Streamflow Composition and the Contradicting Impacts of Anthropogenic Activities and Climatic Change on Streamflow in the Amu Darya Basin, Central Asia

Mei Hou aState Key Laboratory of Tibetan Plateau Earth System, Environment and Resources, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China
bUniversity of Chinese Academy of Sciences, Beijing, China

Search for other papers by Mei Hou in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0003-2486-907X
,
Lan Cuo aState Key Laboratory of Tibetan Plateau Earth System, Environment and Resources, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China
bUniversity of Chinese Academy of Sciences, Beijing, China
cCenter for Excellence in Tibetan Plateau Earth Sciences, Chinese Academy of Sciences, Beijing, China

Search for other papers by Lan Cuo in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0001-9551-3041
,
Amirkhamza Murodov aState Key Laboratory of Tibetan Plateau Earth System, Environment and Resources, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China
bUniversity of Chinese Academy of Sciences, Beijing, China

Search for other papers by Amirkhamza Murodov in
Current site
Google Scholar
PubMed
Close
,
Jin Ding dPublic Meteorological Service Center, China Meteorological Administration, Beijing, China

Search for other papers by Jin Ding in
Current site
Google Scholar
PubMed
Close
,
Yi Luo bUniversity of Chinese Academy of Sciences, Beijing, China
eKey Laboratory of Ecosystem Network Observation and Modeling, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China

Search for other papers by Yi Luo in
Current site
Google Scholar
PubMed
Close
,
Tie Liu bUniversity of Chinese Academy of Sciences, Beijing, China
fState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, China

Search for other papers by Tie Liu in
Current site
Google Scholar
PubMed
Close
, and
Xi Chen bUniversity of Chinese Academy of Sciences, Beijing, China
fState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, China
gResearch Center for Ecology and Environment of Central Asia, Urumqi, China

Search for other papers by Xi Chen in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

Transboundary rivers are often the cause of water-related international disputes. One example is the Amu Darya River, with a catchment area of 470 000 km2, which passes through five countries and provides water resources for 89 million people. Intensified human activities and climate change in this region have altered hydrological processes and led to water-related conflicts and ecosystem degradation. Understanding streamflow composition and quantifying the change impacts on streamflow in the Amu Darya basin (ADB) are imperative to water resources management. Here, a degree-day glacier-melt scheme coupled offline with the Variable Infiltration Capacity hydrological model (VIC-glacier), forced by daily precipitation, maximum and minimum air temperature, and wind speed, is used to examine streamflow composition and changes during 1953–2019. Results show large differences in streamflow composition among the tributaries. There is a decrease in the snowmelt component (−260.8 m3 s−1) and rainfall component (−30.1 m3 s−1) at Kerki but an increase in the glacier melt component (160.0 m3 s−1) during drought years. In contrast, there is an increase in the snowmelt component (378.6 m3 s−1) and rainfall component (12.0 m3 s−1) but a decrease in the glacier melt component (−201.8 m3 s−1) during wet years. Using the VIC-glacier and climate elasticity approach, impacts of human activities and climate change on streamflow at Kerki and Kiziljar during 1956–2015 are quantified. Both methods agree and show a dominant role played by human activities in streamflow reduction, with contributions ranging 103.2%–122.1%; however, the contribution of climate change ranges from −22.1% to −3.2%.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Lan Cuo, lancuo@itpcas.ac.cn

Abstract

Transboundary rivers are often the cause of water-related international disputes. One example is the Amu Darya River, with a catchment area of 470 000 km2, which passes through five countries and provides water resources for 89 million people. Intensified human activities and climate change in this region have altered hydrological processes and led to water-related conflicts and ecosystem degradation. Understanding streamflow composition and quantifying the change impacts on streamflow in the Amu Darya basin (ADB) are imperative to water resources management. Here, a degree-day glacier-melt scheme coupled offline with the Variable Infiltration Capacity hydrological model (VIC-glacier), forced by daily precipitation, maximum and minimum air temperature, and wind speed, is used to examine streamflow composition and changes during 1953–2019. Results show large differences in streamflow composition among the tributaries. There is a decrease in the snowmelt component (−260.8 m3 s−1) and rainfall component (−30.1 m3 s−1) at Kerki but an increase in the glacier melt component (160.0 m3 s−1) during drought years. In contrast, there is an increase in the snowmelt component (378.6 m3 s−1) and rainfall component (12.0 m3 s−1) but a decrease in the glacier melt component (−201.8 m3 s−1) during wet years. Using the VIC-glacier and climate elasticity approach, impacts of human activities and climate change on streamflow at Kerki and Kiziljar during 1956–2015 are quantified. Both methods agree and show a dominant role played by human activities in streamflow reduction, with contributions ranging 103.2%–122.1%; however, the contribution of climate change ranges from −22.1% to −3.2%.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Lan Cuo, lancuo@itpcas.ac.cn

1. Introduction

Transboundary rivers are often the focus of international conflicts and disputes, and because of that the Helsinki Rules were drafted and adopted by the International Law Association in 1966 to resolve international water-right-related disputes. The Amu Darya River is located in central Asia and flows through Kyrgyzstan, Tajikistan, Afghanistan, Uzbekistan, and Turkmenistan. It has a long history of conflicts as the upstream countries where the river originates want to use more water for electricity generation and farming while the downstream countries where little streamflow is generated require more water for their growing agricultural sectors and rising populations (Zhiltsov et al. 2018). On the other hand, climate change could either further exacerbate or ease the existing conflicts depending on the outcome of the change impact on streamflow. Such social and environmental factors that contribute to the water conflicts are also manifested in other regions of the world (Gleick et al. 2020). Quantifying the contributions of various factors that influence streamflow and developing adaptation strategies accordingly in transboundary rivers could alleviate water-related conflicts.

Streamflow originating in cold regions with snow and glacier coverage responds to changes in temperature and precipitation in complicated ways (Barnett et al. 2005; Chen et al. 2017; Lutz et al. 2014; Sorg et al. 2012; Viviroli and Weingartner 2004; Yao et al. 2012). Many studies suggested that warming-induced shrinkage of glacier and snow areas exerted positive impacts on streamflow in densely glacierized river basins but those with less or no glacier coverage exhibited complicated change impacts on streamflow (Chen et al. 2017; Unger-Shayesteh et al. 2013). Specific to the Amu Darya basin (ADB), Khan and Holko (2009) found a positive but statistically insignificant trend in natural streamflow during 1961–2005 at the Kerki station, to which the majority of the upper Amu Darya basin (UADB) drains. The same authors suggested that streamflow increases there could be the result of increasing glacier meltwater. However, Wang et al. (2016) indicated that it was changes in precipitation that resulted in streamflow reduction (increase) in the Pyanj (Zeravshan) subbasin. Seasonally, there was a slight increase in summer (July–September) streamflow in the Vakhsh and Pyanj subbasins during 1966–80 compared to 1959–65 (Konovalov and Shchetinnicov 1994). On the other hand, Olsson et al. (2010) documented positive trends for January–April and negative trends for August–September in the Zeravshan subbasin during 1923–2007.

These different responses of streamflow to climate change could be attributed to the differences in streamflow composition, as streamflow changes vary depending on the dominant streamflow components (Chen et al. 2017; Lutz et al. 2014; Sorg et al. 2012; Unger-Shayesteh et al. 2013; Zhao et al. 2019). Clearly, the quantification of streamflow composition and its changes can shed light on how freshwater has changed and may evolve in the future. A few studies analyzed streamflow composition in the region but only focused on either the outlet of the entire ADB (Armstrong et al. 2019; Immerzeel et al. 2012; Liu et al. 2020b) or individual upland subbasins in the northern and eastern parts (Hagg et al. 2013; Konovalov 2011). To date, streamflow composition has not been extensively analyzed across all subbasins in the region, especially in its southern reaches. Previous estimations of streamflow composition were also often based on simplified assumptions and models or models without subjecting to rigorous validations (Armstrong et al. 2019; Konovalov 2011; Liu et al. 2020b). This study aims to fill the gap by conducting a comparative analysis of streamflow composition among all upland tributaries and the main stem within the ADB.

Apart from climate change, human activities such as water extraction and diversion for agricultural irrigation, land use/cover change, soil conservation, and hydropower engineering can all alter streamflow (Cuo 2016). For the ADB, the impact of human activities on streamflow is mainly through irrigation and the irrigation industry is a major economic sector in central Asia (Micklin 2007, 2016; Su et al. 2021). Due to a dry climate coupled with high water-consuming crops such as cotton and wheat, more than 90% of the crops in the ADB rely on irrigation (FAO 2013; Wang et al. 2020). The irrigated area in the ADB has increased from 2.28 × 106 ha in 1960 to 4.81 × 106 ha in 2010 (Chen et al. 2018), and more than 60 canals have been constructed for irrigation purposes (Adenbaev et al. 2015). The benefits of water consumption for irrigated agriculture increase economic productivity and stability (Micklin 2007, 2016). However, the massive extraction of water from the Amu Darya River caused streamflow decreases at the Tuyamuyun and Kiziljar stations at an annual rate of 7.05 × 109 m3 (10a)−1 and 6.58 × 109 m3 (10a)−1 in 1956–2015, respectively (Jilili 2019). Due to the annual streamflow reduction at these stations, water delivery to the Aral Sea decreased from 1970 to 2015 at an average annual rate of 2.74 × 109 m3 (10a)−1 (Wang et al. 2020).

Those studies that focused on the ADB (e.g., Micklin 2007, 2016; Stulina and Eshchanov 2013; Wang et al. 2016) examined the changes in streamflow in the region and their relation to climate change but failed to include the effects of human activities and to further deconvolve the impacts of climate change and human activities on streamflow. The process- and statistical-based approaches are widely used to quantify the impacts of climate change and human activities on streamflow and their relative importance (Cuo et al. 2009, 2013; Chang et al. 2016; Dey and Mishra 2017; Jiang et al. 2011; Li et al. 2007; Liu et al. 2020a; Zheng et al. 2009). The process-based approach uses physically based hydrological models to quantify the contribution of climate change to streamflow by reconstructing the natural streamflow time series and then differentiating with the observed streamflow time series (Chang et al. 2016; Dey and Mishra 2017; Jiang et al. 2011; Liu et al. 2020a). The statistical approach is applied to estimate the sensitivity of streamflow to precipitation and potential evapotranspiration based on their relationship derived from the historical hydrometeorological data at relatively coarse time steps, e.g., annually (Schaake 1990; Li et al. 2007; Zheng et al. 2009). To our knowledge, Hu et al. (2021) is the only study that applied the climate elasticity method coupled with the Budyko framework to quantitatively analyze streamflow change in the ADB. However, the Budyko hypothesis does not consider the impacts of groundwater storage, soil moisture and other cryospheric components in a basin, making it physically deficient for application in most catchments (Istanbulluoglu et al. 2012; Sankarasubramanian et al. 2001; Zheng et al. 2009). To overcome this shortcoming, a nonparametric estimator of climate elasticity of streamflow was developed (Sankarasubramanian et al. 2001; Zheng et al. 2009). On the other hand, the hydrological model, despite its advantages over the statistical approach, has so far not been considered for quantifying the impacts of climate change and human activities in the ADB.

