
zero, indicating a weak correlation. This is because groundwater changes dominate the GRACE TWSA
Figure 3. (a–i) The time series of RADI, sc‐PDSI, and GDSI for each of the nine regions. The two values represent the R
2
between RADI and sc‐PDSI/GDSI. Due to
the limited coverage of Landsat in the earlier period, the starting year of RADI varies by region.
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 5
signal in these two regions. In the Indus River Basin, the trends for GRACE TWSA, soil moisture, and RADI
are −3.34 mm/year (p< 0.05), 2.99 mm/year (p< 0.05), and 0.03 mm/year (p= 0.22), respectively. Because
there is no significant change in surface water (according to RADI) while the soil moisture has increased
(according to the Global Land Data Assimilation System), the decreased TWSA is most likely induced by
groundwater depletion. This is consistent with the findings by Qureshi et al. (2010) and Rodell et al.
(2018). Similarly, the decreasing TWSA in the Yellow River basin is also caused by groundwater depletion,
given that surface water has an increasing trend (0.10, p< 0.05) while soil moisture shows no significant
changes (−0.37 mm/year, p= 0.54). A more detailed analysis about the TWSA trend and its main drivers
for each of the nine regions can be found in Table S2 and in the supporting information, Text S3.
In addition, the California and Murray‐Darling regions were used as examples to compare the seasonality of
six indices: RSDI, RADI, ESSMI, SRI, sc‐PDSI, and GDSI (Figure S3 and Text S1). These comparisons show
that RADI can reproduce the seasonality of RSDI, and thus is a robust indicator of the water availability in
the reservoir systems. Results for two other regions (i.e., the Orange‐Senqu River basin and the Ohio River
basin) that are less vulnerable to droughts can be found in Figure S4 and in Text S2. Although RSDI is not
calculated due to limited data availability, RADI is found to be consistent with GDSI in both regions. Given
that these two indices are independently calculated, this suggests satisfactory performance of RADI in semi‐
arid to humid regions. It is worth noting, however, that high‐frequency and extensive cloud cover may limit
the quality of GRSAD in humid regions, and thus extra caution is needed to interpret RADI in such areas.
3.3. Drought Responses of Individual Reservoirs
GRSAD can also be used for quantifying the drought responses of individual reservoirs in order to support
optimizing the water supply of multi‐reservoir systems during droughts. Figure 4a shows the distributed area
percentages throughout the California region in October 2015, when RADI reached its lowest value (i.e.,
−2.79) during the 32‐year period. Validations of area percentage values for this month (Figure S5 and
Text S4), and of area values for all of the months (Figure S6), against areas derived from in situ storage values
suggest a satisfactory performance of GRSAD. For all of the 186 reservoirs, the averaged area percentage
value is 68%. There are 180 reservoirs with areas smaller than their normal areas (i.e., A
R
). In particular,
11 reservoirs in northern California and 10 reservoirs in Central Valley suffered significant water losses (area
percentage ≤40%).
The recovery rate (Figure 4b) shows how long it takes, on average, for a reservoir to resume to normal (i.e.,
A
R
). Using two large California reservoirs as examples—Shasta Lake (largest) and New Melones Lake
(fourth largest)—their recovery rates are 3.0% and 0.5% (per month), respectively. These values indicate a
large disparity in drought recovery times. For example, if each of the two reservoirs had a surface area of
70% of its long‐term median (A
R
), it would take New Melones Lake 60 months to fully recover (on average),
while Shasta Lake would recover in 10 months. Along with the area percentage information, the water man-
agers (e.g., river basin authorities) can possibly use this spatially distributed information to optimize water
usage during severe drought events.
4. Discussion and Conclusions
In this study, Landsat based reservoir surface area information was used for monitoring hydrological
drought both at a regional scale and for individual reservoirs.
