Journal of Hydrology-Regional Studies最新文献

{"title":"Dynamic evolution of rainfall erosivity in different rocky desertification areas and analysis of multidimensional driving factors","authors":"Yingshan Zhao ,&nbsp;Dayun Zhu ,&nbsp;Zhigao Wu ,&nbsp;YuRong Han ,&nbsp;WanQing Liu ,&nbsp;YuRong Mu ,&nbsp;Hua Xiao","doi":"10.1016/j.ejrh.2025.102492","DOIUrl":"10.1016/j.ejrh.2025.102492","url":null,"abstract":"<div><h3>Study region</h3><div>Karst rocky desertification areas in southwest China, covering eight control zones.</div></div><div><h3>Study focus</h3><div>This study uses Mann-Kendall test, Gravity Center model, and Optimal Parameter-based Geographical Detectors to analyze spatiotemporal distribution and driving factors of rainfall erosivity (R) in different rocky desertification control zones from 1961 to 2020. It also employs NEX-GDDP-CMIP6 model to predict future changes in R.</div></div><div><h3>New hydrological insights for the region</h3><div>The results indicate that: (1) The long-term average R in region is 4656.20 MJ mm·ha⁻¹ ·h⁻¹ ·a⁻¹ , with significant variations across different control zones. The maximum and minimum R values are found in Karst Peak Forest-Plain (PFP) and Karst Canyon (KC) zones, respectively. Seasonally, summer has the highest erosivity, followed by spring, autumn, and winter, with increasing trends in summer, autumn, and winter. (2) Rainfall is the primary driver of erosivity, with an average explanatory power exceeding 55 %. The interaction between socioeconomic factors and rainfall significantly contributes to explaining rainfall erosivity. (3) Future predictions show that rainfall erosivity will further increase, with changes ranging from −5.78–85.60 %. Notable increases are observed in Karst Hills and Depression (KHD), Karst Trough Valley (KTV), and PFP zones. The center of erosivity is expected to shift northwest, southwest, and southeast. These findings enhance our understanding of rainfall erosivity in rocky desertification areas and inform the update of soil and water conservation measures.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102492"},"PeriodicalIF":4.7,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144166586","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Application of stochastic models for long-range sediment transport during extreme typhoon events","authors":"Yu-Ju Hung,&nbsp;Christina W. Tsai","doi":"10.1016/j.ejrh.2025.102463","DOIUrl":"10.1016/j.ejrh.2025.102463","url":null,"abstract":"<div><h3>Study Region</h3><div>The research focuses on the Shihmen Reservoir in northern Taiwan, a multifunctional rolled rockfill reservoir experiencing reduced capacity due to sediment accumulation, which is further exacerbated by significant flood events during typhoons.</div></div><div><h3>Study Focus</h3><div>Our research employs a novel fractional stochastic diffusion particle tracking model (FSD-PTM), incorporating fractional Brownian motion to account for the long-range dependencies observed in suspended sediment transport during extreme typhoon events. By capturing these dependencies, the model offers improved predictions of sediment dynamics, essential for understanding and managing sediment-related challenges in hydrological reservoir systems under severe weather conditions, thereby supporting operational safety and environmental compliance.</div></div><div><h3>New Hydrological Insights for the Region</h3><div>The application of the FSD-PTM has provided new insights into sediment dynamics during typhoon events in the Shihmen Reservoir. Traditional models based on memoryless diffusion often fail to represent the persistence observed in suspended sediment transport under sustained turbulent flows. By incorporating long-range dependence into the particle motion framework, this study highlights the importance of memory effects in simulating sediment spread. These findings support more accurate predictions of sediment deposition patterns, providing a scientific basis for adaptive reservoir operation strategies in response to intensifying climatic extremes.