Abstract

The ongoing cascading hydropower exploitation in southwestern China has been the subject of debate and conflict in recent years. This study aims to assess the climate change impacts on the hydropower system of Yuan River and to quantify the future potential in operation optimization of Gasa-Madushan (GS-MDS) Reservoir system. The Community Climate System Model version 4 (CCSM4) projections are bias-corrected and downscaled to drive the Soil Water Assessment Tool hydrological model, aiming to predict the climate and runoff changes for the future. Then, an adaptive operation chart model of cascaded reservoirs is established to balance hydropower generation and ecological requirements under climate change. In the future, the decadal average temperature and annual average precipitation will possibly increase by 0.80–2.22 °C and 2.56–4.65%, respectively; the monthly average runoff may increase by 6.89%, 6.17%, and 18.26% for GS Reservoir Basin, and by 8.89%, 8.14%, and 23.14% for MDS Reservoir Basin under Representative Concentration Pathways 2.6, 4.5, and 8.5, respectively. The adaptive operation chart results in a reduction of 52.66–70.77% in the total water shortage at a cost of 2.09–4.54% decrease in total power generation of the GS-MDS cascaded hydropower system compared to that of non-adaptive operation chart.

INTRODUCTION

Extensive cascaded hydropower exploitation in southwestern China has been the subject of debate and conflict in recent years. The cascaded reservoirs' construction plan and operation policy were proposed based on the analysis of historical climate and hydrological observations. However, global mean annual temperature has increased by 0.8 °C since 1880 and future warming is almost certain according to the Intergovernmental Panel on Climate Change Fifth Assessment Report (IPCC AR5) (Flato et al. 2013). The variation in climate may change the balance of global water circulation and energy cycle, resulting in changes in the magnitude and spatial-temporal distribution of runoff, and even unforeseen changes in agriculture, environment, ecology, urbanization, and many other social-economic systems (Nam et al. 2015). Yuan River is a typical mountainous basin in southwestern China, and considering the huge influences of hydropower systems, it is vital to give a reliable picture of long-term climate change effects and the adaptive potential.

The general circulation models (GCMs) under the Coupled Model Intercomparison Project Phase 5 (CMIP5) are essential tools for simulating future changes in climate (Karl & Trenberth 2003). Several GCMs have been widely applied to predict the hydrological processes and assess the impact of climate changes. Xu & Xu (2012) assessed the performances of simulating the present climate over China based on 18 global climate models' simulations during 1961–2005, and demonstrated models can capture the dominant features of the geographic distributions of temperature and precipitation compared with observations. Among all GCMs under CMIP5, the Community Climate System Model version 4 (CCSM4) is a commonly used GCM consisting of atmosphere, land, ocean, and sea ice components linked through a coupler that exchanges state information and fluxes between the components. Its simulation ability has been confirmed in both global and regional scale, including southwestern China. For example, DeFlorio et al. (2013) demonstrated that the climatological precipitation is more realistic in CCSM4, both in spatial structure and seasonal agreement with observations. Shi et al. (2016) pointed out that although small biases exist, the spatial distributions of Asian surface temperatures and precipitation rates are simulated well by CCSM4. Zhang et al. (2017) indicated that CCSM4 performs well in simulating the spatial distribution of surface air temperature climatology and observational precipitation products in China. As well, the CCSM4 output can model temperature and most precipitation indices, including extreme precipitation (Liu et al. 2015). Climate projections are often used to drive hydrological models for the prediction of runoff responses in the future (Nerantzaki et al. 2015; Neupane & Kumar 2015). Distributed hydrological models have been particularly embraced by researchers, including the MIKE SHE model (Moussoulis et al. 2015), Distributed Hydrology Soil-Vegetation model (Du et al. 2016), Variable Infiltration Capacity model (Leng et al. 2016), and Soil Water Assessment Tool (SWAT) model (Serpa et al. 2015).

Water resource management is facing major challenges to balance conflicting objectives in reservoir operation given the uncertainties associated with climate change (Raje & Mujumdar 2010), which has stimulated the search for new adaptive operation rules to mitigate possible negative impacts of climate changes. Zhou & Guo (2013) proposed an integrated optimization model to develop operation rule curves of Danjiangkou Reservoir in a base period and three future periods under the A2 scenario, one of the four scenarios proposed by the Intergovernmental Panel on Climate Change in its Special Report on Emission Scenarios (IPCC SRES). Alvarez et al. (2014) developed a reservoir management tool to investigate the adaptation strategies of three reservoirs in the Lièvre River watershed to climate change under A2 scenario. Ahmadi et al. (2015) established an adaptive model to revise reservoir operating rules for Karoon-4 Reservoir in Iran as an adaptive strategy to climate under A1, A2, B1, and B2 scenarios. Currently, most of the studies in southwestern China are based on CMIP3 projections. The latest CMIP phase 5 (CMIP5) GCMs have advanced our knowledge of climate variability and climate change based on the previous version CMIP3. Considering the huge influences of hydropower systems in southwestern China, a more reliable picture is required for long-term climate change effects and adaptive potential assessment. In similar studies using CMIP5 data, Vliet et al. (2016) developed a coupled hydrological-electricity modeling framework to propose six adaptation options of a reservoir under Representative Concentration Pathways (RCPs) 2.6 and 8.5. Carvalho-Santos et al. (2017) studied the two-reservoir adaptation to water supply problems of a mountain catchment in northeast Portugal under RCPs 4.5 and 8.5. However, most of these studies were qualitative analyses and only gave the adaptive options. It has been rarely explored to establish an optimal operation model to propose the reservoir adaptive policies.

On the basis of previous studies, the purpose of this study is to update the understanding of climate change impacts on hydropower systems and to quantify the future potential in reservoir operation optimization. A systematic framework is proposed for climate change assessment and adaption, including bias correction, hydrological simulation, and adaptive operation. The future runoff responses to climate change are simulated and predicted based on bias-corrected GCM outputs. Finally, the adaptive operation chart model is developed to balance hydropower generation and ecological requirements of the cascaded reservoirs' system under climate change.

