Reservoir Inflow Research Articles

Strategic bidding for a hydro-wind-photovoltaic hybrid system considering the profit beyond forecast time

Hydro-wind-photovoltaic hybrid systems gain profit by bidding in the forecast lead-time. However, the literature focuses on bidding strategy to maximize current profits, while the future utilities (beyond forecast time) for hybrid systems are generally neglected. To address the research gap, the study proposed an integrated bidding strategy for a hydro-wind-photovoltaic hybrid system with a trade-off between current profits and future utilities considered. Steps are as follows: (1) ensemble forecasts are used within the forecast lead-time, within historical data input as stochastic scenarios beyond forecast time, (2) functional forms between the profit beyond forecast time and the reservoir carryover storage, expected price, inflow, wind and photovoltaic power are fitted, and (3) a two-layer nested bidding model is to maximize the total-profit. With China's Sichuan province as the market, the Jinping-I hybrid system was selected as the case study. Results show that: (1) profits beyond forecast time can be quantified as a linear function with R-squared for 10-day above 0.6, and (2) the optimal bidding strategy raises the profit by 0.45% while reduces water-consumption by 10.3% compared with conventional strategy. The framework is effective in gaining more profit with lower risk for hybrid systems.

Early Forecasting Hydrological and Agricultural Droughts in the Bouregreg Basin Using a Machine Learning Approach

Water supply for drinking and agricultural purposes in semi-arid regions is confronted with severe drought risks, which impact socioeconomic development. However, early forecasting of drought indices is crucial in water resource management to implement mitigation measures against its consequences. In this study, we attempt to develop an integrated approach to forecast the agricultural and hydrological drought in a semi-arid zone to ensure sustainable agropastoral activities at the watershed scale and drinking water supply at the reservoir scale. To that end, we used machine learning algorithms to forecast the annual SPEI and we embedded it into the hydrological drought by implementing a correlation between the reservoir’s annual inflow and the annual SPEI. The results showed that starting from December we can forecast the annual SPEI and so the annual reservoir inflow with an NSE ranges from 0.62 to 0.99 during the validation process. The proposed approach allows the decision makers not only to manage agricultural drought in order to ensure pastoral activities “sustainability at watershed scale” but also to manage hydrological drought at a reservoir scale.

Watershed memory amplified the Oroville rain-on-snow flood of February 2017.

Mountain snowpacks are transitioning to experience less snowfall and more rainfall as the climate warms, creating more persistent low- to no-snow conditions. This precipitation shift also invites more high-impact rain-on-snow (ROS) events, which have historically yielded many of the largest and most damaging floods in the western United States. One such sequence of events preceded the evacuation of 188,000 residents below the already-damaged Oroville Dam spillway in February 2017 in California's Sierra Nevada. Prior studies have suggested that snowmelt during ROS dramatically amplified reservoir inflows. However, we present evidence that snowmelt may have played a smaller role than previously documented (augmenting terrestrial water inputs by 21%). A series of hydrologic model experiments and subdaily snow, soil, streamflow, and hydrometeorological measurements demonstrate that direct, "passive" routing of rainfall through snow, and increasingly efficient runoff driven by gradually wetter soils can alternatively explain the extreme runoff totals. Our analysis reveals a crucial link between frequent winter storms and a basin's hydrologic response-emphasizing the role of soil moisture "memory" of within-season storms in priming impactful flood responses. Given the breadth in plausible ROS flood mechanisms, this case study underscores a need for more detailed measurements of soil moisture along with in-storm changes to snowpack structure, extent, energy balance, and precipitation phase to address ROS knowledge gaps associated with current observational limits. Sharpening our conceptual understanding of basin-scale ROS better equips water managers moving forward to appropriately classify threat levels, which are projected to increase throughout the mid-21st century.