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102463"},"PeriodicalIF":4.7,"publicationDate":"2025-05-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144166585","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simulation of the impacts of projected climate change on groundwater resources in the Urban, Semiarid Yucaipa Valley Watershed, Southern California using an integrated hydrologic model","authors":"Derek W. Ryter ,&nbsp;Ayman H. Alzraiee ,&nbsp;Richard G. Niswonger","doi":"10.1016/j.ejrh.2025.102461","DOIUrl":"10.1016/j.ejrh.2025.102461","url":null,"abstract":"<div><div>Managing water resources in semiarid watersheds is challenging due to limited supply and uncertain future climate conditions. This paper examines the impact of future climate changes on an urban watershed in southern California using an integrated hydrologic model. GSFLOW modeling software is used to simulate the nonlinear relationships between climate trends and precipitation partitioning into ET, runoff, and subsurface storage. Four global circulation models (GCMs), each with two greenhouse-gas scenarios, RCP45 and RCP85 are used to project future climate conditions. GCMs include the CanESM2, CNRM-CM5, HadGEM2-ES, and MIROC5 models. The model's simulated hydrologic conditions are compared with historical data to assess changes in water budgets and groundwater supply. Results indicate decreased groundwater storage in most scenarios due to increased natural evapotranspiration, vegetation consumptive use, and streamflow out of the watershed. Only scenarios with substantially increased future precipitation show increased groundwater storage. The study also highlights increased future aridity despite the rise in precipitation and large precipitation events forecast by GCMs, which increase the risk of urban floods and decrease stream leakage and water available to vegetation.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102461"},"PeriodicalIF":4.7,"publicationDate":"2025-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144166584","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dimensionality reduction for groundwater forecasting under drought and intensive irrigation with neural networks","authors":"Tarik Bouramtane ,&nbsp;Ismail Mohsine ,&nbsp;Nourelhouda Karmouda ,&nbsp;Marc Leblanc ,&nbsp;Yannick Estève ,&nbsp;Ilias Kacimi ,&nbsp;Mohamed Hilali ,&nbsp;Salima Mdhaffar ,&nbsp;Sarah Tweed ,&nbsp;Mounia Tahiri ,&nbsp;Nadia Kassou ,&nbsp;Ali El Bilali ,&nbsp;Omar Chafki","doi":"10.1016/j.ejrh.2025.102477","DOIUrl":"10.1016/j.ejrh.2025.102477","url":null,"abstract":"<div><div>Study Region: This study focuses on the Berrechid aquifer system in northern Morocco.</div><div>Study Focus: The research explores Principal Component Analysis (PCA) for optimizing input selection in groundwater level forecasting using neural networks. PCA efficiently reduces input dimensionality while preserving critical information, making it beneficial for neural network modelling of natural systems with extensive input variables in a low-resource scenarios requiring feature engineering. A Long Short-Term Memory (LSTM) model predicted groundwater levels in six monitoring bores using four hydro-climatic variables, precipitation, land surface temperature (LST), actual evapotranspiration (AET), and the normalized difference vegetation index (NDVI). Model performance was compared using two approaches: the LSTM-XGB model with the best-selected input features and the LSTM-PC1 model based on the first principal component (PC1). New Hydrological Insights for the Region: Results showed that NDVI, AET, and LST were the dominant inputs across different monitoring bores. On average, PC1 accounted for 68.3 % of the variance in hydro-climatic variables, with an eigenvalue of 2.75, surpassing the combined variance of two individual hydro-climatic variables. Both models performed effectively, achieving R² values of 0.982–0.999 during training and 0.885–0.999 during validation. The models successfully captured groundwater fluctuations and the declining trend during drought. LSTM-XGB slightly outperformed LSTM-PC1 in certain cases, but the differences were minimal. The use of PC1 not only mitigates overfitting risks but also allows for generalized predictions across multiple monitoring sites, making it a practical choice for large datasets.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102477"},"PeriodicalIF":4.7,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144146858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Air-for-water substitution in characterizing lake heatwaves in the middle and lower reaches of the Yangtze River Basin","authors":"Zihan Zhu ,&nbsp;Wei Wang ,&nbsp;Heng Lyu ,&nbsp;Bo Wang ,&nbsp;Zhiwen Wen ,&nbsp;Qiuxiang Jin","doi":"10.1016/j.ejrh.2025.102475","DOIUrl":"10.1016/j.ejrh.2025.102475","url":null,"abstract":"<div><h3>Study region</h3><div>This study focuses on seventeen lakes with surface area larger than 100 km² located in the middle and lower reaches of the Yangtze River Basin, China.