Accurate runoff inversion is essential for watershed hydrology and water resources management, yet traditional single-model approaches often exhibit substantial biases in complex terrain and nonlinear flow regimes. This study presents a fusion modeling framework for the Jialing River Basin, combining three individual methods—the relationship fitting method, the improved Manning’s formula, and the C/M signal method—through weighted averaging, Bagging–AdaBoost, and Stacking ensemble strategies, with linear regression as a meta-learner for multi-source feature integration. Validation using monthly data from 2015–2020 shows that the Stacking-based model achieves superior performance (Nash–Sutcliffe efficiency = 0.981; RMSE = 38.69 m 3/s; RRMSE = 5.09%), outperforming both Bagging–AdaBoost and weighted averaging. The model effectively reduces systematic errors and variance, providing stable predictions across high- and low-flow regimes. These results demonstrate that ensemble-based fusion models significantly enhance runoff reconstruction and capture flow dynamics more accurately than individual approaches. The proposed framework offers a robust, scalable tool for hydrological estimation in complex basins and has important implications for water resources management and flood mitigation.

Climate and underlying surface changes have a profound impact on runoff in the Qinling Mountains. This study attempts to identify the difference in runoff changes of two rivers in the south and north of China’s south-north transitional zone under climate change. The Pettit test and Mann-Kendall test were used to investigate the abrupt change and trend analysis on runoff in the Ba River watershed and Jinqian River watershed from 1960 to 2014. The coupled energy-water balance equation based on the Budyko hypothesis estimated the climate and landscape elasticity of runoff followed by attribution analysis of runoff in these two watersheds. The results showed that annual runoff in the Jinqian River (in the southern Qinling Mountains) and the Ba River (in the northern Qinling Mountains) exhibited a significant decreasing trend at P<0.05 and P<0.01, respectively. Abrupt runoff changes occurred in 1989 and 1992 in the Ba River and Jinqian River, respectively. The attribution analysis showed that the change in potential evapotranspiration had little impact on runoff in the southern and northern Qinling Mountains. In contrast, the dominant factors leading to runoff reduction were the change in precipitation and catchment landscape. The contributions of climate change and land surface alteration to runoff changes in the Ba River watershed and Jinqian watershed were 38.08% and 61.92%, and 23.95% and 76.05%, respectively. This study can provide a scientific reference for water resource protection in the south-north transitional zone.

Investigation of the variations in runoff and sediment load as well as their dynamic relation is conducive to understanding hydrological regimes changes and supporting channel regulation and fluvial management. This study was undertaken in the Xihanshui catchment, which is known for its high sediment-laden in the Jialing River of the Yangtze River basin, southern China, to evaluate the change characteristics of runoff, sediment load and their relationship at multi-temporal scales from 1966 to 2016. The results showed that the monthly runoff changed significantly for more months whereas the significant changes in monthly sediment load occurred from April to September. The contributions of runoff in summer and autumn and sediment load in summer to their annual value changes were greater. The annual runoff and sediment load in the Xihanshui catchment both exhibited significant decreasing trends (P<0.05) with significant mutation in 1993 (P<0.05). The average annual runoff in the change period (1994-2016) decreased by 49.60% and annual sediment load displayed a substantial decline with a reduction of 77.76% in comparison with the reference period (1966-1993). The variation of the relationship between runoff and sediment load in the catchment was time-dependent. The annual and extreme monthly runoff-sediment relationship could be generally expressed as power function, whereas the monthly runoff-sediment relationships were changeable. Spatially, the relationship between annual runoff and sediment load could be partly attributed to sediment load changes in the upstream and runoff variations in the downstream and it became weaker in the change period due to the impact of existing soil and water conservation measures. Quantitative assessment showed that human activity played a dominant role in annual runoff and sediment load reduction, with the contributions of 67.07% and 87.64%, respectively.