Comparative Study of Coupling Models of Feature Selection Methods and Machine Learning Techniques for Predicting Monthly Reservoir Inflow

Effective reservoir operation under the effects of climate change is immensely challenging. The accuracy of reservoir inflow forecasting is one of the essential factors supporting reservoir operations. This study aimed to investigate coupling models of feature selection (FS) and machine learning (ML) algorithms to predict the monthly reservoir inflow. The study was carried out using data from the Huai Nam Sai reservoir in southern Thailand. Eighteen years of monthly recorded data (i.e., reservoir inflow, reservoir storage, rainfall, and regional climate indices) with up to a 12-month time lag were utilized. Three ML techniques, i.e., multiple linear regression (MLR), support vector regression (SVR), and artificial neural network (ANN)were compared in their capabilities. In addition, two FS techniques, i.e., genetic algorithm (GA) and backward elimination (BE) methods, were studied with four predictable time intervals, consisting of 3, 6, 9, and 12 months in advance. Ten-fold cross-validation was used for model evaluation. Study results revealed that FS methods (i.e., GA and BE) Could improve the performance of SVR and ANN for predicting monthly reservoir inflow forecasting, but they have no effects on MLR. Different developed forecasting models were suitable for different reservoir inflow forecasting time-step-ahead. BE-ANN provided the best performance for three-time-ahead (T + 3) and nine-time-ahead (T + 9) by giving an OI of 0.9885 and 0.8818, NSE of 0.9546 and 0.9815, RMSE of 1.3155 and 1.2172 MCM/month, MAE of 0.9568 and 0.9644 MCM/month, and r of 0.9796 and 0.9804, respectively. The GA-ANN model showed the highest prediction accuracy for six-time-ahead (T + 6), with an OI of 0.8997, NSE of 0.9407, RMSE of 2.1699 MCM/month, MAE of 1.7549 MCM/month, and r of 0.9759. The ANN model showed the best prediction accuracy for twelve-time-ahead (T + 12), with an OI of 0.9515, NSE of 0.9835, RMSE of 1.1613 MCM/month, MAE of 0.9273 MCM/month, and r of 0.9835.

Flow regime changes in the Lancang River, revealed by integrated modeling with multiple Earth observation datasets

The flow regime change of rivers, especially transboundary rivers, affected by reservoir regulations is evident worldwide and has received much attention. Investigating dam-induced flow regime alterations is essential for understanding potential adverse downstream effects and facilitating dialogue around coordinated water use in transboundary basins, such as the Lancang River Basin (LRB). This study explored the value of combining several types of satellite Earth observation (EO) datasets that monitor different water balance components to constrain the parameter space of lumped conceptual hydrological models. Thus, we aimed to reconstruct the natural flow regimes upstream and downstream of the cascade reservoirs. Specifically, reservoir water storage changes were first estimated using satellite imagery and altimetry datasets. Then, storage changes were combined with hydrological model simulations of reservoir inflow to estimate the regulated flow regime downstream. Our results showed that integrated hydrological modeling combined with EO datasets exhibited better overall performance. Continuous warming and drying of the LRB resulted in a decrease in discharge of approximately 47 %. By comparing the simulated natural and regulated flow regimes, we revealed the pivotal role of the Xiaowan and Nuozhadu reservoirs in regulating natural flows. The wet season shortens (approximately 45 days), the flood peak flattens, and the low flow in the dry season has primarily increases. The two reservoirs attenuated 50 % of the flood peaks in the wet seasons and mitigated droughts by releasing up to 100 % of the natural flows in the dry seasons at the China–Laos border. Overall, these results enhance the understanding of upper reservoir operation, and the approaches can be applied to studies of dammed basins under climate change scenarios when knowledge of the upstream area is limited.