STUDY AREA AND DATA COLLECTION

Study area

Yuan River is located in southwestern China (east: 100.01°–105.67°, north: 22.45°–25.53°) with a total length of 692 km and a drainage area of 34,629 km2. This river has several tributaries, including Lixian River, Tengtiao River, Nanxi River, Panlong River, and Nanli River. Principal flow regulation capabilities within the basin are provided by two main cascaded reservoirs, namely, Gasa (GS) Reservoir and Madushan (MDS) Reservoir (Figure 1). MDS Reservoir is located in the downstream of GS Reservoir. Currently, GS and MDS Reservoirs are mainly utilized for hydropower generation, flood control, and ecological conservation. The main parameters of these two reservoirs are listed in Table 1. The conventional operating charts of GS and MDS Reservoirs are shown in Figure 2. There are five zones in the operation chart of GS Reservoir, including the reduced output zone (power output defined as 0.5Nr), guaranteed output zone (Nr), 1.5 times increased output zone (1.5Nr), 2 times increased output zone (2Nr), and full capacity output zone (Nrr), from the bottom to the top. There are three zones in the operation chart of the seasonal MDS Reservoir, including the reduced output zone, guaranteed output zone, and full capacity output zone from the bottom to the top.

Table 1

The main characteristics of the GS-MDS reservoir system

Reservoir Regulation ability Normal storage volume Dead storage volume Power output coefficient Installed capacity Guaranteed capacity 
108 m3 104 kW 
GS Annual 14.9 6.69 8.6 27.0 8.0 
MDS Seasonal 4.82 2.32 8.8 28.8 7.9 
Reservoir Regulation ability Normal storage volume Dead storage volume Power output coefficient Installed capacity Guaranteed capacity 
108 m3 104 kW 
GS Annual 14.9 6.69 8.6 27.0 8.0 
MDS Seasonal 4.82 2.32 8.8 28.8 7.9 
Figure 1

The location and major tributaries of Yuan River.

Figure 1

The location and major tributaries of Yuan River.

Figure 2

Operation charts of GS and MDS reservoirs. Convention operation charts of (a) GS and (b) MDS Reservoir; adaptive operation charts of (c) GS and (d) MDS Reservoir.

Figure 2

Operation charts of GS and MDS reservoirs. Convention operation charts of (a) GS and (b) MDS Reservoir; adaptive operation charts of (c) GS and (d) MDS Reservoir.

Data collection

Geospatial, runoff, meteorological, and climate data were collected in this study. Geospatial data included digital elevation model (DEM), land use and soil information, where the DEM data produced by the United States Geological Survey (USGS) (100*100 m) were used for defining stream, area, and boundary of sub-basins; land use data were developed by the Chinese Academy of Sciences (westdc_lucc_1km_China.asc) in 2000 with a spatial resolution of 1 km soil data in 30 arc-second were derived from the Harmonized World Soil Database 1.2 (Fischer et al. 2008). The historical meteorological data were used to analyze the past climate features of Yuan River Basin, to calibrate the hydrological model, and to downscale the GCM outputs to a finer resolution. The observed variables included daily precipitation, maximum and minimum air temperature, solar radiation, wind speed, and relative humidity, which were collected at six meteorological stations including 56,768, 56,856, 56,875, 56,966, 56,985, and 56,968 (id number defined by China Meteorological Administration) from 1981 to 2007. The locations of these stations are shown in Figure 1. For the future climate projections, we analyzed the daily precipitation, temperature, maximum and minimum near-surface air temperature for the period 2015–2100 based on CCSM4. Then, the future projected temperature and precipitation of CCSM4 were bias corrected and spatial downscaled using BCSDd method. The monthly runoff data collected at the entry of the GS and MDS Reservoirs from January 1981 to December 2007 were used for the calibration and validation of the SWAT model. The future monthly runoff simulated from the SWAT model during the period of 2015–2100 as the inflow of GS-MDS Reservoir was applied to force the operation chart model.

METHODOLOGY

Climate change scenarios

IPCC AR5 has introduced a new way of developing scenarios. These scenarios span the range of plausible radiative forcing scenarios, and are called RCPs (Takemura 2012). There are four RCP scenarios, including RCP2.6, RCP4.5, RCP6.0, and RCP8.5, which are defined and named according to their total radiative forcing in 2100 relative to pre-industrial values. Here, RCPs 2.6, 4.5, and 8.5 were used to reflect the low, medium, and high forcing levels, respectively, as shown in Table 2.

Table 2

Overview of RCPs 2.6, 4.5, and 8.5

Scenario Radiative forcing 
RCP2.6 Peak in radiative forcing at ∼3 W/m² before 2100 and decline 
RCP4.5 Stabilization without overshoot pathway to 4.5 W/m² at stabilization after 2100 
RCP8.5 Rising radiative forcing pathway leading to 8.5 W/m² in 2100 
Scenario Radiative forcing 
RCP2.6 Peak in radiative forcing at ∼3 W/m² before 2100 and decline 
RCP4.5 Stabilization without overshoot pathway to 4.5 W/m² at stabilization after 2100 
RCP8.5 Rising radiative forcing pathway leading to 8.5 W/m² in 2100 

Bias correction and spatial disaggregation daily (BCSDd) method downscaling

When comparing historical simulation results from GCM to observations, comparison often shows that simulations tend to be biased wet/dry, cool/warm, with biases varying by location, season, and variable (Allen & Ingram 2002; Dibike & Coulibaly 2005). To reflect the regional climate change features, the BCSDd method (Thrasher et al. 2012) was performed to establish empirical relationships between GCM-resolution climate variables and local climate and to reproduce the regional climate features. BCSDd consists of two main steps, including bias correction and spatial downscaling: (1) in bias correction, a quantile mapping of daily temperature and precipitation from GCMs to observations re-gridded to the coarse model resolution is used to identify and remove bias (Panofsky & Brier 1958). Bias correction matches the statistical moments of observations and GCM output covering a common time period (e.g., 20th century), and accordingly adjusts for biases in GCM output for projected time periods (e.g., 21st century) by assuming a constant model bias; (2) in spatial disaggregation, daily bias-corrected precipitation and temperature are spatially disaggregated to the fine-resolution grid by interpolation using the synographic mapping system (SYMAP) algorithm (Hietala & Larson 1979), and then applying fine-resolution spatial anomaly patterns derived from the observations. The anomalies calculated as the correction factors between the re-gridded observations and coarsened projected GCM outputs were spatially interpolated to the fine-resolution grid; for precipitation, the fine-resolution daily observations multiplied by the correction factors are calculated as the fine-resolution daily simulations and for temperature, the fine-resolution daily observations plus the correction factors are the fine-resolution daily simulations.

