Study region: The Yellow River Basin (YRB) in China. Study focus: Efficient, equilibrial, and sustainable water allocation is essential for socio-economic development and ecological stability, especially in water-scarce regions. This study proposes a multidimensional assessment framework integrating development efficiency, spatial equilibrium, and temporal sustainability. Efficiency and equilibrium indicators, based on population, GDP, irrigable farmland, and urbanization, reflect regional development levels and disparities. The sustainability indicator, coupling water supply utility with the theory of sustainable orientation, assesses long-term stability across six dimensions: existence, effectiveness, freedom, adaptability, security, and coexistence. New hydrological insights for the region: The evaluation results of water allocation schemes under different scenarios in the YRB demonstrate that the framework can identify and reflect the impacts of changing conditions on water allocation, showing strong stability and sensitivity. The development efficiency, spatial equilibrium, and temporal sustainability exhibit dynamic relationships, without absolute trade-offs or synergies. Increasing water availability can effectively improve development efficiency and temporal sustainability, albeit with a slight reduction in spatial equilibrium. In contrast, blindly raising in-stream ecological water use under scarcity can significantly reduce development efficiency. Adjusting industrial (water use) structure and improving water use efficiency are key to achieving efficient, equilibrial, and sustainable basin development. The framework holds broad potential for application in water-scarce basins worldwide, supporting more balanced and resilient water resource management across diverse socio-economic and ecological contexts.

As a key solution for integrating wind and photovoltaic (PV) plants while reducing emissions, pumped storage hydropower plants (PSHP) must be coordinated effectively with other power resources. To address the challenge of coordinated operation with wind, PV, cascade hydropower, and thermal power plants, this study develops multiple PSHP operation scenarios to quantify both the carbon reduction effects and the residual load fluctuations of different operation modes. The model characterizes the dispatch requirements on the environmental and grid sides in terms of minimizing total carbon emissions and minimizing residual load fluctuation, respectively. In this study, life cycle assessment (LCA) methodology is used to quantify the various energy carbon emission, and the residual load peak-valley difference is used to evaluate the peak shaving effect of PSHP. The case study demonstrates that the proposed method effectively quantifies carbon emissions across various energy sources, reducing total system emissions by 6.3% to 22.6% while maintaining relatively stable electricity generation. Results of the case study show the proposed methodology can effectively compare the carbon emission of different energy resources. And the proposed optimal model framework provides an improvement in the low carbon operation of the combined generation system.

Abstract Flash drought (FD), so‐named for its abrupt and unforeseen onset, poses a significant challenge to forecasting, as current Numerical Weather Prediction (NWP) shows limited skill in the sub‐seasonal to seasonal timescale (S2S, 2‐week to 2‐month range). Here, we present various data‐driven deep learning (DL) frameworks designed to bridge this S2S FD forecasting gap and uncover underlying drought‐inducing mechanisms via interpretability. We developed multiple spatiotemporal DL models (e.g., ConvLSTM, U‐Net) and a Bayesian model averaging (BMA) ensemble to forecast pentad‐scale (5‐day) Standardized Soil‐moisture Index (SSI), serving as the basis for subsequent FD identification. These models leverage diverse drought‐related precursors, including compound drought‐heatwave, evaporative stress, vapor pressure deficit (VPD), and vegetation conditions. Evaluating performance across basins with varied climate regimes, we found that Artificial‐Intelligence‐based methods offer enhanced SSI forecast reliability over NWP, particularly for weather‐scale (1–3 pentads). Notably, the BMA ensemble provided reliable SSI forecasts up to 12 pentads (∼60 days, spanning the entire FD lifecycle), outperforming advanced physics‐based NWP and pixel‐wise benchmark models. Occlusion heatmap reveals that DL models leverage physically plausible precursors for predicting subsequent FD events. Through SHAP analysis, three primary FD‐inducing patterns were identified: water‐dominated (e.g., precipitation), energy‐dominated (e.g., VPD), and multi‐driver composite. Representative regions are arid climate, snow climate, and near‐equatorial areas (e.g., equatorial and warm climate), with widespread interactions among drivers. This study demonstrates explainable DL models potent tools for advancing SSI forecasting and dissecting complex hydro‐climatological drivers relevant to FD assessment, offering novel insights for improved early warning systems.