To fill these knowledge gaps, this study is carried out 1) to estimate streamflow composition among all tributaries in the ADB and analyze how streamflow composition has changed during 1953–2019 and under arid and wet climatic conditions and 2) to combine the hydrological model and climate elasticity method to quantify the impacts of climate change and human activities on streamflow and their relative importance. Applying two approaches with different frameworks would help us understand the uncertainty of the results and improve the credibility of the results.

2. Methodology

a. Study region

The ADB is located within 58°–76°E and 34°–43°N, with elevations ranging from 72 to 7495 m above sea level (Fig. 1). It has an area of 470 000 km2, of which 2.89% (13 600 km2; RGI Consortium 2014) is covered by glacier ice. From the source of the Pyanj River, which is also the main stem, to the Aral Sea where it ends, the Amu Darya River runs 2540 km long (Agal’tseva et al. 2011). The ADB has a continental climate controlled primarily by the westerlies. Its average annual air temperature is 11.2°C, and average annual precipitation is 351.0 mm. Around 62% of annual precipitation falls in December–April. About 77%–80% (8%–13%) of annual streamflow occurs in April–September (December–February) (Agal’tseva et al. 2011). Its land cover types transition from snow and glacier in the east high-mountain region to grassland, cropland, and open shrubland in the middle and west regions. Soil types consist mainly of loam, clay loam, and sand. The UADB above the Kerki station has an area of 288 898 km2, or about 60% of the entire area of the ADB, of which 4.39% (12 681 km2; RGI Consortium 2014) is covered by glacier ice. The long-term average annual discharge in the UADB observed at the Kerki station is 72.6 km3 (Wang et al. 2016).

Fig. 1.
Fig. 1.

The Amu Darya basin (ADB). Black triangles denote the 1) Khorog, 2) Sardem, 3) Shidzh, 4) Hirmanjo, 5) Khirmanjo, 6) Garm, 7) Kishl Pete, 8) Most Dupuli, 9) Kerki, and 10) Chatly hydrological stations whose observations are used for model calibration and validation. Kerki and Kiziljar (red triangle) are where the deconvolving of the anthropogenic activities and climate change impacts on streamflow is carried out. Glaciers are indicated by royal blue shadings, and stream channels are represented by blue lines. Black and red lines in the inset map outline the ADB and UADB, respectively.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

The Amu Darya River contributes two-thirds of the total freshwater to the Aral Sea basin (Agal’tseva et al. 2011) and provides water resources for 89 million people living in five countries in central Asia (FAO 2021). In the second half of the 1950s and the beginning of the 1960s, hydrologically related intensive economic development began in the ADB. To date, more than 60 canals have been built in the middle and low reaches and take water from the Amu Darya River for irrigation, among which the largest are Karakum, Karshi, Аmu-Bukhara, Tashsakin, and Pakhtaarnin (Adenbaev et al. 2015). As a result, water delivery to the Aral Sea from the Amu Darya River displayed a substantial decreasing trend (Wang et al. 2020). The shrinking of the Aral Sea caused the collapse of the fishing industry, compromised drinking water, exacerbated soil salinization, and facilitated the occurrence of dust storms (Micklin 2007, 2016; Yang et al. 2020). Dried sections of the lake also became sources of fine dust particles that have harmful impacts on human health (Micklin 2016). As a transboundary river, the Amu Darya River frequently experienced conflicts about water use at national and local scales.

b. Hydrological model

The Variable Infiltration Capacity (VIC) model (version 4.1.2.g), used in this study, is a physically based macroscale hydrological model developed to solve water and energy balances at a grid cell (Liang et al. 1994, 1996). The VIC model assumes that surface streamflow is generated by the infiltration excess flow and saturation overland flow with soil depth down to 2–3 m below the surface. Baseflow responds both linearly and nonlinearly to soil moisture changes depending on bottom layer soil moisture conditions and flow rates, which is realized using the empirical Arno model (Todini 1996). Evapotranspiration and energy fluxes are calculated at each cell at each time step. After computations of water and energy balances at each cell are finished, surface streamflow and baseflow from all grid cells within the basin boundary are routed to the basin outlet through a routing model forming basin discharge (Lohmann et al. 1998). The routing model assumes that the water can leave the grid cell in one of the eight neighboring grid cells (Lohmann et al. 1998).

One of the appealing features of the VIC model is its capacity to simulate cold region hydrology by incorporating a two-layer energy-balance model to represent snow accumulation and ablation on the ground (Andreadis et al. 2009; Cherkauer and Lettenmaier 1999). In this study, 20 elevation bands, obtained from a finer-resolution (0.0025° × 0.0025°) terrain dataset, are implemented to account for snow accumulation and ablation in the mountainous areas, and the elevation bands vary among the grid cells. Mean temperature of the model grid cell is adjusted to each elevation band using the lapse rate of 6°C km−1, derived from observed temperature in the basin. Precipitation fraction in each elevation band in each grid cell is determined by the ratio of the mean elevation of the elevation band to the mean elevation of the cell, the area fraction of the elevation band within the cell, and the cell’s precipitation. In this way, the orographic effect of elevation on precipitation is roughly accounted for. The total of precipitation fraction in a cell equals to the cell’s precipitation.

However, the glacier-melt process is not considered in the standard version of the VIC model. To overcome this, the 500 m × 500 m grid-based temperature-index model developed by Hock (1999) (see Text S1 in the online supplemental material for details) is incorporated here for calculating glacier melt. This glacier model does not simulate glacier flow.

The total runoff in each grid cell after taking into account glacier melt is calculated as follows (Zhang et al. 2013):
Rtotal=F×M+(1F)×RVIC,
where Rtotal is the total runoff (mm) in the grid cell, F is the glacier area fraction and M is the glacier meltwater (mm) in the cell, RVIC is the runoff (mm) in the cell calculated by the VIC, and 1 − F is the fraction of the ice-free area in the cell. The glacier model is run first and then the results are combined with the VIC simulated runoff in the cells where glaciers exist. This is how the VIC model and the glacier model are coupled, and this coupled system is referred to as the VIC-glacier hereafter. The combined runoff and baseflow are then routed toward the basin outlet. For the VIC-glacier, the spatial resolution is 0.25° × 0.25° and the time step is 1 day.

Like most physically based hydrologic models, the VIC-glacier has various parameters that must be calibrated. We chose and calibrated nine parameters in this study (Table S5). Observed monthly streamflow, satellite-derived snow cover, and glacier area change ratios from previously works are used to calibrate and evaluate the VIC-glacier. Four evaluation statistics, the coefficient of determination (R2), Kling–Gupta efficiency (KGE; Kling et al. 2012), Nash–Sutcliffe efficiency (NSE; Nash and Sutcliffe 1970), and relative error (PBIAS; Gupta et al. 1999), are examined (see Text S2).

c. Data

Daily precipitation, maximum air temperature, minimum air temperature and wind speed for 1951–2019 at 0.25° × 0.25° spatial resolution are required to drive the VIC-glacier model. Multiple gridded precipitation, air temperature and wind speed products with continuous spatial coverage and various time spans are the original sources of the model forcing. A comprehensive evaluation of air temperature and precipitation products (Table S1) against available observations (Fig. S1, Tables S2 and S3) is conducted to select the best quality products. As the observed wind speed is not available, the wind speed product from the Global Land Data Assimilation System (GLDAS) is used as is without evaluation. The analysis results indicate a better performance of the Asian Precipitation–Highly Resolved Observational Data Integration Toward Evaluation of Water Resources (APHRODITE) and Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS) products for precipitation, and the Climate Prediction Center (CPC) and Princeton Global Forcing (PGF) products for air temperature (Table S4 and Fig. S2). These four products are further homogenized to produce consistent spatiotemporal change patterns (see Text S3 for the procedure).

The input data required by the VIC-glacier include topography (e.g., elevation, slope, and aspect), soil texture and depth, and land cover (e.g., vegetation type and glacier distribution). Elevation data (∼500 m) are obtained from the Shuttle Radar Topographic Mission (SRTM; https://srtm.csi.cgiar.org/srtmdata/) and are used to create a digital river network. Soil texture data at 0.0083° × 0.0083° are collected from the Harmonized World Soil Database, version 1.2 (http://webarchive.iiasa.ac.at/Research/LUC/External-World-soil-database/HTML/HWSD_Data.html?sb=4). The MODIS MCD12C1 land cover product at 0.05° × 0.05° is downloaded from https://lpdaac.usgs.gov/products/mcd12c1v006/. The initial glacier distribution dataset is from the Randolph Glacier Inventory (RGI) 4.0 (RGI Consortium 2014). The Northern Hemisphere EASE-Grid 2.0 Weekly Snow Cover and Sea Ice Extent, version 4, data are used to calculate snow cover fraction (SCF; Brodzik and Armstrong 2013).

Observed monthly streamflow data at Khorog, Sardem, Shidzh, Garm, Khirmanjo, Hirmanjo, Kishl Pete, and Most Dupuli are acquired from the Tajikistan Hydrometeorologic Agency. Observations at Kerki and Chatly are publicly accessible from the Global Runoff Data Center (https://www.bafg.de/GRDC/EN/Home/homepage_node.html). Streamflow data at these 10 stations are used to calibrate and validate the VIC-glacier simulated streamflow (see Table 2 for calibration and validation periods) and for streamflow composition analysis. Due to differences in time periods of available records at the 10 stations, calibration and validation periods are also different among the stations, which actually lends credence to model performance because of evaluations under different time periods. Most of the stations without human interference are located in the UADB and are distributed across a broad topographic range and subbasin size. The spatial distributions and detailed information of the stations are shown in Fig. 1 and Table 1. To obtain a comprehensive understanding of natural streamflow composition in the entire ADB, an additional seven tributaries in the southern ADB are chosen and investigated. The details of the outlets of the seven tributaries, which are not used in the model calibration and validation, are shown in Table S6. Annual streamflow at Kerki and Kiziljar for the period of 1956–2015 are used in the deconvolving of the anthropogenic activities and climate change impacts on streamflow in the ADB.

Table 1

Information of the 10 hydrological stations whose records are used to calibrate and validate VIC-glacier.

Table 1
Table 2

Statistics of the observed and simulated monthly streamflow in the calibration and validation period at 10 hydrological stations. An asterisk means statistically significant value.