Dynamics of Reservoir Environment Carrying Capacity (Case of Kedungombo Reservoir, Central Java, Indonesia)

The development of dams or reservoirs requires continuous attention. The dynamics of rain as the primary natural resource and the water utilization may change from the design. Climate change is thought to have caused changes in reservoir inflow patterns where the patterns may change throughout the service life of the reservoir. The way of the utilization of catchment areas and reservoir inundation is also very dynamic, along with the dynamics of water demand which is generally aimed at obtaining better human welfare. At the same time, such activities may affect the quality of the reservoir waters, followed by disturbances in the water utilization for various purposes. This paper presents the results of evaluating the carrying capacity of the reservoir environment based on the data obtained from reservoir sedimentation and water monitoring of Kedungombo Reservoir. The adopted method of the environmental carrying capacity of the Kedungombo Reservoir considers the four environment-related carrying capacities, i.e., reservoir sedimentation, water utilization, recreation, and aquaculture. The term overall carrying capacity is introduced, adopting three approximations of weighting factor α of the four parameters. The analysis results show a promising method to evaluate the reservoir environment carrying capacity that could be dynamics during its service life.

Utilizing deep learning machine for inflow forecasting in two different environment regions: a case study of a tropical and semi-arid region

Reservoir inflow (Qflow) forecasting is one of the crucial processes in achieving the best water resources management in a particular catchment area. Although physical models have taken place in solving this problem, those models showed a noticeable limitation due to their requirements for huge efforts, hydrology and climate data, and time-consuming learning process. Hence, the recent alternative technology is the development of the machine learning models and deep learning neural network (DLNN) is the recent promising methodology explored in the field of water resources. The current research was adopted to forecast Qflow at two different catchment areas characterized with different type of inflow stochasticity, (semi-arid and topical). Validation against two classical algorithms of neural network including multilayer perceptron neural network (MLPNN) and radial basis function neural network (RBFNN) was elaborated and discussed. The research was further investigated the potential of the feature selection algorithm “genetic algorithm (GA)”, for identifying the appropriate predictors. The research finding confirmed the feasibility of the developed DLNN model for the investigated two case studies. In addition, the DLNN model confirmed its capability in solving daily scale Q more accurately in comparison with the monthly scale. The applied GA as feature selection algorithm was reduced the dimension and complexity of the learning process of the applied predictive model. Further, the research finding approved the adequacy of the data span used in the current investigation development of computerized ML algorithm.

Two-step daily reservoir inflow prediction using ARIMA-machine learning and ensemble models

The reservoirs play a crucial role in the development of civilisation as they facilitate the storage of water for multiple purposes like hydroelectric power generation, flood control, irrigation, and drinking water etc. In order to effectively meet these multiple purposes, the knowledge of the inflow in the reservoir is essential. Apart from the historical data, future prediction of the inflows is also necessary specially in context of climate change. A two-step algorithm for the prediction of reservoir inflow to enable meticulous planning and execution of daily reservoir operation keeping the historical variation of inflow in account has been proposed. The developed algorithm takes into account the patterns in the historic inflow data using the time series analysis along with the variability in the climatic patterns using the different predictors in the machine learning model. The first step uses time series model, ARIMA method to forecast the monthly inflows, which are then used as the targets in the second step for the month-wise daily forecasting of the inflows using the two types of ensemble models, namely, averaging and boosting models in machine learning. The test results show that for both the monthly models and daily models the NRMSE and NMAE values were low for the monsoon periods compared to the non-monsoon periods. The averaging ensemble models were found to perform better than the boosting ensemble models for maximum number of months. The yearly results show an error of less than 5% between actual and predicted values for all the test cases, showing the precision in the developed algorithm. Further, the uncertainty analysis shows that the prediction done using the weighted average of the different inflow scenarios performs better than the prediction against the single inflow scenario.