Runoff change directly alters the spatiotemporal distribution of available water resources and disrupts the dynamic balance between water supply and demand, leading to seasonal or even persistent water scarcity in arid inland river basins. This study aims to improve the adaptability of basin water resources systems to runoff change. Therefore, an adaptive water resources management framework is developed, comprising a joint optimal operation model for reservoirs and ponds, a water resource spatial equity allocation model, and integrated adaptive management strategies encompassing water supply security, water loss control, and water-use coordination. The reservoir operation model and the water resources allocation model are coupled to evaluate the impacts of runoff change on the water supply process and to analyze the characteristics and major influencing factors of basin water scarcity. Adaptive management strategies are then implemented stepwise to evaluate their combined effects on reducing the frequency, duration, and severity of water shortages. The framework is applied to the three major source basins of the Tarim River Basin in northwestern China, namely the Aksu River Basin, Hotan River Basin and Yarkant River Basin. The results show that implementation of the adaptive management strategies increases water supply reliability in the above three basins from 0.25, 0.11, and 0.58 to 1.00, respectively. This study provides methodological support and decision-making references for adaptive water resources management in arid inland river basins.

Rational water resources allocation is crucial for achieving the synergistic development of the water-ecology-economy (WEE) nexus in arid basins. Quantitative assessment of individual water users’ benefits under different water shortage rates provides a robust basis for allocation strategies. This study clarifies the pairwise coupling relationships between the water system and other subsystems using a growth curve function. These relationships are integrated as efficiency functions into a multi-objective water resources allocation model that simultaneously optimizes economic benefits, ecological benefits, carbon sequestration, and spatial equilibrium. The water supply volumes allocated to four vegetation types are designated as decision variables. Four scenarios are evaluated: ecological priority, economic priority, balanced optimization, and comprehensive benefit maximization. Allocation performance is assessed using the coupling coordination degree (CCD) method. This research takes the Tarim River Basin, a typical arid basin, as a case study. Key findings reveal that "S"-shaped and "J"-shaped nonlinear relationships exist between crop growth decay rates and water shortage rates. To avert significant losses, it is advisable to maintain water shortage rates below the first inflection point. Notably, most sub-basins perform optimally under the comprehensive benefit maximization scenario, exhibiting the highest CCD in the WEE nexus. These findings provide scientific guidance for both efficient water resource utilization and sustainable development of the WEE nexus in arid basins.

The solution for large-scale multi-objective optimization operation modelling of cascade reservoirs is always one of the difficult issues and hot topics for water resources operation and allocation. However, the multi-objective operation model has the characteristics of complex hydraulic connection, high-dimensionality, and multiconstraint, and how to efficiently and accurately solve and model is always a challenge in decision-making and management of water resources. To avoid the phenomenon and limitations of the conventional studies on the solution of the operation model used to excessively depend on intelligent optimal algorithm without taking the complex hydraulic connections of variables into consideration, in this paper, a new approach of a synergistic optimization strategy combining a decomposition optimization framework (DOF) with an improved Nondominated Sorting Genetic Algorithm II (NSGA-II) was proposed. The new method completely avoided the curse of dimensionality by variable division, boundary condition transmission, and hydraulic connection reconstruction, and significantly improved the computational efficiency and accuracy of the global optimization (GO). The Lancang River (LCR) cascade reservoirs were selected as a case study, and the multi-objective model was constructed to apply the new approach. The results show that compared with the GO, the GO by the DOF (GO-DOF) reduces the computational time by 86.9 %, significantly decreases the degree of ecological change (DEC) by 42.6 % with power generation decreasing by only 0.6 %. It is concluded that the DOF shows excellent performance in both computational efficiency and operation outcomes, providing a new approach to an efficient solution for the operation and allocation optimization of complex water resources systems.