Table 2

d. Analysis methods

It is assumed that when streamflow and climate variables share similar temporal change patterns and abrupt changepoints, then climate change exerts a dominant impact on streamflow change; otherwise, human activities dominate. The assumption is based on the rationale that at an annual time step, climate change determines streamflow change when there is no interference from human activities. The abrupt changepoints of observed streamflow (Q), precipitation (P), potential evapotranspiration (PET), and temperature (T) are detected by the nonparametric Pettitt test (Pettitt 1979) in the R programming environment. If there is an abrupt changepoint in streamflow but not in climate variables, the period prior to and after the changepoint is defined as the baseline period and the changed period, respectively. The difference in the observed streamflow between the baseline and changed periods represent the total streamflow change, which is caused by both climate change and human activities:
ΔQtotal=Q2oQ1o,
ΔQtotal=ΔQhuman+ΔQclimate,
where Q2o and Q1o are the average observed annual streamflow after and before the changepoint, ΔQtotal is the total change in streamflow, ΔQhuman is the change in streamflow caused by human activities, and ΔQclimate is the change in streamflow caused by climate change.
The effects of climate change and human activities on streamflow can be quantified by the hydrological model and climate elasticity methods. The hydrological model, which only simulates natural streamflow without human intervention, is calibrated and validated in the period before the changepoint under natural conditions or at upland stations where human activity is limited. Then the model is used to simulate hydrological processes and reconstruct the natural streamflow for the entire period. The difference in the simulated natural streamflow between the baseline and changed periods represents the effect of climate change as shown in Eq. (4). From Eqs. (2)(4), ΔQhuman can be calculated and the respective contributions of climate change and human activities to streamflow change can be obtained:
ΔQclimate=Q2sQ1s,
where Q2s and Q1s are the average simulated annual streamflow after and before the changepoint.
According to Schaake (1990), the climate elasticity of streamflow is the proportional change in streamflow corresponding to the total proportional changes in precipitation and potential evapotranspiration. As such, the change of streamflow in response to climate change can be determined using the following equation:
ΔQclimate=(εP×ΔPP+εPET×ΔPETPET)×Q,
where εP and εPET are elasticity of observed streamflow with respect to P and PET, respectively; ΔP and ΔPET are the changes in precipitation and potential evapotranspiration; and P, PET, and Q are the average annual values before the occurrence of abrupt changepoints.
The climate elasticity of streamflow can then be estimated using the least squares estimator as follows (Zheng et al. 2009):
ε=X¯Q¯×(XiX¯)×(QiQ¯)(XiX¯)2,
where ε is elasticity of streamflow with respect to P or PET, Xi and Qi are climate variables (P or PET) and observed streamflow at time step i, and X¯ and Q¯ are long-term averages of the climate variables and observed streamflow.

Finally, using Eqs. (2), (3), (5), and (6), ΔQclimate and ΔQhuman and their respective contributions can be calculated.

The temporal trends of streamflow are examined using the Mann–Kendall test (Mann 1945; Kendall 1975) and Sen’s slope approach (Sen 1968). The significance level is set at p < 0.05 for statistical analysis. To evaluate the impacts of extreme precipitation conditions on streamflow changes, streamflow components are investigated in average, arid, and wet years at the outlet of the UADB. One of the most widely used drought indices is the standardized precipitation evapotranspiration index (SPEI). The SPEI is based on the climatology of water balance, i.e., the difference between P and PET (Vicente-Serrano et al. 2010). The Thornthwaite equation is used to calculate PET (Thornthwaite 1948). Following McKee et al. (1993), drought, wet, and average years are defined as years with SPEI values < −1, >1, and around 0, respectively. The SPEI calculation is conducted for the UADB at 0.25° × 0.25° spatial resolution. The areal extent affected by drought or wetness in each year is estimated by counting the numbers of grid cells with SPEI < −1 or >1, respectively. The top three most severe drought and wet events are chosen to analyze streamflow composition under extreme conditions.

3. Results

a. Calibration and validation of the VIC-glacier model

The simulated monthly streamflow in the calibration period matches the historical observations quite well at the Kerki and Chatly stations (Figs. 2a,b) and at the other eight stations in the UADB (Figs. S22 and S23). Specifically, the VIC-glacier simulations capture the magnitude of the observed monthly streamflow at all stations, and the timing of monthly streamflow peaks and troughs is consistent between the observations and simulations. However, some discrepancies do exist, e.g., the observed low flows in cold seasons are underestimated at Khirmanjo (Fig. S23c) and Kishl Pete (Fig. S23d). The R2, KGE, and NSE for monthly streamflow range from 0.71 to 0.91, from 0.72 to 0.95, and from 0.66 to 0.90, respectively, at the 10 stations (Table 2). The averaged R2, KGE, and NSE over the 10 stations are 0.85, 0.83, and 0.82, respectively. Seven out of the 10 stations show an NSE greater than 0.75, which suggests very good model performance (Moriasi et al. 2007). PBIAS values vary from −4.66% to 17.66%, with positive values indicating model underestimation. The reasonable performance of the model at the stations with diverse elevation, climate, and land cover types over various time periods demonstrates that the calibrated model can satisfactorily represent the hydrological processes of a wide range of physical environments in the ADB.

Fig. 2.
Fig. 2.

Observed and simulated monthly streamflow at Kerki and Chatly during the (a),(b) calibration and (c),(d) validation period. Note that y scales vary between panels.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

In the validation period, the simulated monthly streamflow also closely matches the observed streamflow at all stations (Figs. 2c,d, Figs. S24 and S25). The simulated low flows during the validation period are more consistent with the observations when compared to the calibration period. R2, KGE, NSE, and PBIAS at the 10 stations range from 0.72 to 0.94, from 0.65 to 0.92, from 0.59 to 0.89, and from −17.80% to 18.27%, respectively (Table 2). The averaged R2, KGE, and NSE over the 10 stations are 0.84, 0.80, and 0.79, respectively, which further demonstrates that the model performs satisfactorily in representing the diverse flow regimes. However, notable discrepancies exist at Khorog that shows the smallest NSE of 0.59 among the 10 stations.

The EASE-Grid estimated and VIC-glacier simulated monthly and long-term mean monthly SCF are compared during 1967–99/2000–19 over the ADB (Fig. 3). During 1967–99, the VIC-glacier generally overestimates (underestimates) monthly SCF for the winter (summer) season (Fig. 3a), but during 2000–19, the simulated SCF closely matches the EASE-Grid estimations (Fig. 3b). The R2, KGE, NSE, and PBIAS calculated for the monthly SCF during the two periods are 0.79, 0.57, 0.62, 10.36% and 0.92, 0.91, 0.89, −4.36%, respectively, suggesting that in general the performance of the VIC-glacier is acceptable in simulating SCF. Similarly, the VIC-glacier simulated long-term mean monthly SCF is lower than the EASE-Grid estimations during May–October in 1967–99 (Fig. 3c) but the comparison becomes much better in 2000–19 (Fig. 3d).

Fig. 3.
Fig. 3.

VIC-glacier simulated and EASE-Grid estimated SCF in the ADB: monthly SCF (a) in 1967–99 and (b) in 2000–19; long-term mean monthly SCF (c) in 1967–99 and (d) in 2000–19.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

The comparison of simulated annual glacier area change ratios with those of previous works (e.g., Konovalov and Shchetinnicov 1994; Konovalov 2011; Wagner and Hoelzle 2010) provides further evidence for the acceptable performance of the VIC-glacier. Specifically, the VIC-glacier simulated annual glacier area changes show a reduction by 0.61% and 0.35% in the Zeravshan and Pyanj subbasins, respectively, during 1957–80, and by 0.26% in the Vakhsh subbasin during 1961–2000, that all fall within the ranges reported by previous studies (Konovalov and Shchetinnicov 1994; Konovalov 2011; Wagner and Hoelzle 2010) (Fig. 4).

Fig. 4.
Fig. 4.

The annual glacier area decrease ratio in the Vakhsh, Pyanj, and Zeravshan subbasins. The red bars represent simulated values in this study, and the black bars represent the values from previous studies. The annual glacier area decrease ratio data is shown in the first column (1) from this study. Also shown are the annual glacier area decrease ratios in the second column (2) during 1957–80, third column (3) during 1961–2000, and fourth column (4) during 1950–2003 from Konovalov and Shchetinnicov (1994), Konovalov (2011), and Wagner and Hoelzle (2010), respectively.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

In summary, reliable streamflow records that span multiple periods with limited human activities are used to calibrate and validate the simulated streamflow at 10 hydrological stations. These stations are distributed in varied terrain, eight in the upper basin, one in the middle, and one at the outlet. The simulated SCF and glacier area change ratio are also evaluated against available estimations. Although in some specific years the discrepancy between the simulated and observed streamflow tends to be large (see Figs. 2 and S22–S25), the coefficients of determination at all stations during both the calibration and validation periods are consistently greater than 0.70 (Table 2), indicating that 70% of the observed streamflow variance could be explained by the simulation. The satisfactory performance of the VIC-glacier in simulating streamflow, SCF, and annual glacier area change ratio demonstrates the capability of the model in representing cryospheric hydrological processes in the ADB. In the following, we will utilize the simulations to examine streamflow composition.

b. Streamflow composition

We assume that the ADB streamflow is composed of rainfall, glacier melt, and snowmelt only. The simulated monthly and annual natural streamflow components averaged over 1953–2019 at 17 sites in the ADB are shown in Figs. 5 and 6. At Kerki and Chatly, streamflow is dominated by snowmelt, accounting for 43.1% and 50.3%, respectively (Fig. 5a). Glacier melt, contributing 46.6%, is the dominant streamflow component at Khojagar (Fig. 5a). At Pul-I-Chugha and Char Dara, the contribution of glacier melt to streamflow is small due to minimal glacier area there. The rainfall contribution, accounting for 52.4%, dominates annual streamflow at Char Dara; while the snowmelt contribution, accounting for 49.1%, dominates annual streamflow at Pul-I-Chugha (Fig. 5a). Rabat-I-Bala, Asiabad, Sayad, and Pata Baba, located in the Afghanistan side of the southern ADB, are distinct from the other stations in that streamflow at these stations does not contain a glacier melt component and also rainfall contribution exceeds that of snowmelt. Overall, rainfall and snowmelt contributions to annual streamflow range from 56.3% to 75.5% and from 24.5% to 43.7%, respectively, at these southern stations (Fig. 5a). The snow-and-glacier-melt-dominated streamflow regime peaks during summer, when meltwater is at its maximum (Figs. 5b,d). The rainfall-and-snowmelt-dominated streamflow regime peaks during spring (Figs. 5e–j). At Chatly (Fig. 5c), there is a major peak in summer and a secondary but elongated peak in spring and early summer, due to contributions from glacier melt and snowmelt, respectively.

Fig. 5.
Fig. 5.

Average (a) annual and (b)–(j) monthly contributions of rainfall, snowmelt, and glacier melt to streamflow at locations in the southern ADB during 1953–2019.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

Fig. 6.
Fig. 6.