Evaluation of Nominal Energy Storage at Existing Hydropower Reservoirs in the US

AbstractLong‐term planning and operation of hydropower reservoirs require an understanding of both water and energy storage. As energy storage needs of the evolving grid increase, we must account for the water and energy storage potential of these reservoirs. Given the limitations of current data on existing hydropower, we compile statistics related to storage volume and hydraulic head from publicly available data sets and examine differences in descriptions of US hydropower storage. Assembled characteristics are used to calculate nominal energy storage capacity, a simple measure of potential to generate power from a given volume of water, not factoring in detailed constraints. Inventory‐based estimates of energy storage are calculated at 2,075 dams, which helps put the potential for US hydropower to support energy storage in context with similar evaluations in other regions and with other energy storage technologies. The national energy storage capacity ranges between 34.5 and 45.1 TWh depending on the information used, with 52% of energy storage located at the 10 largest reservoirs in the US. Energy storage capacities are also calculated at 236 dams with historical volume and elevation data. Finally, reservoir inflows provide context for the storage volumes and sensitivities to hydrologic variability. Larger reservoirs with greater storage volume to inflow ratios are concentrated in the Western US, but the majority of hydropower reservoirs store less than the annual inflow. We address several infrastructure and water resource informatics challenges and highlight remaining issues, including representing seasonal or shorter variability in water volumes and representing connected hydropower facilities.

A SVM-based implicit stochastic joint scheduling method for ‘wind-photovoltaic-cascaded hydropower stations’ systems

With the gradual expansion of the development scale of wind power and photovoltaic (PV) power plants, the multi-energy complementary power generation system, typically represented by hydro-PV/hydro-wind/hydro-wind-PV, has become an important part of modern power systems. Aiming at the joint operation of the cascaded hydropower stations after wind-PV grid connection, a medium- and long-term implicit stochastic joint dispatching function model for wind-PV-cascaded hydropower stations based on the SVM(support vector machine) method is developed in this paper, which selects the final water levels of the reservoirs as the dependent variables, and the initial water levels of the reservoirs, the reservoir inflow, the interval inflow as well as the wind and PV output are independent variables. First, the optimization of main parameters C (Penalty coefficient), g (Kernel function parameter) and p (Insensitive loss coefficient) of the model are achieved by particle swarm algorithm. The Gaussian radial basis function is then used to fit the scheduling function proposed in this paper. Finally, the rolling simulation calculation and correction of the obtained scheduling function are realized by C# programming language of VS2017 platform. The results show that the proposed scheduling function is an effective method for scheduling decision-making, and the revised water level process, output process as well as annual electricity production of the scheduling model are not significantly different from the optimal scheduling results. Moreover, the simulation results conform to the existing scheduling rules, which has shown it can be used to inform the operation of cascaded hydropower stations under the multi-energy complementary system.

Impacts of climate change on streamflow and reservoir inflows in the Upper Manyame sub-catchment of Zimbabwe

This study focused on the Upper Manyame sub-catchment which covers an area of approximately 3 786 km2 and forms part of the Manyame catchment, one of the seven catchments of Zimbabwe. Manyame catchment has its source in Marondera town and drains into the Zambezi River downstream of the Kariba Dam and upstream of the Cahora Bassa Dam, in the northern part of the country. This study assessed potential climate change impacts on the streamflow and reservoir inflows in the Upper Manyame sub-catchment. Hydrologic simulations for future climate (2030s and 2060s) were carried out using statistically downscaled bias-corrected variables from the HadCM3 (HadCM3A2a and HadCM3B2a scenarios) and CanESM2 (RCP2.6 and RCP8.5) global circulation models. The HEC–HMS hydrological model was set up for two gauged micro-catchments and eight ungauged tributary micro-catchments. Model calibration for gauged micro-catchments of Upper Manyame over the period from 2000–2010 revealed satisfactory model performance of 4.3% (RVE) and 0.1 (bias) for Mukuvisi micro-catchment and 9.5% (RVE) and 0.15 (bias) for Marimba micro-catchment. Model simulations resulted in a projected decrease in streamflow by 7.4–26.4% for HadCM3. For CanESM2, simulations resulted in a projected decrease in streamflow by 2.5–34.7%. Reservoir inflows into Lake Chivero and Lake Manyame, the main water supply sources for Harare, will decrease by 10.5–18% for HadCM3 and by 8–33.6% for CanESM2.