At the regional scale, RADI contributes to drought monitoring by helping characterize the drought propaga-
tion from meteorological and agricultural droughts to hydrological droughts. Hydrological droughts can be
further divided into specific types such as streamflow droughts, reservoir deficit droughts, and groundwater
droughts (Mishra and Singh, 2010). RADI is specifically designed to indicate the degree of water shortage in
reservoir systems, which fills the information gap between streamflow/runoff‐based drought indices and
groundwater based drought indices. Our analysis shows that RADI can better represent the seasonal varia-
tion of reservoir storage than other indices can (Figures 3 and S3). In addition, RADI can provide important
information to help decompose the signal of GDSI, which is based on terrestrial water storage with mixed
signals from soil moisture, surface water, and groundwater (Table S2 and Text S3).
Having the capability to scrutinize individual reservoirs during droughts, the scalable drought quantification
framework formulated in this study also contributes to targeted drought mitigation practices. By evaluating
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 6
the relative area percentage for each individual reservoir during a drought event, the reservoirs with
significantly below‐normal area values can be identified (Figure 4). This can facilitate joint operations of
multiple reservoirs during droughts. For instance, water reallocation among reservoirs based on near real‐
time area monitoring can increase the local water supply resilience (Huang & Chou, 2008). In addition, in
transboundary river basins, information about upstream reservoir conditions under drought can
accommodate downstream water management (Zhang et al., 2014). With the recent launch of multiple
Earth observation satellites (e.g., Landsat 8, Sentinel‐2, and the Joint Polar Satellite System), the accuracy,
temporal resolution, and time‐lag of GRSAD can be further improved. This can help generate more
accurate and up‐to‐date RADI values for reservoir monitoring purposes.
The RADI is affected by multiple factors such as regional climate variability, hydrological conditions, and
anthropogenic water management practices. These factors often act in different “directions”(Brekke
et al., 2009; Trabucco et al., 2008). For example, water managers might tend to store more water to satisfy
the increasing water demand. This is the case for the Yellow River basin, where RADI had a significant
increasing trend (0.1 per year, p< 0.05) while the streamflow was declining from 1999 to 2014 (Huang
et al., 2015; Zhao et al., 2014). By comparing RADI with meteorological drought indices (e.g., sc‐PDSI)
and land surface drought indices (e.g., SRI and ESSMI), the contributions of different factors can be dis-
cerned. For instance, in the Colorado River Basin (Figure 3b), sc‐PDSI started to increase in recent years
(beginning in 2013) while RADI continued to decrease. This indicates that other factors (e.g., water use
and evaporation) were playing important roles in the reservoir depletion (Barnett & Pierce, 2009).
While RADI generally performs well at regional scales, individual reservoir inspection using GRSAD
requires extra caution. Uncertainties of GRSAD are inherited from multiple sources. First, the mixed pixels
on reservoir edges can affect the accuracy of GSWD, especially for small reservoirs. For instance, Ogilvie
et al. (2018) reported significant biases in GSWD for small lakes (~5 ha) that are partially covered by vegeta-
tion. For medium and large size reservoirs, GSWD generally has good performance (Busker et al., 2019;
Huang et al., 2018). Second, the enhancement algorithm for generating GRSAD can introduce uncertainties
both due to its use of a histogram thresholding method (Zhao & Gao, 2018) and because of the biases of reser-
voir masks (Lehner et al., 2011). Third, the monthly time step is not sufficient to resolve the fast flood control
operations, which results in possible biases for regional water availability when the reservoirs in a given sub‐
region are changing quickly (Ogilvie et al., 2015).
Despite the promising applicability of RADI, there are several limitations worth noting: First, RADI is based
on reservoir surface area instead of actual storage. This might limit the use of RADI with regard to represent-
ing the absolute mass change of surface water. This limitation can be addressed by converting surface area
Figure 4. (a) Area percentage values for the 186 individual reservoirs during October 2015 in California. (b) Average
recovery rate calculated for the period from March 1984 to October 2015.
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 7
time series values into storage time series values using techniques such as those described by Li et al. (2019)
and Yigzaw et al. (2018). Second, the size of the region that RADI can be implemented depends on the num-
ber of reservoirs. For instance, the number of reservoirs used for calculating RADI values varies from 5 to 38
for the 8 HUC4 regions shown in Figure S7 (and explained in Text S5). However, for regions that have a very
low number of reservoirs (such as central Australia or western China), analysis of individual reservoir areas
(rather than using RADI) might be more informative.