Reliable medium- to long-term runoff forecasting is of critical importance for water resource management and regional planning. However, the high nonlinearity and non-stationarity of hydrological processes continue to pose significant challenges. To address this issue, this study proposes an integrated forecasting framework termed SVPsEC, which embeds multi-module collaboration and cross-subsequence spatiotemporal coupling within a unified forecasting system. In this framework, Variational Mode Decomposition (VMD) decomposes the original runoff series into intrinsic mode functions with reduced complexity, effectively preserving multi-scale hydrological dynamic characteristics. Parallel CNN–GRU modules collaboratively extract local spatial features and long-term temporal dependencies of each component, achieving early fusion of heterogeneous spatiotemporal information. The Sparrow Search Algorithm (SSA) is employed to optimize the hyperparameters of the decomposition, feature extraction, and forecasting modules, further enhancing the collaborative capability of the framework. Additionally, an error correction strategy is introduced to progressively correct residuals and improve forecasting stability. Evaluations at three hydrological stations demonstrate that SVPsEC consistently produces highly accurate and stable forecasts, with NSC values all exceeding 0.98, R above 0.99, and significantly reduced RMSE compared to benchmark methods. Peak flow forecasting performance is also notably improved. Overall, the decomposition-based multi-module collaboration strategy, combined with coupled spatiotemporal representation, effectively enhances runoff forecasting capability and provides a reliable tool for watershed-scale water resource planning and hydrological risk management. A multi-strategy hybrid model (SVPsEC) has been proposed for regional monthly scale forecasting. The parallel CNN-GRU network architecture is utilized to capture both long-term and short-term spatiotemporal information. The SSA optimization algorithm improves the generalization and predictive accuracy of the parallel CNN-GRU model. The application of error correction strategies enhances the model's ability to predict runoff sequences and extremes. Comparative results demonstrate that the predictive performance of the SVPsEC model exceeds that of other models.

Mountain glaciers play a critical role in freshwater supply and hydrological regulation. Climate warming is projected to accelerate glacier melting in High Mountain Asia, threatening water resource sustainability, particularly in the arid and semi-arid regions. This study employs the integrated ice-dynamic Open Global Glacier Model to investigate glacier responses to past and future climate change. The annual mass balance series of 11,625 glaciers from 1990 to 2019 is reconstructed, revealing their spatiotemporal variability in alpine regions. Additionally, the long-term dynamics of glacier mass, area, and volume are assessed through 2100 under different climate scenarios, focusing on glacier runoff changes and the differences in the timing of peak runoff. Finally, the unit area service pricing method is applied to establish an index for quantifying the ecological value loss resulting from glacier retreat. The results show that climate warming will lead to substantial and irreversible mass loss. This process is characterized by initial thinning followed by retreat, resulting in an overall negative mass balance. By the end of the 21st century, glacier area and volume will decrease by more than 40% in most basins, with peak runoff expected around 2050 under low-emission scenarios. In contrast, delayed peaks are generally associated with high-emission scenarios and regions with substantial glacier reserves. Glacier service values are predicted to decline significantly or be entirely lost, with this trend intensifying from southwest to northeast in the Tarim River Basin. This study provides new insights into glacier dynamical and hydrological responses under climate warming, contributing to regional socio-ecological sustainability and optimized water resource management.