</div></div><div><h3>Study focus</h3><div>This study compares the spatiotemporal differences between lake heatwave and atmospheric heatwave metrics and elucidates the intensity differences between two heatwaves across seventeen lakes in the middle and lower reaches of the Yangtze River Basin.</div></div><div><h3>New hydrological insights for the region</h3><div>This study demonstrated that air-for-water substitution is invalid in characterizing lake heatwaves in this region. Compared to atmospheric heatwaves, lake heatwaves in the study region are characterized by higher frequency, lower intensity, and longer duration. Significantly more frequent and longer-lasting lake heatwaves are observed in the middle and lower reaches of the Yangtze River. At the interannual timescale, lake heatwaves are increasing faster than atmospheric ones by 0.8 d per decade in total annual duration and 1.2 times as fast in frequency, but more slowly by 0.3 K per decade in average intensity. But the difference in long-term change rates between two heatwaves exhibits no discernible spatial pattern. Although air temperature (4.41 K, 94.23 % of the total) is the largest contributor to the average intensity of lake heatwaves, other meteorological variables collectively modulate lake heatwave intensity and cannot be neglected.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102475"},"PeriodicalIF":4.7,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144146857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generation of sub-daily precipitation time series anywhere in Switzerland by mapping the parameters of GWEX-MRC, an at-site weather generator","authors":"Kaltrina Maloku,&nbsp;Guillaume Evin,&nbsp;Benoit Hingray","doi":"10.1016/j.ejrh.2025.102454","DOIUrl":"10.1016/j.ejrh.2025.102454","url":null,"abstract":"<div><div><em>Study Region</em></div><div>Switzerland</div><div><em>Study Focus</em></div><div>Stochastic weather generators (WGENs) are a common tool for generating long precipitation scenarios, also needed at ungauged sites. This study evaluates different methods for obtaining the parameters of a hybrid at-site WGEN at any location within the study area. The hybrid GWEX-MRC model is composed of GWEX, a daily WGEN, and MRC, a disaggregation model based on multiplicative random cascades. Two approaches are considered for obtaining parameter maps. The first approach applies classical spatial interpolation techniques, kriging and thin-plate splines, to parameter estimates derived from rain gauge data. The second approach uses CombiPrecip, an hourly gridded precipitation product from MeteoSwiss, to estimate parameters at grid scale. <em>New Hydrological Insights for the Region</em></div><div>We find that the parameters of GWEX-MRC can be interpolated with satisfactory results across Switzerland. Among interpolation techniques, kriging with elevation as an external drift performs best for GWEX, while thin-plate spline with elevation gives better results for MRC. The comparison of the two approaches, interpolation of site-based estimates and direct parameter estimation using CombiPrecip, showed comparable or slightly different performance depending on the precipitation statistic and season. These findings reveal the feasibility of both approaches and provide insights into their relative strengths and limitations. In addition, this study demonstrates that long precipitation scenarios can be reliably generated throughout Switzerland, which can later be used to feed a hydrological model.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102454"},"PeriodicalIF":4.7,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144139325","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Sediment-flushing equilibrium downstream of tide gates in a coastal region: Findings from a 36-day prototype water-diversion experiment","authors":"Fanzhi Meng ,&nbsp;Long Xiang ,&nbsp;Jianying Zhu ,&nbsp;Haishen Lv ,&nbsp;Cuiling Jiang","doi":"10.1016/j.ejrh.2025.102472","DOIUrl":"10.1016/j.ejrh.2025.102472","url":null,"abstract":"<div><h3>Study region</h3><div>The Lixiahe Region, a coastal area with tide-gate-regulated port channels vulnerable to sedimentation impacts.