Average (a) annual and (b)–(i) monthly contributions of rainfall, snowmelt, and glacier melt to streamflow at locations in the northern and eastern ADB during 1953–2019.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

For stations located in the northern and eastern ADB where elevation is high, streamflow composition is dominated by glacier melt at all stations except Garm, for which snowmelt is dominant (Fig. 6). At the stations except Garm, glacier melt contribution to annual streamflow ranges from 54.5% to 60.1%, while the snowmelt contribution varies from 15.4% to 25.8% (Fig. 6a). At Garm, despite its largest glacierized area compared to the other stations, glacier melt contribution (38.7%) is smaller than that of snowmelt (49.3%) (Fig. 6a). Rainfall contribution to annual streamflow is less than 27% at these stations. Seasonally, all stations show one major July–August streamflow peak when glacier melt peaks (Figs. 6b–i). The rainfall component generally displays two small peaks in June–July and October, respectively.

The absolute amounts of streamflow components and their relative contributions to annual streamflow show interannual variability during 1953–2019. The natural annual total streamflow, rainfall and glacier melt components at Kerki increase significantly at 4.2, 0.5, and 2.6 m3 s−1 yr−1 during 1953–2019, respectively (Fig. 7a). The annual snowmelt component shows an insignificant increasing trend of 0.8 m3 s−1 yr−1 (Fig. 7a). Chatly also exhibits similar interannual variability in streamflow composition to that at Kerki though with slightly different change rates (Fig. 7b). On average, the contributions of rainfall, snowmelt, and glacier melt to annual streamflow at Kerki stay at 17.2% ± 2.0%, 43.1% ± 6.5%, and 39.7% ± 6.9%, respectively, with a negative trend in snowmelt (−0.04% yr−1) and a positive trend in glacier melt (0.03% yr−1) during 1953–2019 (Fig. 7c). At Chatly, the above numbers become 23.0% ± 2.4%, 50.3% ± 5.7%, and 26.7% ± 5.7%, respectively, with a negative trend in snowmelt (−0.01% yr−1) and a positive trend in glacier melt (0.01% yr−1; Fig. 7d). The observed streamflow at Garm, Most Dupuli, and Hirmanjo with little human intervention also increases (Fig. S26), further demonstrating that without human intervention, natural streamflow would increase due to long-term climate change.

Fig. 7.
Fig. 7.

Simulated natural annual total streamflow, rainfall, snowmelt, and glacier melt components at (a) Kerki and (b) Chatly during 1953–2019. Simulated contribution of rainfall, snowmelt, and glacier melt components at (c) Kerki and (d) Chatly during 1953–2019. An asterisk indicates a statistically significant trend.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

Besides long-term changes, climate also undergoes extreme variability and can be generally categorized into average, drought, and wet conditions in terms of the SPEI. Three average years (1959, 1980, and 2004), drought years (1971, 2001, and 2008), and wet years (1958, 1969, and 1992) are selected for analysis (Fig. 8a). The areal extent affected by drought and wetness in the UADB are counted as the numbers of grid cells with SPEI < −1 or >1, respectively, and are shown in Figs. 8b and 8c. More than 60% and 50% of the UADB is affected by the three drought and wet events, respectively.

Fig. 8.
Fig. 8.

(a) Annual SPEI, (b) annual drought area extent, and (c) annual wet area extent during 1953–2019.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

During the three average years, 16.1%, 41.8%, and 42.1% of annual streamflow is from rainfall, snowmelt, and glacier melt, respectively, at Kerki (Fig. S27a), similar to the long-term averages (Fig. 5a). Changes in the absolute amount of rainfall component occur in spring, while changes in snow and glacier melt components occur in April–September (Fig. S27b). Such changes in streamflow components are associated with the changes in precipitation and temperature (Figs. S27c,d). During the drought years, glacier melt contribution to annual streamflow increases to 51.9%, whereas snowmelt contribution diminishes to 31.4% (Fig. 9a) when compared to the average years. The absolute amounts of snowmelt and rainfall components decrease by 260.8 and 30.1 m3 s−1, respectively, during the drought years (Fig. 9b) due to the mean annual precipitation 84 mm lower than the long-term average of 457 mm (Fig. 9c). On the other hand, the glacier melt component increases by 160.0 m3 s−1 (Fig. 9b) because the decreased snowfall results in thinner snow cover that gives rise to the earlier and greater glacier melt. Above average temperatures during most months of the drought years are also conducive to more glacier melt (Fig. 9d). The enhanced glacier melt in the drought years also reported by Pritchard (2019) could alleviate drought stress.

Fig. 9.
Fig. 9.

(a) Average monthly and annual contributions of rainfall, snowmelt, and glacier melt to streamflow and (b) changes in absolute amount of streamflow components relative to the averages in 1953–2019 at Kerki during the drought years. (c) Precipitation departures and (d) temperature departures from the averages of 1953–2019.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

Snowmelt contribution to annual streamflow, with double peaks in March and July, increases to 57.5%, whereas glacier melt contribution diminishes to 26.3% during the wet years (Fig. 10a) when compared to the average years. The absolute amounts of snowmelt and rainfall components increase by 378.6 and 12.0 m3 s−1, respectively (Fig. 10b), because of precipitation during the wet years (Fig. 10c) exceeding by 93 mm the long-term average, primarily in the form of increased snowfall in winter. The glacier melt component is mainly controlled by the amount of winter snowfall and air temperature. Excessive winter snowfall and abnormally low air temperature during the wet years (Figs. 10c,d) result in the reduction of glacier melt component by 201.8 m3 s−1 (Fig. 10b).

Fig. 10.
Fig. 10.

As in Fig. 9, but for the wet years.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

c. Deconvolving the impacts of human activities and climate change on streamflow

1) Abrupt change points

The Pettitt test for P, T, and PET at both Kerki and Kiziljar reveals 1990, 1998, and 1998, respectively, as the changepoints, whereas the changepoint of the observed streamflow at both locations occurred in 1973 (Figs. 11a,b). This discrepancy in abrupt changepoints between climate variables and streamflow suggests that human activities, not climate factors, have drastically altered streamflow. Given that the VIC-glacier can only simulate natural streamflow related to climate and climate change, if a basin is disturbed little by human activities, then the observed and simulated streamflow should be similar (see our calibration results in Fig. 2) during the entire study period. However, significant discrepancies between the simulations and observations are seen since around 1973 (Figs. 11c,d), and this indicates amplified human influences, especially after the abrupt streamflow changepoint of 1973.

Fig. 11.
Fig. 11.

Abrupt changepoints of observed streamflow, P, PET, and T at (a) Kerki and (b) Kiziljar; the left y scale is for observed streamflow, P, PET, and the right y scale is for T. Simulated and observed annual streamflow at (c) Kerki and (d) Kiziljar during 1956–2015.

Citation: Journal of Hydrometeorology 24, 2; 10.1175/JHM-D-22-0040.1

In fact, since the 1960s, progressively increased water consumption associated with human activities has been observed in the ADB (Micklin 2007). This was because during the same time period hundreds of thousands of people migrated to the reclaimed areas to work in the agricultural sector and hydrological facilities were constructed for irrigation. From 1970 to 1989, the irrigated area expanded by a factor of 150% in the ADB (FAO 2013). Irrigation relies upon a system of dams and canals. Since 1960 more than 60 canals have been used to divert water from the Amu Darya River to assist with agricultural production (Adenbaev et al. 2015). The largest canals along the middle reach of the ADB are the Karakum Canal in Turkmenistan and the Karshi Canal in Uzbekistan, constructed in 1967 and 1974 and with water intake reaching 800–850 and 350–375 m3 s−1, respectively. The Karakum and Karshi Canals divert about 10% of the streamflow from the Amu Darya River at Kerki to outside the basin. The water intake along the Kerki–Chatly section has reached 20%–28% of the streamflow at Kerki (Adenbaev et al. 2015). The Nurek Dam (with a capacity of 10.5 km3) along the Vakhsh River began operation in 1972 and supplied irrigation water for about 70 000 ha, which changed the discharge characteristics of the subbasin (Froebrich et al. 2006). Some canals and other hydrological facilities started operations in the 1960s, which may explain the noticeable differences between the observed and simulated runoff at Kiziljar during 1956–73 (Fig. 11d). However, major dams and facilities started operations in the early 1970s and this corresponded well to the detected abrupt changepoints of 1973. Because streamflow data and model forcing before 1950 are not available at Kiziljar, we treat 1956–73 that displays much smaller streamflow differences between the observations and simulations than the period of 1974–2015 as a reference period for comparison with 1974–2015. In this analysis, 1956–2015 is divided into two parts: a baseline period (1956–73) and a changed period (1974–2015). The two periods are used for the deconvolving of change impacts on streamflow in the ADB.

2) VIC-glacier simulation and climate elasticity analysis

At Kerki, the observed average annual runoff was 195.1 mm for 1956–73 and 151.2 mm for 1974–2015, meaning that the total runoff change was −43.9 mm (Table 3). The simulated average annual runoff was 191.4 mm for 1956–73 and 201.1 mm for 1974–2015 (Table 3), meaning that the change in runoff caused by climate change was 9.7 mm (or −22.1% of the observed total runoff change). Thus, the runoff change caused by human activities was −53.6 mm or 122.1% of the observed total runoff change. At Kiziljar, the observed total runoff change was −58.6 mm during 1974–2015, and the change caused by climate change (human activities) was 4.5 mm (−63.1 mm), equal to −7.7% (107.7%) of the observed total runoff change (Table 3).

Table 3

Quantified impacts of climate change and human activities on streamflow based on the VIC-glacier simulations and observed streamflow.

Table 3

The P elasticity of streamflow is 0.71 at Kerki (Table 4), indicating that a 10% change in precipitation would result in a 7.1% change in streamflow. The PET elasticity of streamflow is −0.07, meaning that a 10% increase in PET would result in a 0.7% decrease in streamflow. Following Eq. (5), runoff in 1974–2015 should be around 2.8 mm more than that of 1956–73 because of the 9.0 mm increase in P during 1974–2015 versus 1956–73. On the other hand, the 35.6 mm increase in PET in 1974–2015 would result in a 0.8 mm decrease in runoff. Therefore, the combined impacts of both P and PET would result in a 2.0 mm increase in average annual runoff during 1974–2015, which amounts to −4.6% of the total runoff change. Following Eqs. (2) and (3), human activities would result in a 45.9 mm decrease in average annual runoff, accounting for 104.6% of the total runoff change (Table 4). Similarly at Kiziljar, runoff should have a 2.1 mm increase due to the 10.9 mm increase in P and a 0.2 mm decrease due to the 43.5 mm increase in PET during 1974–2015 versus 1956–73. Therefore, the combined impacts of both P and PET would lead to a 1.9 mm increase in average annual runoff in 1974–2015, accounting for −3.2% of the total runoff change (Table 4). Consequently, human activities should give rise to a 60.5 mm decrease in average annual runoff, amounting to 103.2% of the total runoff change (Table 4).

Table 4

Quantified impacts of climate change and human activities on streamflow based on the climate elasticity approach.