SWAT hydrological modeling

The Soil and Water Assessment Tool (SWAT) was used for this study. Several studies of climate change impacts on water resources have successfully used the SWAT model and the ability of SWAT to predict streamflow has been widely verified around the world. For example, Zhang et al. (2016) applied SWAT to estimate streamflows in the Xin River Basin, China, based on climate change scenarios under three RCPs. Awan et al. (2016) employed SWAT to assess the impact of climate change on consumptive water use in Lower Chenab Canal area, Indus basin. Tamm et al. (2016) modeled future changes in climate on the North-Estonian hydropower production using SWAT. Notably, the SWAT model has yielded high accuracy for long-term simulations of yearly and monthly mean streamflow (Githui et al. 2009; Chen et al. 2011; Liang et al. 2017). Therefore, SWAT can be used for long-term (2015–2100) monthly streamflow prediction in this paper. SWAT is a basin-scale continuous hydrological model. It divides a catchment into several sub-basins with different micro-climatic conditions, and each sub-basin is partitioned into hydrological response units (HRUs) based on land use, soil type, and relief (Nerantzaki et al. 2015). The water flow in each HRU is simulated based on the following water budget formula:  
formula
(1)
where is the soil water content on day t (mm); is the initial soil water content (mm); is the rainfall depth (mm); is the surface runoff (mm); is the evapotranspiration (mm); is the amount of water entering the vadose zone from the soil (mm); is the base flow (mm); and i denotes the ith day, respectively.

SWAT requires as input hydro-meteorological data, a land-cover map, a soil map, and a DEM; after data compilation, ArcSWAT version 510 (Mehdi et al. 2015) running on an ArcGIS 9.3.1 platform was used for watershed delineation and sub-basin discretization. In Yuan River Basin, 107 sub-basins were delimited and then divided into multiple HRUs according to the land cover, soil types, and slope classes, as shown in Figure 3. The slope of the Yuan River Basin with a range from 0 to 30% accounts for more than 90% of the whole basin. Thus, due to the catchment's rather flat topography, single slope option was chosen to simplify the basin slope into two categories: 0–1% and 1–99.99%.

Figure 3

Soil, land use, and slope classes defined for Yuan River Basin: (a) soil; (b) land use; (c) slope.

Figure 3

Soil, land use, and slope classes defined for Yuan River Basin: (a) soil; (b) land use; (c) slope.

The SWAT Calibration and Uncertainty Program (SWATCUP) enables sensitivity analysis, calibration, validation, and uncertainty analysis of the SWAT model (Arnold et al. 2012), and Sequential Uncertainty Fitting 2 (SUFI2) of SWATCUP was used for this study (Abbaspour et al. 2007). The global sensitivity analysis integrated within SUFI2 was applied to test 21 SWAT hydrologic parameters, for surface runoff simulation in parallel with the calibration procedure, shown in Table 3. The derived new parameter values obtained from calibration and confirmation analyses were incorporated with the SWAT database for further simulations. The coefficients of determination (R2) (Herman et al. 2015) and Nash–Sutcliffe efficiency (NSE) (Nerantzaki et al. 2015) were used to evaluate model performance, respectively. The hydrological simulation was considered to be acceptable if R2 > 0.5 or NSE> 0.5 mean.

Table 3

Results of parameter calibration using the SWAT model

Parameter Description Min Max Optimal 
CN2 Curve number for moisture condition II 35.00 98.00 65.17 
GW_DELAY Groundwater delay times (days) 0.00 245.34 222.23 
SFTMP Snowfall temperature (°C) −19.11 6.98 −15.80 
SMTMP Snow melt base temperature (°C) −9.98 10.11 8.37 
SMFMX Maximum melt rate for snow during the year 4.95 14.99 11.30 
SMFMN Minimum melt rate for snow during the year 0.00 12.65 1.23 
TIMP Snow pack temperature lag factor 0.00 0.59 0.17 
GW_REVAP Groundwater ‘revap’ coefficient 0.02 0.12 0.08 
REVAPMN Threshold depth of water in the shallow aquifer for revap (mm) 136.14 408.86 195.26 
HRU_SLP Manning's n value for overland flow 0.00 0.55 0.53 
ALPHA_BF Baseflow alpha factor 0.14 0.71 0.65 
ESCO Soil evaporation compensation factor 0.00 1.00 0.95 
EPCO Plant uptake compensation factor 0.00 1.00 1.00 
CH_N2 Manning's n value for the main channel 0.00 0.00 0.014 
CH_K2 Effective hydraulic conductivity in main channel alluvium (mm/h) 0.00 300.00 78.59 
ALPHA_BNK Baseflow alpha factor for bank storage 0.00 1.00 0.37 
SOL_K Saturated hydraulic conductivity (mm/h) 0.00 1.00 0.35 
SOL_AWC Baseflow alpha factor (mm/mm) 0.00 1.00 0.15 
SLSUBBSN Average slope length 10.00 99.05 35.68 
Parameter Description Min Max Optimal 
CN2 Curve number for moisture condition II 35.00 98.00 65.17 
GW_DELAY Groundwater delay times (days) 0.00 245.34 222.23 
SFTMP Snowfall temperature (°C) −19.11 6.98 −15.80 
SMTMP Snow melt base temperature (°C) −9.98 10.11 8.37 
SMFMX Maximum melt rate for snow during the year 4.95 14.99 11.30 
SMFMN Minimum melt rate for snow during the year 0.00 12.65 1.23 
TIMP Snow pack temperature lag factor 0.00 0.59 0.17 
GW_REVAP Groundwater ‘revap’ coefficient 0.02 0.12 0.08 
REVAPMN Threshold depth of water in the shallow aquifer for revap (mm) 136.14 408.86 195.26 
HRU_SLP Manning's n value for overland flow 0.00 0.55 0.53 
ALPHA_BF Baseflow alpha factor 0.14 0.71 0.65 
ESCO Soil evaporation compensation factor 0.00 1.00 0.95 
EPCO Plant uptake compensation factor 0.00 1.00 1.00 
CH_N2 Manning's n value for the main channel 0.00 0.00 0.014 
CH_K2 Effective hydraulic conductivity in main channel alluvium (mm/h) 0.00 300.00 78.59 
ALPHA_BNK Baseflow alpha factor for bank storage 0.00 1.00 0.37 
SOL_K Saturated hydraulic conductivity (mm/h) 0.00 1.00 0.35 
SOL_AWC Baseflow alpha factor (mm/mm) 0.00 1.00 0.15 
SLSUBBSN Average slope length 10.00 99.05 35.68 