Extreme events in the European renewable power system: Validation of a modeling framework to estimate renewable electricity production and demand from meteorological data

With the need to reduce greenhouse gas emissions, the coming decades will see a transition of Europe's power system, currently mainly based on fossil fuels towards a higher share of renewable sources. Increasing effects of fluctuations in electricity production and demand as a result of meteorological variability might cause compound events with unforeseen impacts. We constructed and validated a modeling framework to examine such extreme impact events on the European power system. This framework includes six modules: i) a reservoir hydropower inflow and ii) dispatch module; iii) a run-of-river hydropower production module; iv) a wind energy production module; v) a photovoltaic solar energy production model; and vi) an electricity demand module. Based on ERA5 reanalysis input data and present-day capacity distributions, we computed electricity production and demand for a set of European countries in the period 2015–2021 and compared results to observed data. The model captures the variability and extremes of wind, photovoltaic and run-of-river production well, with correlations between modelled and observed data for most countries of more than 0.87, 0.68 and 0.65 respectively. The hydropower dispatch module also functions well, with correlations up to 0.82, but struggles to capture reservoir inflows and operating procedures of some countries. A case study into the meteorological drivers of extreme events in Sweden and Spain showed that the meteorological conditions during extreme events selected by the model and extracted from observational data are similar, giving confidence in the application of the modeling framework for (future changes in) extreme event analysis.

Improving Short-range Reservoir Inflow Forecasts with Machine Learning Model Combination

Improving Short-range Reservoir Inflow Forecasts with Machine Learning Model Combination

A model coupling current non-adjustable, coming adjustable and remaining stages for daily generation scheduling of a wind-solar-hydro complementary system

To formulate the daily generation scheduling of a wind-solar-hydro complementary system (WSHCS), the hourly forecasts of the reservoir inflow, wind speed, and sunlight intensity within a day and their longer prediction can be used to assess the benefit of the day-ahead and remaining stages, respectively. Thus, the current two-stage model coupling these forecasts can efficiently maximize the total benefits of day-ahead and remaining stages. However, the two-stage model does not consider the significant difference between non-adjustable and adjustable periods for hydropower. Namely, the hydro unit commitment should be decided in the current stage (non-adjustable period), and it can be adjusted in the coming stage (adjustable period). To address this issue, a three-stage model, including the day-ahead non-adjustable stage, the day-ahead adjustable stage, and the remaining period stage, is proposed to formulate a daily generation scheduling of the WSHCS. The Guandi wind-solar-hydro hybrid power plant on China's Yalong River is selected as a case study. Results show that compared with the two-stage model, the proposed three-stage operation model can increase the average energy production of the WSHCS by 2.23 GWh for typical days, and the average power curtailment rate reduces by 0.26%. A three-layer nested approach can simplify from 576 to 26 decision variables for the high-dimensional three-stage model, and improve the computational efficiency. Thus, the proposed method could guide the daily generation scheduling of a WSHCS.

Daily reservoir inflow forecasting using weather forecast downscaling and rainfall-runoff modeling: Application to Urmia Lake basin, Iran

Study regionThis study develops the first daily runoff forecast system for Bukan reservoir in Urmia Lake basin (ULB), Iran, a region suffering from water shortages and competing water demands. Study focusA weather forecast downscaling model is developed for downscaling large-scale raw weather forecasts of ECMWF and NCEP to small-scale spatial resolutions. Various downscaling methods are compared, including deterministic Artificial Intelligence (AI) techniques and a Bayesian Belief Network (BBN). Downscaled precipitation and temperature forecasts are then fed into a rainfall-runoff model that accounts for daily snow and soil moisture dynamics in the sub-basins upstream of Bukan reservoir. The multi-objective Particle Swarm Optimization (MOPSO) method is used to estimate hydrological model parameters by maximizing the simulation accuracy of observed river flow (NSEQ) and the logarithm of river flow (NSELogQ) in each sub-basin. New hydrological insights for the regionResults of the weather forecast downscaling model show that the accuracy of the BBN is greater than the various deterministic AI methods tested. Calibration results of the rainfall-runoff model indicate no significant trade-off between fitting daily high and low flows, with an average NSEQ and NSELogQ of 0.43 and 0.63 for the calibration period, and 0.54 and 0.57 for the validation period. The entire forecasting system was evaluated using inflow observations for years 2020 and 2021, resulting in an NSE of 0.66 for forecasting daily inflow into Bukan reservoir. The inflow forecasts can be used by policymakers and operators of the reservoir to optimize water allocation between agricultural and environmental demands in the ULB.