Study region: The Middle and Lower branches of Lancang River (LCR) Basin, China. Study focus: Cascade reservoir operations in the LCR are increasingly challenged for balancing hydropower and eco-sustainability under multi-objective and uncertainty-prone conditions. Therefore, it's urgent to investigate difficult-to-quantify dependencies among risks arising in multi-objective operations. A dual-core framework integrating dynamic weight optimization with copula-based risk dependency analysis was proposed. A multi-objective operation model (ModelI) is developed to maximize power generation and minimize the degree of ecological change (DEC), and a single-objective model for maximizing power generation (Model-II) was as the comparison. An innovative solution method was introduced by coupling improved binomial distribution weighting (IBDW) method with entropy weighting (EW), TOPSIS, and Grey Relational Analysis (GRA). The weights for power generation and ecology were determined as 0.785 and 0.215. New hydrological insights for the region: Results show that compared to Model-II, Model-I reduces DEC (the deviation between discharge and ecological flow) from 25.98 % to 17.35 %, with only a 4.06 % reduction in power generation (from 73,213.4 GWh to 70,242.6 GWh), demonstrating significant ecology improvement with minimal power generation loss. Furthermore, Copulabased risk correlation model coupled with three methods (Kendall, Spearman, tail dependence) reveals spatial variations in risk correlations, elucidating risk association mechanisms of cascade reservoirs. The framework provides a new perspective for risk investigation in multi-objective reservoir operations.

Pumped storage hydropower plants (PSHPs) in China have recently begun participating in inter-provincial medium- and long-term (IPMLT) markets. However, the effective utilization of their regulation potential is currently hindered by rigid, fixed-path trading mandated by the administrative two-part tariff system. This "point-to-point" isolation prevents PSHPs from establishing competitive relationships across a broader geographical scope, leading to underutilization of their wide-area regulation capabilities. To bridge this gap, we propose an inter-provincial multi-channel centralized bidding framework specifically designed for the spatiotemporal coupling characteristics of PSHPs. This framework integrates three novel components: (1) two centralized clearing models developed to decouple the pumping and generation processes, establishing a multichannel trading mechanism that maximizes PSHP's arbitrage revenue; (2) a decomposition and pairing method designed to apply merit-order matching to preliminary cleared energy, reconciling operational constraints with market economics in the final clearing results; (3) a multi-stage security verification strategy devised to precisely identify and curtail infeasible energy, thereby avoiding the over-curtailment inherent in pro-rata methods. The proposed framework is validated through a case study of the Zhen'an PSHP in China, demonstrating its capability to ensure physical feasibility while improving rescource allocation efficiency.

Inter-basin water transfer (IBWT) projects represent one of the most effective approaches to alleviating water scarcity caused by the uneven spatiotemporal distribution of water resources. However, the long-term sustainability of these projects critically depends on scientific regulation. Beyond addressing the inherent trade-offs between fairness and efficiency, IBWT projects must be systematically coordinated with existing cascade reservoirs in the receiving basin, rather than functioning in isolation. To this end, this study develops a nested two-layer coordinated regulation framework that integrates fairness and efficiency for IBWT projects and cascade systems. This model is applied to Phase I of the South-to-North Water Transfer Western Route (SNWT-WI) in the Yellow River Basin (YRB). Three scenarios—Basic (without the SNWT-WI), SO (standalone operation), and CO (coordinated operation)—were established for comparative analysis. Compared to the SO scenario, the CO scenario demonstrates significant advantages in securing regional water supply. Although it involves minor trade-offs regarding in-stream ecology and total power generation, the CO water transfer process is substantially more flexible. This flexibility provides superior adaptability to hydrological variability and aligns more closely with the current socio-economic development demands of the YRB. Nevertheless, the augmented inflow from the SNWT-WI increases spillway discharge, presenting a new operational challenge for existing cascade hydroelectric stations. Overall, the proposed coordinated regulation model demonstrates broad applicability, offering valuable methodological insights for the sustainable management of large-scale IBWT projects worldwide

