Quantifying the impact of climate change and human activities on runoff changes is beneficial for developing sustainable water-management strategies within the local ecosystem. Machine-learning models were widely used in scientific research; yet, whether it is applicable for quantifying the contribution of climate change and human activities to runoff changes is not well understood. To provide a new pathway, we quantified the contribution of climate change and human activities to runoff changes using a machine-learning method (random forest model) in two semi-humid basins in this study. Results show that the random forest model provides good performances for runoff simulation; the contributions of climate change and human activities to runoff changes from 1982 to 2014 were found between 6–9% and 91–94% in the Zijinguan basin, and 31–44% and 56–69% in the Daomaguan basin, respectively. Furthermore, the model performances were also compared with those of well-known elasticity-based and double-mass curve methods, and the results of these models are approximate in the investigated basins, which implies that the random forest model has the potential for runoff simulation and for quantifying the impact of climate change and human activities on runoff changes. This study provides a new methodology for studying the impact of climate change and human activities on runoff changes, and the limited numbers of parameters make this methodology important for further applications to other basins elsewhere. Nevertheless, the physical interpretation should be made with caution and more comprehensive comparison work must be performed to assess the model’s applicability.

The multi-model ensemble (MME) approaches are highly regarded in climate prediction and risk assessment for their capacity to integrate multiple global climate models (GCMs) and minimize uncertainties associated with individual models. However, the quantitative impacts of spatial scale, weighted ensemble, and bias correction on the spatiotemporal comprehensive performance of MME remain unknown. In this study, we comprehensively assessed the historical simulation capabilities of 41 CMIP6 GCMs at national, basin, and grid scales. Additionally, we investigated the impact of bias correction and weighted ensemble on enhancing climate simulation performance. The results indicate that CMIP6 models exhibit notable differences in simulating regional climate characteristics of China across different scales. Weighted multi-model ensemble schemes incorporating better-performing models consistently outperform equal-weight approaches, achieving an average 20.67% reduction in the DISO (distance between indices of simulation and observation) index, with temporal performance improvements being particularly pronounced. Bias correction played a critical role in the enhancement of MMEs, reducing DISO values by 41.60% on average, particularly in the spatial dimension. Among all MMEs, the grid-scale optimized ensemble (GBQ), combining bias correction, model selection, and performance-based weighting, demonstrated superior comprehensive performance, achieving the lowest DISO values across spatial and temporal dimensions. These findings provide new insights for enhancing regional climate simulation and evaluation, and they provide more reliable scientific information for investigating climate change and formulating adaptation strategies in China.

The complementary relationship (CR) between actual and potential evaporation has undergone rapid development over the past decades. Commonly, the evaluation of CR models does not comprehensively cover spatial and temporal aspects of model performance; thus, the spatial and temporal accuracy and parameter sensitivity of different CR models remain relatively unknown, especially in ungauged basins. We examine the spatial and temporal performance and parameter sensitivity of four CR models with fixed parameters at the monthly scale over the source region of the Yellow River in China. Additionally, two CR models with distributed parameters were selected as comparisons. Because of the lack of evaporation spatiotemporal “true” values, the corrected water‐balance‐derived evaporation was used as the temporal reference, and the ensemble mean of multiple evaporation products was used as the spatial reference. We find that trade‐offs between the spatial and temporal accuracy of CR models with fixed parameters for basin evaporation should be considered in the application. Although four CR models with fixed parameters exhibited different spatial and temporal performances, their overall performance was generally similar. The parameters of CR models had consistent spatial and temporal sensitivities, and the sensitivity of parameter α_e was stronger than that of other parameters. This study demonstrates that CR models with distributed parameters have a promising future in estimating basin or regional evaporation. An integrated spatial and temporal evaluation framework can better distinguish the performance of CR models. The methodologies can be used as a guideline for evaluating and calibrating CR models that match an application's needs.