Water level–driven agricultural nonpoint source pollution dominated the ammonia variation in China's second largest reservoir

Rainfall-runoff and water flooding are the driving mechanisms of agricultural nonpoint source pollution (ANPSP), but existing research has hardly focused on water level–driven ANPSP. Danjiangkou Reservoir was the second largest reservoir in China, and its water quality was dominated by ANPSP. This study explored the effect of water level on water quality of Danjiangkou Reservoir and aimed to provide basis for water quality management of large reservoirs. The effect of water level–driven ANPSP on the concentration of reservoir ammonia was studied employing the methods of factor decomposition and multiple regression on a extensive time series data of reservoir ammonia, water level, rainfall, fertilizer usage, and inflow river ammonia. The long-term trend revealed the reservoir ammonia peaked in 2011 and the inflow river ammonia peaked in 2012 (Han River) and 2013 (Dan River), which indicated the success of point source control in the past 15 years and the dominant role of ANPSP in the reservoir ammonia in recent years. With the long-term trend series, the multiple regression results showed that 56% of the variation of the reservoir ammonia concentration was due to the water level (standardized regression coefficient 0.422), fertilizer usage (standardized regression coefficient 0.522), and inflow river ammonia (standardized regression coefficient 0.219). However, the rainfall was insignificant. The predominance of water level and fertilizer usage in explanation of the reservoir ammonia variation indicated that water level–driven ANPSP was the primary factor influencing the reservoir ammonia. The effect of water level was primarily reflected in the long-term variation of ammonia concentration rather than the seasonal variation within the year. This study showed that when compared with rainfall–driven ANPSP, water level–driven ANPSP had a greater impact on the reservoir ammonia. Water quality protection should center on the management of the water level–fluctuation zone.

Exploring the role of the long short‐term memory model in improving multi‐step ahead reservoir inflow forecasting

AbstractDaily inflow forecasting is of vital importance in reservoir economic operation. In the context of hydrometeorological forecasting, the effectiveness of the data‐driven models has been demonstrated as bias correctors for physically‐based models or direct forecasting models. However, existing studies only highlight the performance improvements provided by the data‐driven model, lacking a comprehensive investigation on whether the data‐driven model should be used as bias correctors or direct forecasting models. This study constructs long short‐term memory (LSTM)‐based preprocessing and postprocessing techniques for a hydrological model, which are tested by linear scaling preprocessing and autoregressive (AR) postprocessing models. The integrated model is compared with the LSTM‐only model. The Shuibuya and Zuojiang reservoirs in China are selected as case studies. Results indicate that: (1) LSTM‐based bias correctors are effective in both preprocessing and postprocessing and (2) the integrated model is comparable to the LSTM‐only model when trained with four or more years of data, while it is better than the LSTM‐only model when trained with less data. These findings demonstrate that data‐driven methods can effectively correct the bias in physically‐based model output, and integrating the physical and data‐driven models is useful in improving multi‐step ahead reservoir inflow forecasting if limited data can be obtained.

Multi-source rainfall merging and reservoir inflow forecasting by ensemble technique and artificial intelligence