Intensified drought significantly shifts the structure and function of the ecosystem, driving asynchronous changes between them. However, understanding the relationship between terrestrial ecosystem responses to drought and vegetation growth remains a persistent challenge due to limited direct observations. Here, we used gross primary productivity as a proxy for carbon sink, the normalized difference vegetation index for canopy structure, and the standardized ecological water shortage index to disclose this relationship across 24 ecological-climatic regions and 6 vegetation types. Integrating climate, vegetation, soil, and topography factors, the ecological-climatic regions were classified using Fuzzy C-Means method combined with the ant colony algorithm. The results indicated that 55.7 % of the vegetated areas in the Yangtze River Basin (YRB) have experienced inconsistent vegetation growth in canopy structure and the ecosystem carbon sink. More than 66 % of vegetated areas displayed short-term (<= 3 months) responses to ecological drought. Notably, forest ecosystems showed much longer lagged responses, with mean NDVI lag time exceeding 7 months in significantly decreasing regions. The ecosystem carbon sink is more sensitive to ecological drought than canopy structure. Grassland is the most sensitive vegetation type in the YRB, and forests express the most pronounced ecological drought impacts. Generally, vegetation in arid regions is more sensitive to ecological drought than in humid areas. Ecosystem carbon sink in areas of increased growth shows greater sensitivity to ecological drought than in areas of decreased growth. Furthermore, across 9 vegetation growth pattens between ecosystem carbon sink and canopy structure, the sensitivity of ecosystem carbon sink and canopy structure to ecological drought also varies distinctly. The lowest sensitivity of vegetation to ecological drought was observed when ecosystem carbon sink declined, and canopy structure increased within the YRB.

The operation of hydropower plants must balance multiple demands. Specifically, plant operators prioritize economic benefit. Grid managers require enhanced peak-shaving capability to address source-load uncertainty. Basin managers focus on reservoir safety to prevent overtopping accidents. Coordinating these objectives is particularly challenging for weakly regulated cascade hydropower plants(WRCHP), which consist of multiple adjacent plants with limited regulation capacity. This study proposes a multi-objective bilevel risk-economic dispatch model. At the plant level, economic benefit is quantified by combining electricity revenue and ancillary service compensation, while peak-shaving capability is evaluated using a CVaR-based residual load variance risk metric. Furthermore, a method for calculating emergency time based on the dynamic control of daily maximum water level is proposed to quantify dam safety margins against overtopping. At the unit level, the model minimizes water consumption and avoids unit vibration zones. The model is applied to the ZM and JC hydropower plants in the YZ River basin. Results reveal trade-offs among economic benefit, peak-shaving capability, and overtopping risk, and identify key influencing factors such as plant output, forebay level, generation head, and grid load. In addition, this study also summarizes dispatching strategies for WRCHP under multi-objective coordination, providing valuable guidance for operators.

In arid regions, the limited water resources are bringing challenges to the basin sustainable development, and the increasing diverse water requirements intensifies the water supply-demand imbalances. The typical case of reservoir group located in the Irtysh River Basin (containing 3 rivers, 4 reservoirs, and 18 water consumers in 6 categories) in northwest China was selected. The coordinated optimization of conflicting multi-objectives has been the critical bottleneck for such a complex system. To alleviate the problems above, the comprehensive water resources utilization rules are mathematized and introduced into three multi-objective operation models (Model-I, Model-II and Model-III) which stands for key decision-making preferences (water supply, ecological water and hydropower generation). To solve the complex multi-dimensionality (3816 variables in 53 hydrological years) and multi-constraint models, a new solution approach with reducing the search space and time series was proposed to achieve the optimal schemes. To deeply explore the relationship and interactions among objectives, the Vector Autoregressive (VAR) model was constructed. It is concluded that the new solution approach achieves good efficiency and operation performance, and all the optimal schemes reached the comprehensive water utilization rules (even beyond the designed guarantee rates of 95%, 90%, 75% and 50%). In addition, based on limited water resources, the water supply and eco-water supply competed significantly, except in BEJSK, which means regulation capacity of reservoir impacted the relationship among the objectives a lot. Finally, the VAR model is available to quantify the interactions among objectives of reservoir group operation, providing a new idea for the related research.