In summary, a new drought index, RADI—which is based on a satellite water mapping dataset—has demon-
strated skills in monitoring regional hydrological droughts. Comparison between RADI and sc‐PDSI—and
between RADI and GDSI—not only shows the robustness of the new index in representing reservoir storage
but also suggests that it can provide additional information about the drought propagation chain.
Meanwhile, the reservoir surface area dataset can also be employed to characterize the drought responses
of individual reservoirs. These methods can benefit regional and local scale water resources management
practices and can improve our understanding about the partitioning of surface water and groundwater.
References
AghaKouchak, A., Farahmand, A., Melton, F., Teixeira, J., Anderson, M., Wardlow, B. D., & Hain, C. (2015). Remote sensing of drought:
Progress, challenges and opportunities. Reviews of Geophysics,53, 452–480. https://doi.org/10.1002/2014RG000456
Alley, W. M. (1984). The Palmer drought severity index: Limitations and assumptions. Journal of Climate and Applied Meteorology,23(7),
1100–1109. https://doi.org/10.1175/1520‐0450(1984)023<1100:TPDSIL>2.0.CO;2
Barnett, T. P., & Pierce, D. W. (2009). Sustainable water deliveries from the Colorado River in a changing climate. Proceedings of the
National Academy of Sciences,106(18), 7334–7338. https://doi.org/10.1073/pnas.0812762106
Birkett, C. (1994). Radar altimetry: A new concept in monitoring lake level changes, Eos. Transactions of the American Geophysical Union,
75(24), 273–275. https://doi.org/10.1029/94EO00944
Bond, N. R., Lake, P., & Arthington, A. H. (2008). The impacts of drought on freshwater ecosystems: An Australian perspective.
Hydrobiologia,600(1), 3–16. https://doi.org/10.1007/s10750‐008‐9326‐z
Brekke, L. D., Maurer, E. P., Anderson, J. D., Dettinger, M. D., Townsley, E. S., Harrison, A., & Pruitt, T. (2009). Assessing reservoir
operations risk under climate change. Water Resources Research,45, W04411. https://doi.org/10.1029/2008WR006941
Busker, T., Roo, A. d., Gelati, E., Schwatke, C., Adamovic, M., Bisselink, B., et al. (2019). A global lake and reservoir volume analysis
using a surface water dataset and satellite altimetry. Hydrology and Earth System Sciences,23(2), 669–690. https://doi.org/10.5194/hess‐
23‐669‐2019
Carrão, H., Russo, S., Sepulcre‐Canto, G., & Barbosa, P. (2016). An empirical standardized soil moisture index for agricultural drought
assessment from remotely sensed data. International Journal of Applied Earth Observation and Geoinformation,48,74–84. https://doi.
org/10.1016/j.jag.2015.06.011
Crétaux, J.‐F., Jelinski, W., Calmant, S., Kouraev, A., Vuglinski, V., Bergé‐Nguyen, M., et al. (2011). SO LS: A lake database to monitor in the
Near Real Time water level and storage variations from remote sensing data. Advances in Space Research,47(9), 1497–1507. https://doi.
org/10.1016/j.asr.2011.01.004
Dai, A. (2011). Drought under global warming: A review. Wiley Interdisciplinary Reviews: Climate Change,2(1), 45–65.
Gao, H., Birkett, C., & Lettenmaier, D. P. (2012). Global monitoring of large reservoir storage from satellite remote sensing. Water Resources
Research,48, W09504. https://doi.org/10.1029/2012WR012063
Gassert, F., Reig, P., Luo, T., & Maddocks, A. (2013). A weighted aggregation of spatially distinct hy drological indicators, Working paper.