Adaptive operation chart model of cascaded reservoirs

An adaptive operation chart model consisting of objective functions and constraints was proposed in this study to estimate the long-term potential in hydropower system optimization with multiple conflicting objectives in a changing environment. The GS and MDS are multi-objective reservoirs, but the long-term functions including water supply and entertainment are associated with large uncertainty, as little reliable information is available now. Therefore, we mainly considered the power production and ecological conservation as two main objectives of the GS-MDS hydropower system. The objective function was to maximize energy production with a penalty function for violation of ecological requirements, as shown in Equation (2). The reservoir storage was considered as the decision variable:  
formula
(2)
where E is the total power production of the cascaded reservoirs' system (kW·h); n is the index of the reservoir, ; M is the number of the reservoir, ; is the output power coefficient of the nth hydropower station; is the average water head of the hydropower station at time step t (m), ; is the outflow of the nth reservoir at time step t (m3/s); is the minimum ecological flow of the hydropower station at time step t (m3/s); and is the time step (s).
The penalty function pun was calculated using Equation (3). pun was not a constant, but increased as the ecological runoff shortage increased:  
formula
(3)
where is the penalty factor, which is generally regarded as a large number; and is the outflow of the hydropower station at time step t (m3/s).

The Tennant method (Tennant 1975) was used to determine the minimum ecological flow of the downstream area of the GS and MDS reservoirs. It is one of the classical hydrological methods developed to compute environmental flow. The Tennant method is simple as it requires no field work and is based on a single hydrologic statistic (mean annual discharge) (Palmer et al. 2008), which can be employed to calculate future ecological flow with predicted runoff under climate change. The commonly used threshold varied from 10% (poor or minimum condition) to 60–100% (optimum range) (Tharme 2003). Based on the investigation of key aquatic species of Yuan River Basin and their preference analysis (Wen et al. 2016), 60% of monthly average runoff was identified as the minimum ecological flow requirement.

The adaptive operation charge model of cascaded reservoirs was subjected to the following constraints:

  • (1)
    Water balance constraint:  
    formula
    (4)
  • (2)
    Water storage constraint:  
    formula
    (5)
  • (3)
    Outflow constraint:  
    formula
    (6)
  • (4)
    Turbine release constraint:  
    formula
    (7)
  • (5)
    Power output constraint:  
    formula
    (8)
    where m is the index of reservoir or hydropower station, ; t is the index of time step; w and q are the inflow and outflow of reservoir (m3/s), respectively; S is the storage of reservoir (m3); I is the water loss of reservoir (m3); u is the turbine release of hydropower station (m3/s); N is the power output of hydropower station (KW); min and max are the minimum and maximum limits of variables, respectively. The values of model inputs are all shown in Table 4.

Dynamic programming successive approximation was applied to solve the chart model (Shi et al. 2015). The monthly guiding level of operation chart was taken as the decision variable.

Table 4

The values of the model constraints

Constraints Water storage value (108 m3)
 
Outflow value (m3/s)
 
Turbine release value (m3/s)
 
Power output value (104 kW)
 
Min Max Min Max Min Max Min Max 
GS 6.7 14.9 1,400 270 27.0 
MDS 2.3 4.8 3,000 490 28.8 
Constraints Water storage value (108 m3)
 
Outflow value (m3/s)
 
Turbine release value (m3/s)
 
Power output value (104 kW)
 
Min Max Min Max Min Max Min Max 
GS 6.7 14.9 1,400 270 27.0 
MDS 2.3 4.8 3,000 490 28.8 

RESULTS AND DISCUSSION

BCSD downscaling performance assessment

The CCSM4 temperature and precipitation future projections were bias corrected and spatial downscaled to the grid scale of 0.5° × 0.5° using the BCSDd method. Figure 4 shows the downscaled CCSM4 projected temperature and precipitation for the future. Figure 5 indicates the spatial distribution of deviation of historical GCM simulations and BCSDd downscaled results to climate observations. The bias in CCSM4 simulated precipitation is mainly distributed in 10–40 mm. After downscaling, the variation of bias decreases to 5–20 mm. The improvements in the temperature downscaling are more significant. The variation of bias is 1–4 °C in CCSM4 simulations and 0.0–0.6 °C after downscaling.

Figure 4

The multi-year average of downscaled CCSM4 projections during 2015–2100. Downscaled precipitation under (a) RCP2.6; (b) RCP4.5; (c) RCP8.5. Ddownscaled temperature under (d) RCP2.6; (e) RCP4.5; (f) RCP8.5.

Figure 4

The multi-year average of downscaled CCSM4 projections during 2015–2100. Downscaled precipitation under (a) RCP2.6; (b) RCP4.5; (c) RCP8.5. Ddownscaled temperature under (d) RCP2.6; (e) RCP4.5; (f) RCP8.5.

Figure 5

The spatial distribution of bias of historical CCSM4 simulations and BCSDd downscaled simulations of Yuan River Basin. Bias in (a) CCSM4 simulated precipitation; (b) downscaled precipitation; (c) CCSM4 simulated temperature; (d) downscaled temperature.

Figure 5

The spatial distribution of bias of historical CCSM4 simulations and BCSDd downscaled simulations of Yuan River Basin. Bias in (a) CCSM4 simulated precipitation; (b) downscaled precipitation; (c) CCSM4 simulated temperature; (d) downscaled temperature.

Calibration and validation of the SWAT model

The SWAT model was calibrated and validated using the inflow of GS and MDS Reservoirs for the period 1981–2000 and 2001–2007, respectively. The optimal value of parameters used for the calibration procedures are shown in Table 3. A significant correlation is observed between observed and simulated runoff with R2 > 0.75 and NSE> 0.75 (before removing seasonal cycle) with R2 > 0.4 and NSE> 0.4 (after removing seasonal cycle), as shown in Figure 6 and Table 5. The simulated and observed monthly average flow are 143.06 and 143.66 m3/s for GS Reservoir, and 295.16 and 296.44 m3/s for MDS Reservoir, respectively. However, most of the high flows (>500 m3/s for GS Reservoir and >1,000 m3/s for MDS Reservoir) are underestimated by 10–50%. The calibration strategy prioritizes the overall simulation accuracy, also the high flow simulation is always the weak point of the SWAT model (Ning et al. 2012).

Table 5

The performance of SWAT model in the hydrological simulation

Items Basin Calibration
 
Validation
 
R2 NSE R2 NSE 
Before removing seasonal cycle GS Reservoir 0.83 0.83 0.78 0.77 
MDS Reservoir 0.84 0.83 0.75 0.75 
After removing seasonal cycle GS Reservoir 0.62 0.59 0.56 0.53 
MDS Reservoir 0.62 0.57 0.54 0.51 
Items Basin Calibration
 
Validation
 
R2 NSE R2 NSE 
Before removing seasonal cycle GS Reservoir 0.83 0.83 0.78 0.77 
MDS Reservoir 0.84 0.83 0.75 0.75 
After removing seasonal cycle GS Reservoir 0.62 0.59 0.56 0.53 
MDS Reservoir 0.62 0.57 0.54 0.51 
Figure 6

The simulated runoff results at the entry of (a) GS Reservoir and (b) MDS Reservoir.

