Simulation and prediction of hydrological processes in Kaidu River Basin based on DHSVM model

The Kaidu River Basin in Xinjiang, China. This study is focused on simulating future climate and underlying surface changes using CMIP6 climate model data and the PLUS model, and modeling historical and future runoff in the study region using the Distributed Hydrology Soil Vegetation Model (DHSVM). The double cumulative curve method is applied to quantify the respective contributions of climate change and reservoir operations to runoff variation, and to reveal the mechanisms by which reservoir cluster regulation alters the spatiotemporal distribution of runoff. Our findings reveal that the PLUS model exhibits high reliability in simulating land use in the study region, providing accurate land surface inputs for hydrological modeling. The performance of the DHSVM model was significantly improved through parameter optimization, with the Nash–Sutcliffe efficiency (NSE) during the validation period increasing to 0.64 at the daily scale, and reaching 0.77 at the monthly scale. This confirms its suitability for simulating hydrological processes in small to medium-sized arid basins. The cascade reservoirs adopt a multi-stage winter-spring storage and summer-autumn coordinated release operation strategy, which shifts the natural runoff peak from June–July to July–August, effectively aligning peak water supply with agricultural water demand. Human activities, represented by reservoir operations, account for 20.30 % of the runoff variation, while climatic factors, primarily precipitation, contribute 79.70 %, highlighting the regulatory role of reservoirs in regional hydrological processes. • We simulated future runoff using validated DHSVM model with CMIP6 climate data and PLUS land use projections. • Climate factors dominated runoff impacts (79.70 %), while reservoir operations accounted for 20.30 %. • Cascade reservoirs regulate flow patterns, shifting peak runoff timing to match agricultural irrigation needs.