Washington, DC: World Resources Institute, available online at. wri.org/publication/aqueduct‐country‐river‐basin‐rankings
Griffin, D., & Anchukaitis, K. J. (2014). How unusual is the 2012–2014 California drought? Geophysical Research Letters,41, 9017–9023.
https://doi.org/10.1002/2014GL062433
Griggs, D., Stafford‐Smith, M., Gaffney, O., Rockström, J., Öhman, M. C., Shyamsundar, P., et al. (2013). Policy: Sustainable development
goals for people and planet. Nature,495(7441), 305–307. https://doi.org/10.1038/495305a
Heim, R. R. Jr. (2002). A review of twentieth‐century drought indices used in the United States. Bulletin of the American Meteoro logical
Society,83(8), 1149–1166. https://doi.org/10.1175/1520‐0477‐83.8.1149
Huang, C., Chen, Y., Zhang, S., & Wu, J. (2018). Detecting, extracting, and monitoring surface water from space using optical sensors: A
review. Reviews of Geophysics,56, 333–360. https://doi.org/10.1029/2018RG000598
Huang, S., Huang, Q., Chang, J., Zhu, Y., Leng, G., & Xing, L. (2015). Drought structure based on a nonparametric multivariate standar-
dized drought index across the Yellow River basin, China. Journal of Hydrology,530, 127–136. https://doi.org/10.1016/j.
jhydrol.2015.09.042
Huang, W.‐C., & Chou, C.‐C. (2005). Drought early warning system in reservoir operation: Theory and practice. Water Resources Research,
41, W11406. https://doi.org/10.1029/2004WR003830
Huang, W.‐C., & Chou, C.‐C. (2008). Risk‐based drought early warning system in reservoir operation. Advances in Water Resources,31(4),
649–660. https://doi.org/10.1016/j.advwatres.2007.12.004
Humphries, P., & Baldwin, D. S. (2003). Drought and aquatic ecosystems: An introduction. Freshwater Biology,48(7), 1141–1146. https://
doi.org/10.1046/j.1365‐2427.2003.01092.x
Klein, I., Gessner, U., Dietz, A. J., & Kuenzer, C. (2017). Global WaterPack—A 250 m resolution dataset revealing the daily dynamics of
global inland water bodies. Remote Sensing of Environment,198, 345–362. https://doi.org/10.1016/j.rse.2017.06.045
Lehner, B., Liermann, C. R., Revenga, C., Vörösmarty, C., Fekete, B., Crouzet, P., et al. (2011). High‐resolution mapping of the world's
reservoirs and dams for sustainable river‐flow management. Frontiers in Ecology and the Environment,9(9), 494–502. https://doi.org/
10.1890/100125
Li, Y., Gao, H., Jasinski, M. F., Zhang, S., & Stoll, J. D. (2019). Deriving high‐resolution reservoir bathymetry from ICESat‐2 prototype
photon‐counting Lidar and Landsat imagery. IEEE Transactions on Geoscience and Remote Sensing,57(10), 7883–7893.
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 8
Acknowledgments
This research was supported by the
NASA Science of Terra, Aqua, and
Suomi NPP (TASNPP) Program
(80NSSC18K0939) provided to Texas
A&M University. This research has
benefitted from the usage of the Google
Earth Engine platform (https://
earthengine.google.com) and the Texas
A&M Supercomputing Facility (http://
sc.tamu.edu). A script to help calculate
different indices can be found at:
https://github.tamu.edu/hgao/
drought‐indices. We appreciate the help
from Miss Cheryl Holmes that
improved the presentation of the
manuscript. We would also like to
thank Dr. Valeriy Ivanov (the Editor)
and two anonymous reviewers for the
detailed suggestions toward helping to
improve the manuscript.
Long, D., Scanlon, B. R., Longuevergne, L., Sun, A. Y., Fernando, D. N., & Save, H. (2013). GRACE satellite monitoring of large depletion in
water storage in response to the 2011 drought in Texas. Geophysical Research Letters,40, 3395–3401. https://doi.org/10.1002/grl.50655
Mendicino, G., Senatore, A., & Versace, P. (2008). A Groundwater Resource Index (GRI) for drought monitoring and forecasting in a
mediterranean climate. Journal of Hydrology,357(3‐4), 282–302. https://doi.org/10.1016/j.jhydrol.2008.05.005
Mishra, A. K., & Singh, V. P. (2010). A review of drought concepts. Journal of Hydrology,391(1), 202–216. https://doi.org/10.1016/j.
jhydrol.2010.07.012
Mu, Q., Zhao, M., Kimball, J. S., McDowell, N. G., & Running, S. W. (2013). A remotely sensed global terrestrial drought severity index.