</div></div><div><h3>Study focus</h3><div>Investigating sedimentation challenges induced by tide gates in port channels and testing a sediment management method combining water transfer strategies (informed by mountainous reservoir practices) with optimized gate operations through a 36-day prototype experiment.</div></div><div><h3>New hydrological insights for the region</h3><div>The study quantified critical water volumes required to achieve sediment-flushing equilibrium and identified optimal timing for tide gate operations, providing actionable thresholds for balancing sedimentation and hydraulic efficiency. Furthermore, it demonstrated that coordinated water transfer (inspired by mountainous reservoir management) and adaptive gate operations outperform conventional dredging in mitigating sedimentation while reducing operational costs and ecological disruption. Building on these findings, a transferable sediment management scheme was developed, integrating dynamic water allocation with gate control protocols. This framework was validated against decade-long operational data (2012–2021), revealing its adaptability to seasonal hydrological variability and long-term sediment dynamics. The scheme proposes actionable strategies, including pre-flooding water diversion to preemptively flush sediments and synchronized gate adjustments during tidal cycles, to enhance channel stability. These insights offer coastal regions with tide-gate systems a scalable, eco-sensitive alternative to dredging, aligning sediment control with sustainable water resource management under changing hydrological conditions.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102472"},"PeriodicalIF":4.7,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A three-parameter runoff probability model for long-slope croplands in the black soil region of Northeast China","authors":"Yiting Huang ,&nbsp;Yuhao Gao ,&nbsp;Junxi Chen ,&nbsp;Ziheng Feng ,&nbsp;Yi Zhang ,&nbsp;Yangbo He ,&nbsp;Zhengchao Tian ,&nbsp;Lirong Lin ,&nbsp;Jiazhou Chen","doi":"10.1016/j.ejrh.2025.102471","DOIUrl":"10.1016/j.ejrh.2025.102471","url":null,"abstract":"<div><h3>Study region</h3><div>The Mangang black soil region of Northeast China, characterized by long or ultra-long slopes (100–4000 m), is experiencing severe water erosion.</div></div><div><h3>Study focus</h3><div>Soil erosion has worsened, particularly on long slopes, due to precipitation variability and tillage disturbance. Existing models struggle to predict runoff because they fail to capture nonlinear shifts among multiple runoff-generation mechanisms. To address this, a probabilistic runoff model with three parameters (<em>β</em>, <em>μ</em>, <em>τ</em>) was developed, coupling infiltration, re-infiltration, and exfiltration with Hortonian, Betsonian, and Duncanian regimes. By extending the Sheridan framework with an exponential exfiltration function, the model estimates runoff coefficients based on slope length and soil saturation. It was validated using rainfall plots, field observations (1–600 m), and literature data, and benchmarked against the Sheridan model and HYDRUS-2D.</div></div><div><h3>New hydrological insights</h3><div>The model reveals a “hook-shaped” nonlinear relationship between slope length and runoff coefficient (R2 =0.92–0.99, KGE=0.87–0.98), with a turning point (T_p, 100–200 m) marking the shift from decreasing to increasing runoff. A new threshold, <em>τ</em> (incomplete saturation slope length), reflects saturation-driven exfiltration feedback, dividing slopes into: (1) Wetting zone (0-T_p): dominated by Hortonian runoff; (2) Transition zone (T_p-<em>τ</em>) with rapid exfiltration and gully initiation; (3) Saturated zone (&gt;<em>τ</em>): dominated by Duncanian runoff. Compared with HYDRUS-2D, the model improves runoff distance estimation by 25 % and reduces runoff errors by 37–62 %, supporting adaptive land management.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102471"},"PeriodicalIF":4.7,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparative analysis of lumped and semi-distributed hydrological models for an upland watershed in Ethiopia","authors":"Gebiaw T. Ayele,&nbsp;Bofu Yu","doi":"10.1016/j.ejrh.2025.102486","DOIUrl":"10.1016/j.ejrh.2025.102486","url":null,"abstract":"<div><h3>Study regions</h3><div>The Koga watershed is in the Blue Nile River Basin in Ethiopia.