Table 4

The above hydrological model and elasticity analysis indicates that the contributions of climate change (human activities) to streamflow reduction between 1974–2015 and 1956–73 range from −22.1% to −3.2% (from 103.2% to 122.1%) at Kerki and Kiziljar. Here a negative percentage represents a beneficial impact that increases streamflow while a positive percentage signifies reduced streamflow. Though there are differences in exact numbers, the hydrological model and climate elasticity approaches corroborate each other by showing that the significant impacts of human activity dominate over those from climate change, with the net result of significant annual streamflow reduction. Additionally, the absolute runoff reduction caused by human activities from both approaches shows that detrimental human activities impacts appear more severe in the downstream (Kiziljar) than in the upper stream (Kerki) of the ADB.

4. Discussion

a. Streamflow composition by other studies

Savoskul and Smakhtin (2013) studied streamflow components in the ADB for the period of 1961–90 and documented a glacier melt component (including glacier melt, snowmelt, and rainfall in glacierized areas) of 25%, a snowmelt component of 39%, a rainfall component of 1%, and a baseflow component of 35% of the annual streamflow. Armstrong et al. (2019) used areas of seasonal snow cover and ice derived from the MODIS products as inputs to a temperature index model to calculate the contributions of glacier melt, snowmelt, and rainfall to streamflow at Chatly. They showed that the glacier melt, snowmelt, and rainfall components accounted for 8%, 69%, and 23% of the annual streamflow, respectively. These estimated glacier melt contributions were much smaller than what we obtained. Using a distributed cryospheric hydrological model, Immerzeel et al. (2012) reported the contributions of glacier melt (including clean ice melt and debris-covered glacier melt), snowmelt, rainfall, and baseflow in the UADB to be 38%, 26.9%, 16.5%, and 18.6%, respectively, similar to our results in terms of glacier melt and rainfall contributions. A recent study by Liu et al. (2020b) suggested that the combined glacier and snowmelt contributed to about 70% of the annual streamflow in the source region of the Amu Darya River. Other published estimates of glacier melt contribution in the ADB were for smaller upper subbasins. For example, based on the physical and statistical models, Konovalov (2011) estimated the contribution of glacier melt to streamflow in the Pyanj, Vakhsh, and Zeravshan subbasins to be 26.2%, 20.9%, and 24.4%, respectively. Hagg et al. (2013) reported that glacier melt contributed to about 32% of the annual streamflow in the upper Pyanj subbasin. Differences between this study and previous publications (Table 5) arise because of differences in the derivation of glacier and snowmelt components, glacier source data, study areas and approaches. Here, glacier data were from RGI 4.0 and glacier melt contribution was derived as all streamflow coming from the entire glacierized area without separating ice from snow on the glacier surface. As such, more studies must be undertaken to fully partition streamflow composition. Only after using the same derivation of glacier and snowmelt components and the same data sources, should the results be fairly compared. Different from previous studies, we also investigated and compared streamflow composition at locations that collect freshwater from all mountainous subbasins.

Table 5

Comparison of streamflow composition in the ADB basin among the published data. Note: “—” means missing information.

Table 5

b. Comparison of hydrological modeling with climate elasticity approach

We used the climate elasticity and VIC-glacier model approaches to deconvolve the impacts of climate change and human activities on streamflow in the ADB. The two approaches have both advantages and shortcomings on their own (Cuo 2016; Chang et al. 2016; Dey and Mishra 2017). The climate elasticity approach is a relatively simple statistical method with fewer data and parameter requirements compared to hydrological modeling. But the elasticity approach operates on annual time scale because of the inherent assumption of mass balance and is thus unable to include the variability at subannual time scales. In contrast, the VIC-glacier model is physically based and can simulate hydrological processes at finer time scales. However, it requires detailed spatial input data and its calibration and validation also remain a challenge. As such, uncertainties exist in both methods when estimating the impacts of climate change and human activities on streamflow.

The two methods agree that human activities have drastically reduced streamflow and that climate change has contributed modestly to streamflow increase in the ADB, consistent with the results by Hu et al. (2021). These authors used the elasticity coefficient method coupled with the Budyko framework to quantitatively analyze streamflow change during 1960–2017 and reported 11.79%–13.38% streamflow increase due to climate change and 111.79%–113.38% streamflow reduction because of human activities at Kerki (Hu et al. 2021). However, there are still differences in the magnitude of change impacts calculated individually by the two methods. Compared to the hydrological modeling result, streamflow changes caused by climate change and human activities using the elasticity-based method are smaller in magnitude. Note that the two methods use the same meteorological forcing so the discrepancy is solely due to the difference in the approaches. The elasticity-based method uses a simple statistical analysis to reveal the complex nonlinear relationship between streamflow and climate variables while neglecting soil moisture, snow and glacier accumulation, and melt processes. These processes can be considered in the VIC-glacier, which may partially explain the difference in the magnitude between the two methods. But due to the inherent imperfection in the model parameters, model spatial and temporal input data, initial conditions, and model structure, the simulated result also contains a certain degree of uncertainty. Clearly, there are advantages and shortcomings, as well as various sources of errors and uncertainties in the hydrological model and elasticity-based methods. Whenever possible, a comprehensive analysis using both approaches is recommended to improve the fidelity of the analysis results.

It should also be noted here that the framework used to isolate the impacts of climate change and human activities on streamflow is based on the assumption that land use and land cover change is independent of climate change. However, in reality, the land surface and climate system interact with each other in a dynamic way. Especially at a catchment scale, climate change may play an important role in land cover change which may subsequently change streamflow. Therefore in this study, the quantified human activities impacts on streamflow are limited to direct human impacts but the potential indirect feedbacks of human activities to P, PET, and streamflow are not considered.

5. Conclusions

In this study, a glacio-hydrological modeling framework, VIC-glacier, is developed by linking a macroscale physically based VIC model with a degree-day glacier-melt module. The monthly observed streamflow at 10 hydrological stations, satellite-derived snow cover fraction, and glacier area change ratio are simulated by the VIC-glacier with reasonable accuracy. The calibrated model is subsequently used to assess streamflow composition of all tributaries and the main branch within the ADB. The SPEI indices calculated with the gridded precipitation and temperature data are used to categorize average, dry, and wet years for the analysis of streamflow composition changes under extreme conditions. Furthermore, by combining the VIC-glacier model with the climate elasticity method, we quantify the relative contributions of climate change and human activities to streamflow in the ADB with reduced uncertainty.

The contributions of rainfall, snowmelt, and glacier melt to the total streamflow vary greatly across the ADB. In the eastern and northern subbasins with high glacier coverage, streamflow composition is dominated by snow and glacier melt components. Rainfall and snowmelt are the dominant streamflow components in the southern lowland subbasins. The trend analysis indicates that all components of natural streamflow at Kerki and Chatly increased during 1953–2019. However, there is substantial variability in natural streamflow composition under extreme precipitation conditions in the UADB. Compared to the long-term averages, a decrease in the snowmelt (−260.8 m3 s−1) and rainfall (−30.1 m3 s−1) contributions but an increase in glacier melt contribution (160.0 m3 s−1) to annual streamflow are revealed during drought years. In contrast, there are increases in the snowmelt (378.6 m3 s−1) and rainfall (12.0 m3 s−1) contributions but a decrease in glacier melt contribution (−201.8 m3 s−1) to annual streamflow during wet years.

The observed annual streamflow showed decreasing trends at Kerki and Kiziljar during 1956–2015. The climate change contributions of −22.1% and −7.7% to streamflow decrease at Kerti and Kiziljar, respectively, are calculated with the VIC-glacier model, while the contributions of −4.6% and −3.2% are obtained from the elasticity method. The contributions to streamflow reduction by human activities are 122.1% and 107.7% based on the VIC-glacier model and 104.6% and 103.2% based on the elasticity method at Kerki and Kiziljar, respectively. Despite the differences in the magnitude of the contributions between the two methods, both methods show the substantial water losses due to direct human intervention that are much more than modest water gains from the impacts of climate change, causing annual streamflow to decline drastically after 1973. Given that there are different types of uncertainties associated with predefined assumptions, input data, parameters, mathematical framework, initial conditions in the hydrological model, and elasticity-based method, a combination of both approaches is recommended for a robust analysis of the deconvolvement of human activities and climate change impacts on streamflow.

Acknowledgments.

This work is supported by the joint program from Chinese Academy of Sciences and Sanjiangyuan National Park (Grant LHZX-2020-10-4), the “Strategic Priority Research Program” of the Chinese Academy of Sciences (Grant XDA20060202), and the Second Tibetan Plateau Scientific Expedition and Research Program (Grant 2019QZKK0203).

Data availability statement.

Observed monthly streamflow data at Kerki and Chatly used for model calibration and validation are from the Global Runoff Data Center via https://www.bafg.de/GRDC/EN/Home/homepage_node.html. Other streamflow observations are purchased from the Hydrometeorological Agency of Tajikistan under a nondisclosure agreement. Observed temperature and precipitation are from the Global Historical Climate Network (GHCN-Daily, https://www.ncei.noaa.gov/products/land-based-station/global-historical-climatology-network-daily). Gridded precipitation and temperature and other model input data are available from their websites described in the text. Simulation results can be accessed by contacting the corresponding author.

REFERENCES

  • Adenbaev, B. Y., F. H. Hikmatov, and O. A. Kxaydarova, 2015: Hydrological regime of the lower reaches of Amudarja River in the conditions of intensive economical use of water resources. Eur. Sci. Rev, 9–10, 1821.

    • Search Google Scholar
    • Export Citation
  • Agal’tseva, N. A., M. V. Bolgov, T. Y. Spektorman, M. D. Trubetskova, and V. E. Chub, 2011: Estimating hydrological characteristics in the Amu Darya River basin under climate change conditions. Russ. Meteor. Hydrol., 36, 681689, https://doi.org/10.3103/S1068373911100062.

    • Search Google Scholar
    • Export Citation
  • Andreadis, K. M., P. Storck, and D. P. Lettenmaier, 2009: Modeling snow accumulation and ablation processes in forested environments. Water Resour. Res., 45, W05429, https://doi.org/10.1029/2008WR007042.

    • Search Google Scholar
    • Export Citation
  • Armstrong, R. L., and Coauthors, 2019: Runoff from glacier ice and seasonal snow in High Asia: Separating melt water sources in river flow. Reg. Environ. Change, 19, 12491261, https://doi.org/10.1007/s10113-018-1429-0.

    • Search Google Scholar
    • Export Citation
  • Barnett, T. P., J. C. Adam, and D. P. Lettenmaier, 2005: Potential impacts of a warming climate on water availability in snow-dominated regions. Nature, 438, 303309, https://doi.org/10.1038/nature04141.

    • Search Google Scholar
    • Export Citation
  • Brodzik, M. J., and R. Armstrong, 2013: Northern Hemisphere EASE-Grid 2.0 weekly snow cover and sea ice extent, version 4. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 23 September 2020, https://doi.org/10.5067/P7O0HGJLYUQU.