Figure 6

The simulated runoff results at the entry of (a) GS Reservoir and (b) MDS Reservoir.

Future changes in precipitation and temperature

The annual average precipitation of the Yuan River Basin is 1,102.86 mm for the period 1981–2007, and is expected to increase by 2.56–4.65% for the period 2015–2100. Figure 7(a) shows the future changes in decadal average projected precipitation under three RCP scenarios based on CCSM4 of the Yuan River Basin from 2015 to 2100 with respect to that from 1981 to 2007. After 2030, the annual precipitation may exceed the historical average level, and increases with increasing strength of the radiative forcing at a rate ranging from 5.83 mm/10a (RCP2.6) to 11.05 mm/10a (RCP4.5) and then to 18.15 mm/10a (RCP8.5).

Figure 7

Future changes in (a) decadal and (b) monthly average projected precipitation of Yuan River Basin from 2015 to 2100 compared with that from 1981 to 2007.

Figure 7

Future changes in (a) decadal and (b) monthly average projected precipitation of Yuan River Basin from 2015 to 2100 compared with that from 1981 to 2007.

Figure 7(b) shows the future changes in monthly average projected precipitation. The monthly precipitation could increase by 0.01–11.1 mm in spring (MAM), summer (JJA) and October. In particular, the Yuan River Basin may experience more serious flood events in the wet period in the future, since the monthly precipitation is expected to rise by 7.82–11.11 mm in June, the main flood period of the basin. Meanwhile, a decrease in the monthly precipitation is observed in winter months (DJF), September and November. The most significant reduction in monthly rainfall (2.56–4.06 mm) is observed in September, indicating that the Yuan River Basin may become dryer in the dry period in the future, and the transition between flood and drought may become faster and more obvious. In addition, the change in monthly precipitation is not sensitive to the radiation intensity.

Figure 8 shows the future changes in decadal and monthly average projected temperature based on CCSM4 of the Yuan River Basin from 2015 to 2100, respectively. The decadal average temperature is expected to increase by 0.80–2.22 °C in the future. Not surprisingly, the increasing rate ranges from 0.18 °C/10a in RCP2.6 to 1.33 °C/10a in RCP8.5. The monthly average temperature will also increase in a similar way, and the largest increase (0.80–2.50 °C) is detected in April and May.

Figure 8

Future changes in (a) decadal and (b) monthly average projected temperature of Yuan River Basin from 2015 to 2100 compared with that from 1981 to 2007.

Figure 8

Future changes in (a) decadal and (b) monthly average projected temperature of Yuan River Basin from 2015 to 2100 compared with that from 1981 to 2007.

Hydrologic responses modeling

The climate projections were employed to drive the calibrated SWAT hydrological model to predict the runoff responses to climate changes of the Yuan River Basin in the future. The upstream area of GS Reservoir is unregulated. The inflow of MDS Reservoir is composed of the release of GS Reservoir and the runoff flowing into the GS-MDS river reach. Figure 9 shows the decadal mean simulated average runoff change under RCPs 2.6, 4.5, and 8.5 based on CCSM4 of the GS and MDS Reservoir Basin during the period of 2015–2100, respectively. Generally, the upward variation trend is detected in the future decadal mean runoff. For the GS Reservoir Basin, the decadal average discharge is expected to increase by 2.93%, 5.36%, and 10.02% under RCPs 2.6, 4.5, and 8.5, respectively, compared to the historical baseline of 140.20 m3/s. For the MDS Reservoir Basin, the future average runoff is expected to increase by 0.79%, 4.04%, and 8.31% under RCPs 2.6, 4.5, and 8.5, respectively, compared to the historical mean discharge of 292.62 m3/s. In addition, the stream flow will be reduced after the 2060s under RCPs 2.6 and 4.5.

Figure 9

Future changes in decadal average simulated runoff at (a) GS Reservoir Basin and (b) MDS Reservoir Basin from 2015 to 2100 compared with that from 1981 to 2007.

Figure 9

Future changes in decadal average simulated runoff at (a) GS Reservoir Basin and (b) MDS Reservoir Basin from 2015 to 2100 compared with that from 1981 to 2007.

The boxplots of monthly mean simulated runoff under RCPs based on CCSM4 of the GS and MDS Basin from 2015 to 2100 compared with that from 1981 to 2007 are shown in Figures 10 and 11, respectively. Generally, the differences between wet and dry periods become more pronounced. The runoff in the flood season (June–September) will increase by 16.52–36.87% and 29.10–55.03% for the GS and MDS Basin, respectively. However, the dry period is expected to be much drier than before, since the monthly runoff will decrease by 8.41–15.72% and 22.60–29.63% for the GS and MDS Basin, respectively. The largest reduction is observed in October and November. The range of the monthly runoff variation increases as the radiative forcing intensity increases from RCP2.6 to RCP8.5, that is, the high forcing level may increase the possibility and the intensity of flood and drought. In addition, the temporal distribution of runoff is not sensitive to the variation of radiation intensity. These findings are consistent with the results in future precipitation analysis. The largest reduction is observed between September (the ending of wet period) and October (the beginning of dry period). For the GS Basin, October is decreased by 41.15–65.29% compared with September, while there is only 9.30% reduction for the historical scenario. For the MDS Basin, October is decreased by 52.31–69.08% compared with September, while there is only 15.86% reduction for the historical scenario. The same situation occurs between May (the ending of dry period) and June (the beginning of wet period). It can be concluded that the transition between flood and drought may become faster and more obvious in the future.

Figure 10

Boxplots of monthly average simulated runoff of GS Reservoir Basin: (a) historical condition; (b) RCP2.6; (c) RCP4.5; and (d) RCP8.5.

Figure 10

Boxplots of monthly average simulated runoff of GS Reservoir Basin: (a) historical condition; (b) RCP2.6; (c) RCP4.5; and (d) RCP8.5.

Figure 11

Boxplots of monthly average simulated runoff of MDS Reservoir Basin: (a) historical condition; (b) RCP2.6; (c) RCP4.5; and (d) RCP8.5.