Study regionThe Shihmen reservoir, the second largest reservoir in northern Taiwan, is flooded frequently, and featured by short period of flood peak due to uneven distribution of rainfall and mountainous topography. Study focusWe proposed a novel streamflow-oriented ensemble recurrent neural networks (ERNN) method to merge three precipitation products, namely gauge measurements, radar and satellite rainfall products. We analyzed (1) the effect of artificial intelligence method for bias correction of precipitation products with reference to streamflow, (2) the comparison of two widely used arithmetic average (AA) and Bayesian model averaging (BMA) methods with ERNN, (3) the performance of merged rainfalls and the original three precipitation products in inflow forecasting during 13 typhoon events. New hydrological insightsWe found that all precipitation products are biased and can be appropriately adjusted with an improvement in inflow forecasting of about 2.5 %, 21.8 % and 60.7 % for gauge, radar and satellite rainfall products in terms of RMSE. After rainfall merging, RMSE for radar and satellite are further improved by 14% and 36% at least. ERNN merging method indicates that the optimal merging weights for gauge, radar and satellite are within the range of 0.52–0.56, 0.35–0.39 and 0.08–0.09, respectively according to 95 % confidence interval. The merged rainfall product skillfully predicts the inflow with a lead time of five hours, and ERNN merging method is superior to traditional AA and BMA methods, with NS up to 0.85 and CRPS reduced by near 50 % compared to BMA. Additionally, ERNN is found to capture the peak times and peak flows with the smallest error. Overall, this study provides a potential approach to merge multiple rainfall products and to obtain the effective rainfall by fitting the hydrological responses over mountainous watersheds, where observational biases may frequently occur in gauge, radar and satellite measurements.

Machine Learning Assisted Reservoir Operation Model for Long‐Term Water Management Simulation

ABSTRACTThis study explores strategies for long‐term reservoir simulations by combining generic rule‐based reservoir management model (RMM) and machine learning (ML) models for two major multipurpose reservoirs — Allatoona Lake and Lake Sidney Lanier in the southeastern United States. First, a standalone RMM is developed to simulate daily release and storage during Water Year 1981–2015. Next, using Long‐Short Term Memory (LSTM) as the ML technique, a standalone LSTM model is trained based on reservoir inflow and meteorological observations to simulate reservoir release and estimate reservoir storage through water balance calculation. Three hybrid modeling strategies are developed, one using RMM output as an additional LSTM input (H1), another using LSTM as the initial release estimate in RMM (H2), and the third combining the first two strategies (H3). The Nash–Sutcliffe efficiency (NSE) for release (NSE‐r), storage (NSE‐s), and their mean (NSE‐avg) are used for model evaluation. Overall, H1 improves NSE‐r to 0.65 and 0.54 for Allatoona and Lanier, respectively, compared to standalone RMM (0.44 and 0.21); however, its storage trajectory did not produce a physically feasible solution, similar to LSTM. H2 and especially H3 show that they can retain the best features from RMM and LSTM, with H3 NSE‐avg being 0.695 and 0.55 for Allatoona and Lanier outperforming RMM (0.615 and 0.29). The findings suggest a robust simulation capacity for large‐scale water management in future studies.

Future runoff forecast in Hanjiang River Basin based on Wetspa model and CMIP6 model

In order to comprehensively consider the impact of human activities on runoff simulation and improve the accuracy of runoff simulation, so as to make a more accurate prediction of the future runoff of the Hanjiang River Basin, this study improved the reservoir module of the Wespa model, adding two parts: reservoir inflow data correction and water storage and outflow data calculation without measured data. Use the improved model to verify its applicability to the Hanjiang River Basin, then, choose the ones who has the most familiar trend with the historical data in the future climate model data (CMIP6). Put the selected data in the model to predict the runoff of Hanjiang River from 2021 to 2060. By analyzing the future runoff trend of Ankang, Huangjiagang and Huangzhuang in the Hanjiang River Basin from 2021 to 2060 and the changes of average runoff, seasonal runoff and monthly runoff compared with the historical period (1981–2020), the conclusions drawn are as follows: 1) The improved Wetspa model has good applicability in the Hanjiang River Basin; 2) The future runoff of Ankang section is decreasing, while that of Huangjiagang and Huangzhuang sections is increasing; 3) Compared with the reference period, the average runoff of the three sections in the future shows an increasing trend, which indicates that there will be flood risk in the future; 4) Compared with the reference period, the runoff proportion of the three sections will increase in spring and winter, and decrease in autumn. Attention should be paid to the risk of drought in autumn. In terms of months, the proportion of runoff from April to June increases, decreases from September to November, and increases and decreases in other months are uncertain.