  • Chang, J., H. Zhang, Y. Wang, and Y. Zhu, 2016: Assessing the impact of climate variability and human activities on streamflow variation. Hydrol. Earth Syst. Sci., 20, 15471560, https://doi.org/10.5194/hess-20-1547-2016.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., W. Li, G. Fang, and Z. Li, 2017: Review article: Hydrological modeling in glacierized catchments of central Asia–Status and challenges. Hydrol. Earth Syst. Sci., 21, 669684, https://doi.org/10.5194/hess-21-669-2017.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., Z. Li, G. Fang, and W. Li, 2018: Large hydrological processes changes in the transboundary rivers of Central Asia. J. Geophys. Res. Atmos., 123, 50595069, https://doi.org/10.1029/2017JD028184.

    • Search Google Scholar
    • Export Citation
  • Cherkauer, K. A., and D. P. Lettenmaier, 1999: Hydrologic effects of frozen soils in the upper Mississippi River basin. J. Geophys. Res., 104, 19 59919 610, https://doi.org/10.1029/1999JD900337.

    • Search Google Scholar
    • Export Citation
  • Cuo, L., 2016: Land use/cover change impacts on hydrology in large river basins: A review. Terrestrial Water Cycle and Climate Change: Natural and Human-Induced Impacts, Geophys. Monogr., Vol., 221, Amer. Geophys. Union, 103–134, https://doi.org/10.1002/9781118971772.ch6.

  • Cuo, L., D. P. Lettenmaier, M. Alberti, and J. E. Richey, 2009: Effects of a century of land cover and climate change on hydrology in Puget Sound. Hydrol. Processes, 23, 907933, https://doi.org/10.1002/hyp.7228.

    • Search Google Scholar
    • Export Citation
  • Cuo, L., Y. Zhang, Y. Gao, Z. Hao, and L. Cairang, 2013: The impacts of climate change and land cover transition on the hydrology in the upper Yellow River basin, China. J. Hydrol., 502, 3752, https://doi.org/10.1016/j.jhydrol.2013.08.003.

    • Search Google Scholar
    • Export Citation
  • Dey, P., A. Mishra, 2017: Separating the impacts of climate change and human activities on streamflow: A review of methodologies and critical assumptions. J. Hydrol., 548, 278290, https://doi.org/10.1016/j.jhydrol.2017.03.014.

    • Search Google Scholar
    • Export Citation
  • FAO, 2013: Irrigation in Central Asia in figures. FAO Water Rep. 39, 246 pp., http://www.fao.org/docrep/018/i3289e/i3289e.pdf.

  • FAO, 2021: AQUASTAT core database. Food and Agriculture Organization of the United Nations, accessed 7 September 2021, http://www.fao.org/nr/water/aquastat/data/query/index.html?lang=en.

  • Froebrich, J., O. Olsson, M. Bauer, I. Normatov, and G. Petrov, 2006: Improved dam operation in the Amu Darya Basin including transboundary aspects. Dams and Reservoirs, Societies and Environment in the 21st Century, L. Berga et al., Eds., Taylor and Francis Group, 97–103.

  • Gleick, P., C. Iceland, and A. Trivedi, 2020: Ending conflicts over water: Solutions to water and security challenges. Water, Peace and Security Rep., 128 pp., https://pacinst.org/wp-content/uploads/2020/09/Ending-Conflicts-Over-Water-Pacific-Institute-Sept-2020.pdf.

  • Gupta, H. V., S. Sorooshian, and P. O. Yapo, 1999: Status of automatic calibration for hydrologic models: Comparison with multilevel expert calibration. J. Hydrol. Eng., 4, 135143, https://doi.org/10.1061/(ASCE)1084-0699(1999)4:2(135).

    • Search Google Scholar
    • Export Citation
  • Hagg, W., M. Hoelzle, S. Wagner, E. Mayr, and Z. Klose, 2013: Glacier and runoff changes in the Rukhk catchment, upper Amu-Darya basin until 2050. Global Planet. Change, 110, 6273, https://doi.org/10.1016/j.gloplacha.2013.05.005.

    • Search Google Scholar
    • Export Citation
  • Hock, R., 1999: A distributed temperature-index ice-and snowmelt model including potential direct solar radiation. J. Glaciol., 45, 101111, https://doi.org/10.3189/S0022143000003087.

    • Search Google Scholar
    • Export Citation
  • Hu, Y., W. Duan, Y. Chen, S. Zou, P. M. Kayumba, and N. Sahu, 2021: An integrated assessment of runoff dynamics in the Amu Darya River Basin: Confronting climate change and multiple human activities, 1960–2017. J. Hydrol., 603, 126905, https://doi.org/10.1016/j.jhydrol.2021.126905.

    • Search Google Scholar
    • Export Citation
  • Immerzeel, W. W., A. F. Lutz, and P. Droogers, 2012: Climate change impacts on the upstream water resources of the Amu and Syr Darya River Basins. Future Water Rep. 107, 103 pp., https://www.futurewater.nl/wp-content/uploads/2012/03/Upstream_Report_FW_web.pdf.

  • Istanbulluoglu, E., T. Wang, O. M. Wright, and J. D. Lenters, 2012: Interpretation of hydrologic trends from a water balance perspective: The role of groundwater storage in the Budyko hypothesis. Water Resour. Res., 48, W00H16, https://doi.org/10.1029/2010WR010100.

    • Search Google Scholar
    • Export Citation
  • Jiang, S., L. Ren, B. Yong, V. P. Singh, X. Yang, and F. Yuan, 2011: Quantifying the effects of climate variability and human activities on runoff from the Laohahe basin in Northern China using three different methods. Hydrol. Processes, 25, 24922505, https://doi.org/10.1002/hyp.8002.

    • Search Google Scholar
    • Export Citation
  • Jilili, R., 2019: Study on the impacts of climate change and human activities on runoff in the Amu Darya River, Central Asia (in Chinese). M.S. thesis, Dept. of Geography and Tourism, Xinjiang Normal University, 61 pp.

  • Kendall, M. G., 1975: Rank Correlation Methods. 4th ed. Charles Griffin, 202 pp.

  • Khan, V. M., and L. Holko, 2009: Snow cover characteristics in the Aral Sea Basin from different data sources and their relation with river runoff. J. Mar. Syst., 76, 254262, https://doi.org/10.1016/j.jmarsys.2008.03.012.

    • Search Google Scholar
    • Export Citation
  • Kling, H., M. Fuchs, and M. Paulin, 2012: Runoff conditions in the upper Danube basin under an ensemble of climate change scenarios. J. Hydrol., 424425, 264277, https://doi.org/10.1016/j.jhydrol.2012.01.011.

    • Search Google Scholar
    • Export Citation
  • Konovalov, V. G., 2011: Past and prospective change in state of Central Asian glaciers. Ice Snow, 3, 6068.

  • Konovalov, V. G., and A. S. Shchetinnicov, 1994: Evolution of glaciation in the Pamiro-Alai Mountains and its effect on river runoff. J. Glaciol., 40, 149157, https://doi.org/10.1017/S0022143000003920.

    • Search Google Scholar
    • Export Citation
  • Li, L. J., L. Zhang, H. Wang, J. Wang, J. W. Yang, D. J. Jiang, J. Y. Li, and D. Y. Qin, 2007: Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China. Hydrol. Processes, 21, 34853491, https://doi.org/10.1002/hyp.6485.

    • Search Google Scholar
    • Export Citation
  • Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges, 1994: A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophys. Res., 99, 14 41514 428, https://doi.org/10.1029/94JD00483.

    • Search Google Scholar
    • Export Citation
  • Liang, X., E. F. Wood, and D. P. Lettenmaier, 1996: Surface soil moisture parameterization of the VIC-2L model: Evaluation and modification. Global Planet. Change, 13, 195206, https://doi.org/10.1016/0921-8181(95)00046-1.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., L. Cuo, Q. Li, X. Liu, X. Ma, L. Liang, and J. Ding, 2020a: Impacts of climate change and land use/cover change on streamflow in Beichuan River Basin in Qinghai Province, China. Water, 12, 1198, https://doi.org/10.3390/w12041198.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., Z. Yao, R. Wang, and G. Yu, 2020b: Estimation of the Qinghai-Tibetan plateau runoff and its contribution to large Asian rivers. Sci. Total Environ., 749, 141570, https://doi.org/10.1016/j.scitotenv.2020.141570.

    • Search Google Scholar
    • Export Citation
  • Lohmann, D., E. Raschke, B. Nijssen, and D. P. Lettenmaier, 1998: Regional scale hydrology: I. Formulation of the VIC-2L model coupled to a routing model. Hydrol. Sci. J., 43, 131141, https://doi.org/10.1080/02626669809492107.

    • Search Google Scholar
    • Export Citation
  • Lutz, A. F., W. W. Immerzeel, A. B. Shrestha, and M. F. P. Bierkens, 2014: Consistent increase in High Asia’s runoff due to increasing glacier melt and precipitation. Nat. Climate Change, 4, 587592, https://doi.org/10.1038/nclimate2237.

    • Search Google Scholar
    • Export Citation
  • Mann, H. B., 1945: Nonparametric tests against trend. Econometrica, 13, 245259, https://doi.org/10.2307/1907187.

  • McKee, T. B., N. J. Doeskin, and J. Kleist, 1993: The relationship of drought frequency and duration to time scales. 8th Conf. on Applied Climatology, Boston, MA, Amer. Meteor. Soc., 179–184.

  • Micklin, P., 2007: The Aral Sea disaster. Annu. Rev. Earth Planet. Sci., 35, 4772, https://doi.org/10.1146/annurev.earth.35.031306.140120.

    • Search Google Scholar
    • Export Citation
  • Micklin, P., 2016: The future Aral Sea: Hope and despair. Environ. Earth Sci., 75, 844, https://doi.org/10.1007/s12665-016-5614-5.

  • Moriasi, D. N., J. G. Arnold, M. W. Van Liew, R. L. Bingner, R. D. Harmel, and T. L. Veith, 2007: Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Trans. ASABE, 50, 885900, https://doi.org/10.13031/2013.23153.

    • Search Google Scholar
    • Export Citation
  • Nash, J. E., and J. V. Sutcliffe, 1970: River flow forecasting through conceptual models: Part 1. A discussion of principles. J. Hydrol., 10, 282290, https://doi.org/10.1016/0022-1694(70)90255-6.

    • Search Google Scholar
    • Export Citation
  • Olsson, O., M. Gassmann, K. Wegerich, and M. Bauer, 2010: Identification of the effective water availability from streamflows in the Zerafshan river basin, Central Asia. J. Hydrol., 390, 190197, https://doi.org/10.1016/j.jhydrol.2010.06.042.