</div></div><div><h3>Study focus</h3><div>The study conducted in a hilly dominated tropical catchment in Ethiopia adopted a multi-model approach by comparing six hydrological models, including the semi-distributed Soil and Water Assessment Tool (SWAT) and five lumped conceptual models, namely: Australian Water Balance Model (AWBM), Sacramento, Simplified HYDROLOG (SimHYD), Soil Moisture Accounting and Routing (SMAR), and Tank models. The study aimed at evaluating model performance and sensitivity of parameters in predicting streamflow for tropical watersheds. Calibration and uncertainty analysis (UA) for SWAT was performed using four UA techniques available in the SWAT (SWAT-CUP) and the genetic algorithm was used for parameter estimation for conceptual models. The model comparative analysis was supplemented by multi-model averaging using Sum of Squared Errors (SSEW) and Performance-Based Weights (PBW-NSE) to improve prediction reliability.</div></div><div><h3>New hydrological insights for the region</h3><div>Results showed that the semi-distributed scheme using SWAT-SUFI-2 provided practical applications in land use change modelling and provided better calibration efficiency with limited computational resources. Sacramento offers improved performance with cost-effective alternative to hydrological modelling in a data-scarce environment. For monthly flows, SWAT- SUFI-2 achieved NSE of 0.60 and PBIAS of 24 % for calibration (P-factor = 73, R-factor = 93) and 0.60 and 9 % for validation (P-factor = 70, R-factor = 88). Sacramento model yielded NSE of 0.55 and PBIAS of 16 % during calibration, and 0.72 and −4 % for validation, respectively. Compared to the individual models, multi-model averaging enhanced overall performance, with monthly NSE ranging from 0.60 to 0.67 and R² from 0.71 to 0.74. While similar in terms of model performance, SWAT takes into account the spatial variability in soil, land use, and terrain, hence has the capacity for impact assessment of environment change in comparison to lumped conceptual models.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102486"},"PeriodicalIF":4.7,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134842","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A combined prediction model for middle and long-term water and sediment forecasting based on causal analysis feature selection","authors":"XuanZhao Kong ,&nbsp;KeBin Shi ,&nbsp;HaiKun Yao ,&nbsp;SiHai Liu","doi":"10.1016/j.ejrh.2025.102464","DOIUrl":"10.1016/j.ejrh.2025.102464","url":null,"abstract":"<div><h3>Study region</h3><div>Upper reaches of the Weigan River Basin, southern foothills of the Tianshan Mountains, Xinjiang, China</div></div><div><h3>Study focus</h3><div>The stochastic characteristics of water-sediment fluxes in the basin pose significant challenges for accurate mid-to-long term predictions. In complex hydrological systems, collected multivariate time series contain critical system features, yet traditional feature selection methods fail to consider inter-variable causal relationships. This study proposes an ensemble prediction model incorporating causal feature selection, developing 24 individual prediction models for water-sediment processes and selecting optimal combinations to enhance prediction accuracy for reservoir inflow water and sediment.</div></div><div><h3>New hydrological insights for the region</h3><div>By employing causal analysis instead of conventional feature selection and applying SHAP for LSTM gradient interpretation, we identify key causal variables: historical runoff, temperature, and evapotranspiration primarily govern reservoir inflow (with snowmelt being the dominant water source), while runoff, historical sediment load, and Nino1+2 most influence sediment inflow. For higher-quality runoff series, GRMIG-MCI-based models achieve KGE&gt;0.85, whereas PCMCI proves more suitable for noisy sediment series (KGE≈0.4). The monthly-weighted ensemble approach improves NSE by 0.35%–3.65% for runoff and over 20% for sediment compared to individual optimal models.</div></div>","PeriodicalId":48620,"journal":{"name":"Journal of Hydrology-Regional Studies","volume":"60 ","pages":"Article 102464"},"PeriodicalIF":4.7,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}