With the rapid increase in renewable energy penetration and the ongoing reform of electricity markets, the coordinated operation of hydro-wind-PV integrated energy bases has become a critical issue for enhancing system flexibility and market efficiency. However, under continuously evolving market mechanisms, how to effectively coordinate electricity pricing, generation operation, and demand-side participation remains a core challenge. To address this issue, this study develops a bi-level optimization framework based on Stackelberg game theory for the coordinated operation and pricing strategy of an integrated hydro-wind-PV energy base in the day-ahead electricity market, in which the energy base operator acts as the leader by determining bundled electricity prices, while the hydro-wind-PV generators and end-users act as followers, making generation and demand response decisions, respectively. The proposed framework is validated using a real-world case study in the upper reaches of the Yellow River Basin, China. The results demonstrate that: (1) After implementing price-based demand response, the peak load shaving rate reaches 28.16%. (2) The proportions of electricity purchased from and sold to the external grid decrease by 1.04% and 2.24%, respectively. (3) Over the 24-hour day-ahead operation horizon, the revenue of the hydro-wind-PV generators increases by RMB 32.5 million, while consumer costs decrease by RMB 189.08 million. These findings offer policymakers a strategic coordination framework for designing pricing and demand-response mechanisms that reconcile economic benefits with the promotion of local renewable energy consumption.

The rapid expansion of wind and photovoltaic energy has significantly driven the growth of pumped-storage hydropower plants (PSHP). As a prominent PSHP configuration, pump-back PSHP (PSHPPB) utilizes cascade in-stream reservoirs from conventional hydropower plants (HP) by integrating reversible pump-turbines technology. However, this configuration imposes considerable operational pressure on cascade reservoirs, necessitating a delicate balance between fulfilling water supply demands, managing PSHPPB pumping and generation cycles, and maintaining conventional HP output. This study: (1) investigates the coordinated operation of PSHPPB-HP system to enhance photovoltaic (PV) power integration under water supply priorities; and (2) evaluates the optimal PV capacity that can be integrated into the PSHPPB-HP system, accounting for long-term basin-scale water supply variability. A case study of the PSHPPB located in the upper Yellow River Basin of China, utilizing the cascade reservoirs of Longyangxia (LYX) and Laxiwa (LXW) hydropower stations, is presented. Results indicate that: (1) an increase in the daily average release from the LYX reservoir shifts the PSHPPB operation from a pumping-generating mode to a generation-only mode; (2) energy consumption of the PSHPPB decreases to zero as the daily average reservoir release increases, while its power generation follows a distinct dip-and-climb pattern; (3) two critical daily average reservoir release thresholds, approximately 900 m3/s and 1500 m3/s, are identified for optimizing the PV integration in the PSHPPB-HP system; (4) the PSHPPB-HP system can support up to 4141.6 MW of PV capacity (62% of the total installed capacity of the PSHPPB-HP system) with PV curtailment limited to 5%, accounting for long-term water supply variability in the Yellow River Basin. These findings provide valuable insights for PSHPPB planning and operation, not only in the Yellow River Basin but also in other global regions.

Transitioning from passive drought control to active risk management requires a systematic understanding of drought propagation beyond conventional short chains. This study established a novel framework to characterize high-resolution, cascading drought dynamics across meteorological, hydrological, agricultural, ecological, and socioeconomic dimensions in the drought-prone Yellow River Basin. By integrating SWAT and AquaCrop simulations with electrical network-inspired “series–parallel–hybrid” theory, we identified dominant propagation pathways and thresholds across 403 sub-basins. Results revealed distinct spatiotemporal groupings where meteorological–ecological and meteorological–agricultural droughts shared similar propagation times, as did meteorological–socioeconomic and meteorological–hydrological droughts. Meanwhile, when ecological drought of varying severity occurred in phase 1, the cascade consistently progressed from meteorological to ecological, then to agricultural, further to hydrological, and ultimately to socioeconomic drought. Series and hybrid modes dominated drought propagation pathways, with series propagation increasing under intensified meteorological drought. As drought severity rose, propagation thresholds declined while exhibiting clear spatial clustering patterns. Driving force analysis indicated that dynamic cumulative processes rely more on environmental drivers than static threshold conditions. This study fills a critical gap in long-chain drought propagation research and supports the development of cascade-based early warning and targeted regulation systems.