Figure 11

Boxplots of monthly average simulated runoff of MDS Reservoir Basin: (a) historical condition; (b) RCP2.6; (c) RCP4.5; and (d) RCP8.5.

Adaptive operation of GS-MDS reservoirs

Figure 2 shows the adaptive operation chart of GS and MDS reservoirs generated from adaptive operation chart model under climate change. It indicates new regulation policies that allow for a better adaptation to the future flow regime under climate change for cascaded reservoirs' system in Yuan River Basin. Compared with the convention operation chart, the adaptive charts reduce the guiding level to produce more energy output for both GS and MDS reservoirs. Figure 12 shows the monthly adaptive operation results of cascaded reservoirs' system guided with the operation chart. It can be seen that GS Reservoir and MDS Reservoir have similar storage with different inflow under three scenarios, respectively. GS Reservoir and MDS Reservoir operate smoothly throughout the year. The only exception is that the monthly storage is higher in September of MDS Reservoir, which has small storage volume, due to the lesser outflow than inflow in this period.

Figure 12

Monthly adaptive operation results for the period 2015–2100 of (a) GS Reservoir and (b) MDS Reservoir under RCP2.6; (c) GS Reservoir and (d) MDS Reservoir under RCP4.5; (e) GS Reservoir and (f) MDS Reservoir under RCP8.5.

Figure 12

Monthly adaptive operation results for the period 2015–2100 of (a) GS Reservoir and (b) MDS Reservoir under RCP2.6; (c) GS Reservoir and (d) MDS Reservoir under RCP4.5; (e) GS Reservoir and (f) MDS Reservoir under RCP8.5.

The adaptive operation chart model without ecological penalty function was used (termed as nonpenalty-adaptive scenario) to explore the tradeoffs between hydropower and ecological conservation in the future. Future runoff following current operation rules was also simulated as the baseline condition (termed as non-adaptive scenario). Table 6 shows the results of non-adaptive (historical convention), nonpenalty-adaptive, and adaptive operation, respectively. For the future, the adaptive operation technique could significantly improve the ecological condition of the Yuan River Basin by reducing 52.66–70.77% of the total ecological streamflow shortage, at a cost of 2.09–4.54% decrease in total power generation of the GS-MDS cascaded hydropower system compared to that of the non-adaptive operation model. More water resources are allocated to satisfy the ecological demands, and the total spills and average storage are reduced by 5.81–13.18% and 1.54–8.63%, respectively, compared to the baseline scenario. In addition, the average storage changes by −5.72 to 3.98% for GS Reservoir, and is reduced by 19.24 to 22.96% for MDS Reservoir. Compared to the nonpenalty-adaptive model that prioritizes the hydropower objective, severe ecological deterioration could be avoided in the adaptive operation model, where the ecological runoff shortage is reduced by 71.16–83.01% at a cost of 11–13.97% decrease in the total power generation amount.

Table 6

Comparison of non-adaptive, nonpenalty-adaptive, and adaptive operation results under operation chart for the period 2015–2100

  RCP2.6
 
RCP4.5
 
RCP8.5
 
GS MDS Total GS MDS Total GS MDS Total   
Total power generation (108 kW·h) Non-adaptive 1,100.74 1,141.93 2,242.67 1,122.52 1,185.30 2,307.82 1,136.89 1,198.52 2,335.41 
Nonpenalty-adaptive 1,184.70 1,255.34 2,440.04 1,215.09 1,293.15 2,508.24 1,231.83 1,299.42 2,531.25 
Adaptive 1,121.53 1,019.39 2,140.92 1,206.54 1,053.08 2,259.62 1,221.75 1,056.96 2,278.71 
Total ecological water shortage (108 m3Non-adaptive 118.96 437.19 556.15 87.57 437.19 524.77 95.99 404.79 500.78 
Nonpenalty-adaptive 250.16 677.00 927.17 236.44 645.30 881.75 230.43 631.01 861.44 
Adaptive 41.81 221.46 263.27 0.00 172.34 172.34 0.00 146.35 146.35 
Total spill (108 m3Non-adaptive 462.56 1,797.28 2,259.84 474.19 1,919.98 2,394.17 824.58 2,898.27 3,722.84 
Nonpenalty-adaptive 236.64 1,682.81 1,919.45 290.78 1,835.23 2,126.01 574.71 2,782.78 3,357.50 
Adaptive 236.42 1,725.64 1,962.06 354.51 1,900.67 2,255.18 573.86 2,850.24 3,424.10 
Average storage (108 m3Non-adaptive 12.11 3.33 – 12.17 3.40 – 12.18 3.40 – 
Nonpenalty-adaptive 13.05 4.55 – 13.29 4.56 – 13.48 4.57 – 
Adaptive 11.42 2.69 – 12.28 2.62 – 12.67 2.67 – 
  RCP2.6
 
RCP4.5
 
RCP8.5
 
GS MDS Total GS MDS Total GS MDS Total   
Total power generation (108 kW·h) Non-adaptive 1,100.74 1,141.93 2,242.67 1,122.52 1,185.30 2,307.82 1,136.89 1,198.52 2,335.41 
Nonpenalty-adaptive 1,184.70 1,255.34 2,440.04 1,215.09 1,293.15 2,508.24 1,231.83 1,299.42 2,531.25 
Adaptive 1,121.53 1,019.39 2,140.92 1,206.54 1,053.08 2,259.62 1,221.75 1,056.96 2,278.71 
Total ecological water shortage (108 m3Non-adaptive 118.96 437.19 556.15 87.57 437.19 524.77 95.99 404.79 500.78 
Nonpenalty-adaptive 250.16 677.00 927.17 236.44 645.30 881.75 230.43 631.01 861.44 
Adaptive 41.81 221.46 263.27 0.00 172.34 172.34 0.00 146.35 146.35 
Total spill (108 m3Non-adaptive 462.56 1,797.28 2,259.84 474.19 1,919.98 2,394.17 824.58 2,898.27 3,722.84 
Nonpenalty-adaptive 236.64 1,682.81 1,919.45 290.78 1,835.23 2,126.01 574.71 2,782.78 3,357.50 
Adaptive 236.42 1,725.64 1,962.06 354.51 1,900.67 2,255.18 573.86 2,850.24 3,424.10 
Average storage (108 m3Non-adaptive 12.11 3.33 – 12.17 3.40 – 12.18 3.40 – 
Nonpenalty-adaptive 13.05 4.55 – 13.29 4.56 – 13.48 4.57 – 
Adaptive 11.42 2.69 – 12.28 2.62 – 12.67 2.67 – 

Figure 13 shows the monthly hydropower production and ecological runoff shortage of GS and MDS reservoirs for the future. The adaptive monthly hydropower production of GS Reservoir is expected to increase by 31.60, 21.30, and 25.90 million kW·h during May–September and decrease by 16.00, 14.20, and 9.40 million kW·h during October–April under RCPs 2.6, 4.5, and 8.5, respectively. The monthly hydropower production of the MDS hydropower station will decrease by 13.30, 14.60, and 15.70 million kW·h under RCPs 2.6, 4.5, and 8.5, respectively. The only exception is that the monthly hydropower production increases by about 14.50 million kW·h in October due to the increasing requirement of the ecological streamflow in this period.