Bulletin of the American Meteorological Society,94(1), 83–98. https://doi.org/10.1175/BAMS‐D‐11‐00213.1
Ogilvie, A., Belaud, G., Delenne, C., Bailly, J.‐S., Bader, J.‐C., Oleksiak, A., et al. (2015). Decadal monitoring of the Niger Inner Delta flood
dynamics using MODIS optical data. Journal of Hydrology,523, 368–383. https://doi.org/10.1016/j.jhydrol.2015.01.036
Ogilvie, A., Belaud, G., Massuel, S., Mulligan, M., Le Goulven, P., & Calvez, R. (2016). Assessing floods and droughts in ungauged small
reservoirs with long‐term Landsat imagery. Geosciences,6(4), 42. https://doi.org/10.3390/geosciences6040042
Ogilvie, A., Belaud, G., Massuel, S., Mulligan, M., Le Goulven, P., & Calvez, R. (2018). Surface water monitoring in small water bodies:
Potential and limits of multi‐sensor Landsat time series. Hydrology and Earth System Sciences,22(8), 4349–4380. https://doi.org/10.5194/
hess‐22‐4349‐2018
Pekel, J.‐F., Cottam, A., Gorelick, N., & Belward, A. S. (2016). High‐resolution mapping of global surface water and its long‐term changes.
Nature,540(7633), 418–422. https://doi.org/10.1038/nature20584
Pittock, J., & Finlayson, C. M. (2011). Australia's Murray–Darling Basin: Freshwater ecosystem conservation options in an era of climate
change. Marine and Freshwater Research,62(3), 232–243. https://doi.org/10.1071/MF09319
Qureshi, A. S., McCornick, P. G., Sarwar, A., & Sharma, B. R. (2010). Challenges and prospects of sustainable groundwater management in
the Indus Basin, Pakistan. Water Resources Management,24(8), 1551–1569. https://doi.org/10.1007/s11269‐009‐9513‐3
Rodell, M., Famiglietti, J., Wiese, D., Reager, J., Beaudoing, H., Landerer, F., & Lo, M.‐H. (2018). Emerging trends in global freshwater
availability. Nature,557(7707), 651–659. https://doi.org/10.1038/s41586‐018‐0123‐1
Screen, J. A., & Simmonds, I. (2014). Amplified mid‐latitude planetary waves favour particular regional weather extremes. Nature Climate
Change,4(8), 704–709. https://doi.org/10.1038/nclimate2271
Shafer, B. (1982), Developemnet of a surface water supply index (SWSI) to assess the severity of drought conditions in snowpack runoff
areas, paper presented at Proceedings of the 50th Annual Western Snow Conference, Colorado State University, Fort Collins, 1982.
Sheffield, J., & Wood, E. F. (2012), Drought: Past problems and future scenarios, Routledge.
Shukla, S., & Wood, A. W. (2008). Use of a standardized runoff index for characterizing hydrologic drought. Geophysical Research Letters,
35, L02405. https://doi.org/10.1029/2007GL032487
Tallaksen, L. M., & Van Lanen, H. A. (2004). Hydrological drought: Processes and estimation methods for streamflow and groundwater.
Amsterdam, Netherlands: Elsevier.
Thomas, A. C., Reager, J. T., Famiglietti, J. S., & Rodell, M. (2014). A GRACE‐based water storage deficit approach for hydrological drought
characterization. Geophysical Research Letters,41,1537–1545. https://doi.org/10.1002/2014GL059323
Trabucco, A., Zomer, R. J., Bossio, D. A., van Straaten, O., & Verchot, L. V. (2008). Climate change mitigation through
afforestation/reforestation: A global analysis of hydrologic impacts with four case studies, Agriculture. Ecosystems & Environment,
126(1‐2), 81–97. https://doi.org/10.1016/j.agee.2008.01.015
van der Schrier, G., Barichivich, J., Briffa, K., & Jones, P. (2013). A scPDSI‐based global data set of dry and wet spells for 1901‐2009. Journal
of Geophysical Research: Atmospheres,118, 4025–4048. https://doi.org/10.1002/jgrd.50355
van Dijk, A. I., Beck, H. E., Crosbie, R. S., de Jeu, R. A., Liu, Y. Y., Podger, G. M., et al. (2013). The Millennium Drought in southeast
Australia (2001–2009): Natural and human causes and implications for water resources, ecosystems, economy, and society. Water
Resources Research,49, 1040–1057. https://doi.org/10.1002/wrcr.20123
Van Loon, A. F. (2015). Hydrological drought explained. Wiley Interdisciplinary Reviews Water,2(4), 359–392. https://doi.org/10.1002/
wat2.1085
Vicente‐Serrano, S. M., López‐Moreno, J. I., Beguería, S., Lorenzo‐Lacruz, J., Azorin‐Molina, C., & Morán‐Tejeda, E. (2011). Accurate
computation of a streamflow drought index. Journal of Hydrologic Engineering,17(2), 318–332.