    • Search Google Scholar
    • Export Citation
  • Pettitt, A. N., 1979: A non-parametric approach to the change point problem. J. Appl. Stat., 28, 126135, https://doi.org/10.2307/2346729.

    • Search Google Scholar
    • Export Citation
  • Pritchard, H. D., 2019: Asia’s shrinking glaciers protect large populations from drought stress. Nature, 569, 649654, https://doi.org/10.1038/s41586-019-1240-1.

    • Search Google Scholar
    • Export Citation
  • RGI Consortium, 2014: Randolph Glacier Inventory-A dataset of global glacier outlines, version 4.0. National Snow and Ice Data Center, accessed 1 December 2014, https://doi.org/10.7265/bse7-r330.

  • Sankarasubramanian, A., R. M. Vogel, and J. F. Limbrunner, 2001: Climate elasticity of streamflow in the United States. Water Resour. Res., 37, 17711781, https://doi.org/10.1029/2000WR900330.

    • Search Google Scholar
    • Export Citation
  • Savoskul, O. S., and V. Smakhtin, 2013: Glacier systems and seasonal snow cover in six major Asian river basins: Hydrological role under changing climate. International Water Management Institute Research Rep. 150, 45 pp., https://www.iwmi.cgiar.org/2013/07/glacier-systems-and-seasonal-snow-cover-in-six-major-asian-river-basins-hydrological-role-under-changing-climate/.

  • Schaake, J. C., 1990: From climate to flow. Climate Change and U.S. Water Resources, P. E. Waggoner, Ed., John Wiley, 177–206.

  • Sen, P. K., 1968: Estimates of regression coefficient based on Kendall’s tau. J. Amer. Stat. Assoc., 63, 13791389, https://doi.org/10.1080/01621459.1968.10480934.

    • Search Google Scholar
    • Export Citation
  • Sorg, A., T. Bolch, M. Stoffel, O. Solomina, and M. Beniston, 2012: Climate change impacts on glaciers and runoff in Tien Shan (Central Asia). Nat. Climate Change, 2, 725731, https://doi.org/10.1038/nclimate1592.

    • Search Google Scholar
    • Export Citation
  • Stulina, G., and O. Eshchanov, 2013: Climate change impacts on hydrology and environment in the Pre-Aral region. Quat. Int., 311, 8796, https://doi.org/10.1016/j.quaint.2013.07.015.

    • Search Google Scholar
    • Export Citation
  • Su, Y., and Coauthors, 2021: High agricultural water consumption led to the continued shrinkage of the Aral Sea during 1992–2015. Sci. Total Environ., 777, 145993, https://doi.org/10.1016/j.scitotenv.2021.145993.

    • Search Google Scholar
    • Export Citation
  • Thornthwaite, C. W., 1948: An approach toward a rational classification of climate. Geogr. Rev., 38, 5594, https://doi.org/10.2307/210739.

    • Search Google Scholar
    • Export Citation
  • Todini, E., 1996: The ARNO rainfall-runoff model. J. Hydrol., 175, 339382, https://doi.org/10.1016/S0022-1694(96)80016-3.

  • Unger-Shayesteh, K., S. Vorogushyn, D. Farinotti, A. Gafurov, D. Duethmann, A. Mandychev, and B. Merz, 2013: What do we know about past changes in the water cycle of Central Asian headwaters? A review. Global Planet. Change, 110, 425, https://doi.org/10.1016/j.gloplacha.2013.02.004.

    • Search Google Scholar
    • Export Citation
  • Vicente-Serrano, S. M., S. Beguería, and J. I. López-Moreno, 2010: A multiscalar drought index sensitive to global warming: The standardized precipitation evapotranspiration index. J. Climate, 23, 16961718, https://doi.org/10.1175/2009JCLI2909.1.

    • Search Google Scholar
    • Export Citation
  • Viviroli, D., and R. Weingartner, 2004: The hydrological significance of mountains: From regional to global scale. Hydrol. Earth Syst. Sci., 8, 10171030, https://doi.org/10.5194/hess-8-1017-2004.

    • Search Google Scholar
    • Export Citation
  • Wagner, S., and M. Hoelzle, 2010: Glacier volume changes in the Panj and Vaksh basins. Application of simple parameterizations to estimate past and future glacier change in the Panj and Vakhsh River Basins. Tech. Rep., ITos GmbH, 103 pp.

  • Wang, X., Y. Luo, L. Sun, C. He, Y. Zhang, and S. Liu, 2016: Attribution of runoff decline in the Amu Darya River in central Asia during 1951–2007. J. Hydrometeor., 17, 15431560, https://doi.org/10.1175/JHM-D-15-0114.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., Y. Chen, Z. Li, G. Fang, F. Wang, and H. Liu, 2020: The impact of climate change and human activities on the Aral Sea Basin over the past 50 years. Atmos. Res., 245, 105125, https://doi.org/10.1016/j.atmosres.2020.105125.

    • Search Google Scholar
    • Export Citation
  • Yang, W., N. Wang, A. Chen, J. He, T. Hua, and Y. Qie, 2020: Changes in area and water volume of the Aral Sea in the arid Central Asia over the period of 1960–2018 and their causes. Catena, 191, 104566, https://doi.org/10.1016/j.catena.2020.104566.

    • Search Google Scholar
    • Export Citation
  • Yao, T., and Coauthors, 2012: Different glacier status with atmospheric circulations in Tibetan Plateau and surroundings. Nat. Climate Change, 2, 663667, https://doi.org/10.1038/nclimate1580.

    • Search Google Scholar
    • Export Citation
  • Zhang, L., F. Su, D. Yang, Z. Hao, and K. Tong, 2013: Discharge regime and simulation for the upstream of major rivers over Tibetan Plateau. J. Geophys. Res. Atmos., 118, 85008518, https://doi.org/10.1002/jgrd.50665.

    • Search Google Scholar
    • Export Citation
  • Zhao, Q., and Coauthors, 2019: Projecting climate change impacts on hydrological processes on the Tibetan Plateau with model calibration against the glacier inventory data and observed streamflow. J. Hydrol., 573, 6081, https://doi.org/10.1016/j.jhydrol.2019.03.043.

    • Search Google Scholar
    • Export Citation
  • Zheng, H., L. Zhang, R. Zhu, C. Liu, Y. Sato, and Y. Fukushima, 2009: Responses of streamflow to climate and land surface change in the headwaters of the Yellow River Basin. Water Resour. Res., 45, W00A19, https://doi.org/10.1029/2007WR006665.

    • Search Google Scholar
    • Export Citation
  • Zhiltsov, S. S., I. S. Zonn, O. E. Grishin, V. G. Egorov, and M. S. Ruban, 2018: Transboundary Rivers in Central Asia: Cooperation and conflicts among countries. Water Resources in Central Asia: International Context, S. Zhiltsov et al., Eds., Springer, 61–80.

Supplementary Materials

Save
  • Adenbaev, B. Y., F. H. Hikmatov, and O. A. Kxaydarova, 2015: Hydrological regime of the lower reaches of Amudarja River in the conditions of intensive economical use of water resources. Eur. Sci. Rev, 9–10, 1821.

    • Search Google Scholar
    • Export Citation
  • Agal’tseva, N. A., M. V. Bolgov, T. Y. Spektorman, M. D. Trubetskova, and V. E. Chub, 2011: Estimating hydrological characteristics in the Amu Darya River basin under climate change conditions. Russ. Meteor. Hydrol., 36, 681689, https://doi.org/10.3103/S1068373911100062.

    • Search Google Scholar
    • Export Citation
  • Andreadis, K. M., P. Storck, and D. P. Lettenmaier, 2009: Modeling snow accumulation and ablation processes in forested environments. Water Resour. Res., 45, W05429, https://doi.org/10.1029/2008WR007042.

    • Search Google Scholar
    • Export Citation
  • Armstrong, R. L., and Coauthors, 2019: Runoff from glacier ice and seasonal snow in High Asia: Separating melt water sources in river flow. Reg. Environ. Change, 19, 12491261, https://doi.org/10.1007/s10113-018-1429-0.

    • Search Google Scholar
    • Export Citation
  • Barnett, T. P., J. C. Adam, and D. P. Lettenmaier, 2005: Potential impacts of a warming climate on water availability in snow-dominated regions. Nature, 438, 303309, https://doi.org/10.1038/nature04141.

    • Search Google Scholar
    • Export Citation
  • Brodzik, M. J., and R. Armstrong, 2013: Northern Hemisphere EASE-Grid 2.0 weekly snow cover and sea ice extent, version 4. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 23 September 2020, https://doi.org/10.5067/P7O0HGJLYUQU.

  • Chang, J., H. Zhang, Y. Wang, and Y. Zhu, 2016: Assessing the impact of climate variability and human activities on streamflow variation. Hydrol. Earth Syst. Sci., 20, 15471560, https://doi.org/10.5194/hess-20-1547-2016.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., W. Li, G. Fang, and Z. Li, 2017: Review article: Hydrological modeling in glacierized catchments of central Asia–Status and challenges. Hydrol. Earth Syst. Sci., 21, 669684, https://doi.org/10.5194/hess-21-669-2017.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., Z. Li, G. Fang, and W. Li, 2018: Large hydrological processes changes in the transboundary rivers of Central Asia. J. Geophys. Res. Atmos., 123, 50595069, https://doi.org/10.1029/2017JD028184.

    • Search Google Scholar
    • Export Citation
  • Cherkauer, K. A., and D. P. Lettenmaier, 1999: Hydrologic effects of frozen soils in the upper Mississippi River basin. J. Geophys. Res., 104, 19 59919 610, https://doi.org/10.1029/1999JD900337.

    • Search Google Scholar
    • Export Citation
  • Cuo, L., 2016: Land use/cover change impacts on hydrology in large river basins: A review. Terrestrial Water Cycle and Climate Change: Natural and Human-Induced Impacts, Geophys. Monogr., Vol., 221, Amer. Geophys. Union, 103–134, https://doi.org/10.1002/9781118971772.ch6.

  • Cuo, L., D. P. Lettenmaier, M. Alberti, and J. E. Richey, 2009: Effects of a century of land cover and climate change on hydrology in Puget Sound. Hydrol. Processes, 23, 907933, https://doi.org/10.1002/hyp.7228.

    • Search Google Scholar
    • Export Citation
  • Cuo, L., Y. Zhang, Y. Gao, Z. Hao, and L. Cairang, 2013: The impacts of climate change and land cover transition on the hydrology in the upper Yellow River basin, China. J. Hydrol., 502, 3752, https://doi.org/10.1016/j.jhydrol.2013.08.003.

    • Search Google Scholar
    • Export Citation
  • Dey, P., A. Mishra, 2017: Separating the impacts of climate change and human activities on streamflow: A review of methodologies and critical assumptions. J. Hydrol., 548, 278290, https://doi.org/10.1016/j.jhydrol.2017.03.014.