Figure 13

Monthly power generation of adaptive and non-adaptive operation models for the period 2015–2100 under (a) RCP2.6; (c) RCP4.5; (e) RCP8.5 for GS hydropower station and (b) RCP2.6; (d) RCP4.5; (f) RCP8.5 for MDS hydropower station.

Figure 13

Monthly power generation of adaptive and non-adaptive operation models for the period 2015–2100 under (a) RCP2.6; (c) RCP4.5; (e) RCP8.5 for GS hydropower station and (b) RCP2.6; (d) RCP4.5; (f) RCP8.5 for MDS hydropower station.

Figure 14 shows the monthly ecological runoff shortage of GS and MDS reservoirs for the future. Under current rules, runoff shortage may occur in the downstream area of GS Reservoir in July–October in the future. However, after adaptive operation, it could be reduced by 64.81% under RCP 2.6 and totally eliminated under RCPs 4.5 and 8.5. However, the problem of ecological runoff shortage in the downstream area of MDS Reservoir is most pronounced in October, but less severe in January–May. Correspondingly, the adaptive ecological water shortage will be reduced by 55.60%, 65.20%, and 66.20% in October and 43.60%, 35.30%, and 33.80% from January to May under RCPs 2.6, 4.5, and 8.5, respectively.

Figure 14

Monthly ecological runoff requirement and shortage of adaptive and non-adaptive operation models for the period 2015–2100 under (a) RCP2.6; (c) RCP4.5; (e) RCP8.5 for GS hydropower station and (b) RCP2.6; (d) RCP4.5; (f) RCP8.5 for MDS hydropower station.

Figure 14

Monthly ecological runoff requirement and shortage of adaptive and non-adaptive operation models for the period 2015–2100 under (a) RCP2.6; (c) RCP4.5; (e) RCP8.5 for GS hydropower station and (b) RCP2.6; (d) RCP4.5; (f) RCP8.5 for MDS hydropower station.

CONCLUSIONS

Global warming is altering the global climate and hydrological cycle, thus the cascaded hydropower system needs to be changed correspondingly to better adapt to the new environment. This study aims to assess the climate change impacts on the hydropower system of the Yuan River and to quantify the future potential in operation optimization of Gasa-Madushan (GS-MDS) Reservoir system. A systematic framework is proposed for climate change assessment and adaption including bias correction, hydrological simulation, and adaptive operation. The SWAT model was applied to simulate the hydrological process, and then the BCSDd downscaled CCSM4 projections were used to drive the calibrated SWAT model to predict the future (2015–2100) runoff under RCPs 2.6, 4.5, and 8.5, respectively. An adaptive operation chart model was proposed to improve the hydropower system performance and quantify the potential optimization under climate change. Compared with the baseline condition (1981–2007), the warming and wetting trend was detected across the Yuan River with 0.80–2.22 °C increase in annual average temperature and 2.56–4.65% growth in annual precipitation. Also, the monthly average runoff may increase by 2.93%, 5.36%, and 10.02% for the GS Reservoir Basin, and by 0.79%, 4.04%, and 8.31% for the MDS Reservoir Basin under RCPs 2.6, 4.5, and 8.5, respectively. The Yuan River Basin may experience more serious flood events in the wet period and drought events in the dry period, and the transition between flood and drought may become faster and more obvious in the future. For the future, the adaptive operation technique could significantly improve the ecological condition of the Yuan River Basin by reducing 52.66–70.77% of the ecological streamflow shortage, at a cost of 2.09–4.54% decrease in total power generation of the GS-MDS cascaded hydropower system compared to that of the non-adaptive operation model.

ACKNOWLEDGEMENTS

This research is funded by National Natural Science Foundation of China (U1765201, 51609061), the Fundamental Research Funds for the Central Universities (2018B11314), National Key R&D Program of China (2018YFC0407902), the CRSRI Open Research Program (CKWV2016370/KY), Open Fund Research of State Key Laboratory of Hydraulics and Mountain River Engineering (SKHL1621), the Jiangsu Province Ordinary University Graduate Student Research Innovation Project (2016B05327), a project funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). We would like to thank the United States Geological Survey (USGS), Chinese Academy of Sciences (CAS), International Union of Soil Sciences (IUSS) and China Metrological Data Sharing Service for providing geospatial and metrological data. Also, we acknowledge the World Climate Research Program's Working Group on Coupled Modeling and the climate modeling groups for producing and making available their model output.