Wang, W., Ertsen, M. W., Svoboda, M. D., & Hafeez, M. (2016). Propagation of drought: From meteorological drought to agricultural and
hydrological drought. Advances in Meteorology,2016,1–5. https://doi.org/10.1155/2016/6547209
Wells, N., Goddard, S., & Hayes, M. J. (2004). A self‐calibrating Palmer drought severity index. Journal of Climate,17(12), 2335–2351.
https://doi.org/10.1175/1520‐0442(2004)017<2335:ASPDSI>2.0.CO;2
Wilhite, D. A., & Glantz, M. H. (1985). Understanding: the drought phenomenon: The role of definitions. Water International,10(3),
111–120. https://doi.org/10.1080/02508068508686328
Yigzaw, W., Li, H. Y., Demissie, Y., Hejazi, M. I., Leung, L. R., Voisin, N., & Payn, R. (2018). A new global storage‐area‐depth dataset for
modeling reservoirs in land surface and earth system models. Water Resources Research,54,10–372. https://doi.org/10.1029/
2017WR022040
Zhang, S., Gao, H., & Naz, B. S. (2014). Monitoring reservoir storage in South Asia from multisatellite remote sensing. Water Resources
Research,50, 8927–8943. https://doi.org/10.1002/2014WR015829
Zhao, G., & Gao, H. (2018). Automatic correction of contaminated images for assessment of reservoir surface area dynamics. Geophysical
Research Letters,45,6092–6099. https://doi.org/10.1029/2018GL078343
Zhao, G., & Gao, H. (2019). Estimating reservoir evaporation losses for the United States: Fusing remote sensing and modeling approaches.
Remote Sensing of Environment,226, 109–124. https://doi.org/10.1016/j.rse.2019.03.015
Zhao, G., Gao, H., Kao, S.‐C., Voisin, N., & Naz, B. S. (2018). A modeling framework for evaluating the drought resilience of a surface water
supply system under non‐stationarity. Journal of Hydrology,563,22–32. https://doi.org/10.1016/j.jhydrol.2018.05.037
Zhao, G., Tian, P., Mu, X., Jiao, J., Wang, F., & Gao, P. (2014). Quantifying the impact of climate variability and human activities on
streamflow in the middle reaches of the Yellow River basin, China. Journal of Hydrology,519, 387–398. https://doi.org/10.1016/j.
jhydrol.2014.07.014
Zhao, M., Velicogna, I., & Kimball, J. S. (2017). A global gridded dataset of grace drought severity index for 2002–14: Comparison with PDSI
and SPEI and a case study of the Australia millennium drought. Journal of Hydrometeorology,18(8), 2117–2129. https://doi.or g/10.1175/
JHM‐D‐16‐0182.1
Zou, Z., Dong, J., Menarguez, M. A., Xiao, X., Qin, Y., Doughty, R. B., et al. (2017). Continued decrease of open surface water body area in
Oklahoma during 1984–2015. Science of the Total Environment,595, 451–460. https://doi.org/10.1016/j.scitotenv.2017.03.259
10.1029/2019GL085345
Geophysical Research Letters
ZHAO AND GAO 9
A preview of this full-text is provided by Wiley.
Content available from Geophysical Research Letters
This content is subject to copyright. Terms and conditions apply.