    • Search Google Scholar
    • Export Citation
  • FAO, 2013: Irrigation in Central Asia in figures. FAO Water Rep. 39, 246 pp., http://www.fao.org/docrep/018/i3289e/i3289e.pdf.

  • FAO, 2021: AQUASTAT core database. Food and Agriculture Organization of the United Nations, accessed 7 September 2021, http://www.fao.org/nr/water/aquastat/data/query/index.html?lang=en.

  • Froebrich, J., O. Olsson, M. Bauer, I. Normatov, and G. Petrov, 2006: Improved dam operation in the Amu Darya Basin including transboundary aspects. Dams and Reservoirs, Societies and Environment in the 21st Century, L. Berga et al., Eds., Taylor and Francis Group, 97–103.

  • Gleick, P., C. Iceland, and A. Trivedi, 2020: Ending conflicts over water: Solutions to water and security challenges. Water, Peace and Security Rep., 128 pp., https://pacinst.org/wp-content/uploads/2020/09/Ending-Conflicts-Over-Water-Pacific-Institute-Sept-2020.pdf.

  • Gupta, H. V., S. Sorooshian, and P. O. Yapo, 1999: Status of automatic calibration for hydrologic models: Comparison with multilevel expert calibration. J. Hydrol. Eng., 4, 135143, https://doi.org/10.1061/(ASCE)1084-0699(1999)4:2(135).

    • Search Google Scholar
    • Export Citation
  • Hagg, W., M. Hoelzle, S. Wagner, E. Mayr, and Z. Klose, 2013: Glacier and runoff changes in the Rukhk catchment, upper Amu-Darya basin until 2050. Global Planet. Change, 110, 6273, https://doi.org/10.1016/j.gloplacha.2013.05.005.

    • Search Google Scholar
    • Export Citation
  • Hock, R., 1999: A distributed temperature-index ice-and snowmelt model including potential direct solar radiation. J. Glaciol., 45, 101111, https://doi.org/10.3189/S0022143000003087.

    • Search Google Scholar
    • Export Citation
  • Hu, Y., W. Duan, Y. Chen, S. Zou, P. M. Kayumba, and N. Sahu, 2021: An integrated assessment of runoff dynamics in the Amu Darya River Basin: Confronting climate change and multiple human activities, 1960–2017. J. Hydrol., 603, 126905, https://doi.org/10.1016/j.jhydrol.2021.126905.

    • Search Google Scholar
    • Export Citation
  • Immerzeel, W. W., A. F. Lutz, and P. Droogers, 2012: Climate change impacts on the upstream water resources of the Amu and Syr Darya River Basins. Future Water Rep. 107, 103 pp., https://www.futurewater.nl/wp-content/uploads/2012/03/Upstream_Report_FW_web.pdf.

  • Istanbulluoglu, E., T. Wang, O. M. Wright, and J. D. Lenters, 2012: Interpretation of hydrologic trends from a water balance perspective: The role of groundwater storage in the Budyko hypothesis. Water Resour. Res., 48, W00H16, https://doi.org/10.1029/2010WR010100.

    • Search Google Scholar
    • Export Citation
  • Jiang, S., L. Ren, B. Yong, V. P. Singh, X. Yang, and F. Yuan, 2011: Quantifying the effects of climate variability and human activities on runoff from the Laohahe basin in Northern China using three different methods. Hydrol. Processes, 25, 24922505, https://doi.org/10.1002/hyp.8002.

    • Search Google Scholar
    • Export Citation
  • Jilili, R., 2019: Study on the impacts of climate change and human activities on runoff in the Amu Darya River, Central Asia (in Chinese). M.S. thesis, Dept. of Geography and Tourism, Xinjiang Normal University, 61 pp.

  • Kendall, M. G., 1975: Rank Correlation Methods. 4th ed. Charles Griffin, 202 pp.

  • Khan, V. M., and L. Holko, 2009: Snow cover characteristics in the Aral Sea Basin from different data sources and their relation with river runoff. J. Mar. Syst., 76, 254262, https://doi.org/10.1016/j.jmarsys.2008.03.012.

    • Search Google Scholar
    • Export Citation
  • Kling, H., M. Fuchs, and M. Paulin, 2012: Runoff conditions in the upper Danube basin under an ensemble of climate change scenarios. J. Hydrol., 424425, 264277, https://doi.org/10.1016/j.jhydrol.2012.01.011.

    • Search Google Scholar
    • Export Citation
  • Konovalov, V. G., 2011: Past and prospective change in state of Central Asian glaciers. Ice Snow, 3, 6068.

  • Konovalov, V. G., and A. S. Shchetinnicov, 1994: Evolution of glaciation in the Pamiro-Alai Mountains and its effect on river runoff. J. Glaciol., 40, 149157, https://doi.org/10.1017/S0022143000003920.

    • Search Google Scholar
    • Export Citation
  • Li, L. J., L. Zhang, H. Wang, J. Wang, J. W. Yang, D. J. Jiang, J. Y. Li, and D. Y. Qin, 2007: Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China. Hydrol. Processes, 21, 34853491, https://doi.org/10.1002/hyp.6485.

    • Search Google Scholar
    • Export Citation
  • Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges, 1994: A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophys. Res., 99, 14 41514 428, https://doi.org/10.1029/94JD00483.

    • Search Google Scholar
    • Export Citation
  • Liang, X., E. F. Wood, and D. P. Lettenmaier, 1996: Surface soil moisture parameterization of the VIC-2L model: Evaluation and modification. Global Planet. Change, 13, 195206, https://doi.org/10.1016/0921-8181(95)00046-1.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., L. Cuo, Q. Li, X. Liu, X. Ma, L. Liang, and J. Ding, 2020a: Impacts of climate change and land use/cover change on streamflow in Beichuan River Basin in Qinghai Province, China. Water, 12, 1198, https://doi.org/10.3390/w12041198.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., Z. Yao, R. Wang, and G. Yu, 2020b: Estimation of the Qinghai-Tibetan plateau runoff and its contribution to large Asian rivers. Sci. Total Environ., 749, 141570, https://doi.org/10.1016/j.scitotenv.2020.141570.

    • Search Google Scholar
    • Export Citation
  • Lohmann, D., E. Raschke, B. Nijssen, and D. P. Lettenmaier, 1998: Regional scale hydrology: I. Formulation of the VIC-2L model coupled to a routing model. Hydrol. Sci. J., 43, 131141, https://doi.org/10.1080/02626669809492107.

    • Search Google Scholar
    • Export Citation
  • Lutz, A. F., W. W. Immerzeel, A. B. Shrestha, and M. F. P. Bierkens, 2014: Consistent increase in High Asia’s runoff due to increasing glacier melt and precipitation. Nat. Climate Change, 4, 587592, https://doi.org/10.1038/nclimate2237.

    • Search Google Scholar
    • Export Citation
  • Mann, H. B., 1945: Nonparametric tests against trend. Econometrica, 13, 245259, https://doi.org/10.2307/1907187.

  • McKee, T. B., N. J. Doeskin, and J. Kleist, 1993: The relationship of drought frequency and duration to time scales. 8th Conf. on Applied Climatology, Boston, MA, Amer. Meteor. Soc., 179–184.

  • Micklin, P., 2007: The Aral Sea disaster. Annu. Rev. Earth Planet. Sci., 35, 4772, https://doi.org/10.1146/annurev.earth.35.031306.140120.

    • Search Google Scholar
    • Export Citation
  • Micklin, P., 2016: The future Aral Sea: Hope and despair. Environ. Earth Sci., 75, 844, https://doi.org/10.1007/s12665-016-5614-5.

  • Moriasi, D. N., J. G. Arnold, M. W. Van Liew, R. L. Bingner, R. D. Harmel, and T. L. Veith, 2007: Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Trans. ASABE, 50, 885900, https://doi.org/10.13031/2013.23153.

    • Search Google Scholar
    • Export Citation
  • Nash, J. E., and J. V. Sutcliffe, 1970: River flow forecasting through conceptual models: Part 1. A discussion of principles. J. Hydrol., 10, 282290, https://doi.org/10.1016/0022-1694(70)90255-6.

    • Search Google Scholar
    • Export Citation
  • Olsson, O., M. Gassmann, K. Wegerich, and M. Bauer, 2010: Identification of the effective water availability from streamflows in the Zerafshan river basin, Central Asia. J. Hydrol., 390, 190197, https://doi.org/10.1016/j.jhydrol.2010.06.042.

    • Search Google Scholar
    • Export Citation
  • Pettitt, A. N., 1979: A non-parametric approach to the change point problem. J. Appl. Stat., 28, 126135, https://doi.org/10.2307/2346729.

    • Search Google Scholar
    • Export Citation
  • Pritchard, H. D., 2019: Asia’s shrinking glaciers protect large populations from drought stress. Nature, 569, 649654, https://doi.org/10.1038/s41586-019-1240-1.

    • Search Google Scholar
    • Export Citation
  • RGI Consortium, 2014: Randolph Glacier Inventory-A dataset of global glacier outlines, version 4.0. National Snow and Ice Data Center, accessed 1 December 2014, https://doi.org/10.7265/bse7-r330.

  • Sankarasubramanian, A., R. M. Vogel, and J. F. Limbrunner, 2001: Climate elasticity of streamflow in the United States. Water Resour. Res., 37, 17711781, https://doi.org/10.1029/2000WR900330.

    • Search Google Scholar
    • Export Citation
  • Savoskul, O. S., and V. Smakhtin, 2013: Glacier systems and seasonal snow cover in six major Asian river basins: Hydrological role under changing climate. International Water Management Institute Research Rep. 150, 45 pp., https://www.iwmi.cgiar.org/2013/07/glacier-systems-and-seasonal-snow-cover-in-six-major-asian-river-basins-hydrological-role-under-changing-climate/.

  • Schaake, J. C., 1990: From climate to flow. Climate Change and U.S. Water Resources, P. E. Waggoner, Ed., John Wiley, 177–206.

  • Sen, P. K., 1968: Estimates of regression coefficient based on Kendall’s tau. J. Amer. Stat. Assoc., 63, 13791389, https://doi.org/10.1080/01621459.1968.10480934.

    • Search Google Scholar
    • Export Citation
  • Sorg, A., T. Bolch, M. Stoffel, O. Solomina, and M. Beniston, 2012: Climate change impacts on glaciers and runoff in Tien Shan (Central Asia). Nat. Climate Change, 2, 725731, https://doi.org/10.1038/nclimate1592.

    • Search Google Scholar
    • Export Citation
  • Stulina, G., and O. Eshchanov, 2013: Climate change impacts on hydrology and environment in the Pre-Aral region. Quat. Int., 311, 8796, https://doi.org/10.1016/j.quaint.2013.07.015.

    • Search Google Scholar
    • Export Citation
  • Su, Y., and Coauthors, 2021: High agricultural water consumption led to the continued shrinkage of the Aral Sea during 1992–2015. Sci. Total Environ., 777, 145993, https://doi.org/10.1016/j.scitotenv.2021.145993.