REFERENCES

REFERENCES
Abbaspour
K. C.
,
Yang
J.
,
Maximov
I.
,
Siber
R.
,
Bogner
K.
,
Mieleitner
J.
,
Zobrist
J.
&
Srinivasan
R.
2007
Modelling hydrology and water quality in the pre-alpine/alpine Thur watershed using SWAT
.
Journal of Hydrology
333
(
2
),
413
430
.
Ahmadi
M.
,
Haddad
O. B.
&
Loáiciga
H. A.
2015
Adaptive reservoir operation rules under climatic change
.
Water Resources Management
29
(
4
),
1247
1266
.
Arnold
J.
,
Moriasi
D.
,
Gassman
P.
,
Abbaspour
K.
,
White
M.
,
Srinivasan
R.
,
Santhi
C.
,
Harmel
R.
,
Van Griensven
A.
&
Van Liew
M.
2012
SWAT: Model use, calibration, and validation
.
Transactions of the ASABE
55
(
4
),
1491
1508
.
Carvalho-Santos
C.
,
Monteiro
A. T.
,
Azevedo
J. C.
,
Honrado
J. P.
&
Nunes
J. P.
2017
Climate change impacts on water resources and reservoir management: uncertainty and adaptation for a mountain catchment in northeast Portugal
.
Water Resources Management
31
(
11
),
3355
3370
.
Chen
L.
,
Liang
Z.
&
Zhu
J.
2011
Application of SWAT model to runoff simulation in Huangyahe Basin
.
Water Resources and Power
10
,
8
11
.
Deflorio
M. J.
,
Pierce
D. W.
,
Cayan
D. R.
&
Miller
A. J.
2013
Western U.S. extreme precipitation events and their relation to ENSO and PDO in CCSM4
.
Journal of Climate
26
(
12
),
4231
4243
.
Fischer
G.
,
Nachtergaele
F.
,
Prieler
S.
,
Van Velthuizen
H.
,
Verelst
L.
&
Wiberg
D.
2008
Global Agro-Ecological Zones Assessment for Agriculture (GAEZ 2008)
.
IIASA, Laxenburg, Austria and FAO
,
Rome
,
Italy
.
Flato
G.
,
Marotzke
J.
,
Abiodun
B.
,
Braconnot
P.
,
Chou
S. C.
,
Collins
W. J.
,
Cox
P.
,
Driouech
F.
,
Emori
S.
&
Eyring
V.
2013
Evaluation of climate models
.
In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
Climate Change
5
,
741
866
.
Githui
F.
,
Gitau
W.
,
Mutua
F.
&
Bauwens
W.
2009
Climate change impact on SWAT simulated streamflow in western Kenya
.
International Journal of Climatology
29
(
12
),
1823
1834
.
Herman
M. R.
,
Nejadhashemi
A. P.
,
Daneshvar
F.
,
Ross
D. M.
,
Woznicki
S. A.
,
Zhang
Z.
&
Esfahanian
A.-H.
2015
Optimization of conservation practice implementation strategies in the context of stream health
.
Ecological Engineering
84
,
1
12
.
Hietala
H. J.
&
Larson
P. A.
1979
SYMAP analyses in archaeology: intrasite assumptions and a comparison with TREND analysis
.
Norwegian Archaeological Review
12
(
1
),
57
64
.
Karl
T. R.
&
Trenberth
K. E.
2003
Modern global climate change
.
Science
302
(
5651
),
1719
1723
.
Leng
G.
,
Tang
Q.
,
Huang
S.
,
Zhang
X.
&
Cao
J.
2016
Assessments of joint hydrological extreme risks in a warming climate in China
.
International Journal of Climatology
36
(
4
),
1632
1642
.
Liang
Z.
,
Tang
T.
,
Li
B.
,
Liu
T.
,
Wang
J.
&
Hu
Y.
2017
Long-term streamflow forecasting using SWAT through the integration of the random forests precipitation generator: case study of Danjiangkou Reservoir
.
Hydrology Research
DOI: 10.2166/nh.2017.085
.
Liu
C.
,
Hofstra
N.
&
Leemans
R.
2015
Preparing suitable climate scenario data to assess impacts on local food safety
.
Food Research International
68
,
31
40
.
Mehdi
B.
,
Ludwig
R.
&
Lehner
B.
2015
Evaluating the impacts of climate change and crop land use change on streamflow, nitrates and phosphorus: a modeling study in Bavaria
.
Journal of Hydrology: Regional Studies
4
,
60
90
.
Moussoulis
E.
,
Mallinis
G.
,
Koutsias
N.
&
Zacharias
I.
2015
Modelling surface runoff to evaluate the effects of wildfires in multiple semi-arid, shrubland-dominated catchments
.
Hydrological Processes
29
(
20
),
4427
4441
.
Nerantzaki
S.
,
Giannakis
G.
,
Efstathiou
D.
,
Nikolaidis
N.
,
Sibetheros
I.
,
Karatzas
G.
&
Zacharias
I.
2015
Modeling suspended sediment transport and assessing the impacts of climate change in a karstic Mediterranean watershed
.
Science of the Total Environment
538
,
288
297
.
Ning
J. C.
,
Liu
G. H.
,
Yu
Y. E.
,
Liu
Q. S.
&
Xie
C. J.
2012
Study on precipitation parameters input of SWAT model
.
Journal of Natural Resources
27
(
5
),
866
875
.
Palmer
M. A.
,
Reidy Liermann
C. A.
,
Nilsson
C.
,
Flörke
M.
,
Alcamo
J.
,
Lake
P. S.
&
Bond
N.
2008
Climate change and the world's river basins: anticipating management options
.
Frontiers in Ecology and the Environment
6
(
2
),
81
89
.
Panofsky
H. A.
&
Brier
G. W.
1958
Some Applications of Statistics to Meteorology
.
Mineral Industries Extension Services, College of Mineral Industries, Pennsylvania State University
,
USA
.
Raje
D.
&
Mujumdar
P.
2010
Reservoir performance under uncertainty in hydrologic impacts of climate change
.
Advances in Water Resources
33
(
3
),
312
326
.
Serpa
D.
,
Nunes
J.
,
Santos
J.
,
Sampaio
E.
,
Jacinto
R.
,
Veiga
S.
,
Lima
J.
,
Moreira
M.
,
Corte-Real
J.
&
Keizer
J.
2015
Impacts of climate and land use changes on the hydrological and erosion processes of two contrasting Mediterranean catchments
.
Science of the Total Environment
538
,
64
77
.
Shi
Z.
,
Cheng
J.
,
Yang
S.
,
Chang
B.
&
Jiang
Y.
2015
Research on methods of optimal water resources allocation for single reservoir-single pump station system using DPSA
.
Journal of Irrigation and Drainage
34
(
2
),
37
40
.
Tamm
O.
,
Luhamaa
A.
&
Tamm
T.
2016
Modeling future changes in the North-Estonian hydropower production by using SWAT
.
Hydrology Research
47
(
4
),
835
846
.
Thrasher
B.
,
Maurer
E. P.
,
Mckellar
C.
&
Duffy
P. B.
2012
Technical note: bias correcting climate model simulated daily temperature extremes with quantile mapping
.
Hydrology & Earth System Sciences
16
(
9
),
3309
3314
.
Vliet
M. T. H. v.
,
Wiberg
D.
,
Leduc
S.
&
Riahi
K.
2016
Power-generation system vulnerability and adaptation to changes in climate and water resources
.
Nature Climate Change
6
,
375
380
.
Wen
X.
,
Fang
G.-h.
,
Guo
Y.-x.
&
Zhou
L.
2016
Adapting the operation of cascaded reservoirs on Yuan River for fish habitat conservation
.
Ecological Modelling
337
,
221
230
.
Xu
Y.
&
Xu
C.
2012
Preliminary assessment of simulations of climate changes over China by CMIP5 multi-models
.
Atmospheric and Oceanic Science Letters
5
(
6
),
489
494
.