Socioeconomic drought is essentially a supply–demand imbalance, yet the cumulative water stress generated by this imbalance remains insufficiently quantified in human-modified river basins. This study proposes a non-stationary assessment framework by integrating the GAMLSS model, Copula functions, and Bayesian conditional probability inference, using the lower Yellow River basin (1980–2022) as a case study. We introduce the Socioeconomic Drought Potential Vulnerability Threshold (SEDVT) to quantify the "initial risk load"—defined as the maximum Antecedent Cumulative Water Deficit (ACWD) a system carries at the moment of drought onset. The Drought Resistance Capacity (DR) is further evaluated using the ratio of Conditional Return Periods. The findings reveal: (1) A significant regime shift in drought patterns occurred, with the frequency of severe and extreme droughts decreasing by nearly 40

Cascade development is a key approach for modern hydropower utilization. However, differences in ownership and scheduling authority lead to multiple operation modes in cascade hydropower plants, including cooperative scheduling (CSM), game-based scheduling (GSM), and independent scheduling (ISM). These modes influence both operational efficiency and the accommodation of wind and PV, thereby affecting the power system's carbon reduction potential. This study proposes a framework to evaluate carbon reduction benefits under different scheduling modes. A load generalization method is developed to capture daily variability and peak shaving, based on which a multi-energy complementary carbon reduction model (MECR-PS) is established. Using the LYX, LXW, and NN cascade hydropower plant as a case study, the results show: (1) CSM improves carbon reduction benefits by 3.5 % compared to ISM and GSM. (2) GSM benefits upstream plants, while ISM favors downstream plants. (3) Wind generation and hydropower discharge are positively correlated with carbon reduction. (4) Under high-flow conditions, GSM for the upstream, ISM for the midstream, and CSM for the downstream plants yield the highest individual power generation. Under other conditions, CSM consistently delivers the highest generation across all plants. These findings offer valuable insights for cascade hydropower operation and lowcarbon power system development worldwide.

Study region The cascade reservoirs in the lower Lancang River, southwestern China. Study focus Natural river ecosystems rely on specific thermal regimes, which are frequently disrupted by deep reservoirs releasing cold hypolimnetic water. This study investigated how interannual hydrological variability controls thermal stratification, and how selective withdrawal via stoplog gates influences downstream thermal habitats and hydropower generation. By coupling a three-dimensional hydrodynamic model with a streamwise temperature model, the thermal dynamics were simulated. A systematic assessment was then conducted to compare the thermal habitat improvement and the hydropower losses between bottom-outlet operation and selective withdrawal. New hydrological insights for the region Results indicated persistent vertical stratification in the reservoir forebay across typical hydrological years, with a stable hypolimnion temperature of 15 °C and a distinct stratified regime from May to August. Utilizing multi-layer stoplog gates shifted withdrawal flows to warmer surface layers, raising downstream temperatures by a maximum of 8.34 °C. Moreover, the thermal guarantee rate for the spawning of the indicator species (Tor sinensis) increased from 14.29% to 66.67%. This thermal habitat improvement incurred negligible hydropower losses, yielding a 0.33% reduction in total generation and a 0.058% decrease in guaranteed output. These results confirmed that coupling selective withdrawal with cascade optimization effectively balances thermal habitat suitability and hydropower generation, providing information to support reservoir management in highly regulated river basins.