Reservoir operation and irrigation drive contrasting responses in hydrological drought predictability across the Yellow River Basin

Against the backdrop of intensifying climate change, the predictability of hydrological drought—a devastating extreme event, is undergoing significant changes. However, the analysis of the spatiotemporal evolution laws of hydrological drought predictability and the quantitative contribution of human activities in existing research were still very limited. Here we developed a Mean Square Error Weighted Averaging Vine Copula (MSEWAVC) model to dynamically assess hydrological drought predictability in the drought-prone Yellow River Basin (YRB). The study showed: (1) Hydrological drought predictability exhibited marked spatial heterogeneity: the predictability of tributary areas was significantly better than that of main streams. The hydrological processes in the densely populated areas of water conservancy projects have become seriously “unnatural”. (2) Under different forecast periods, the overall hydrological drought predictability throughout the basin exhibited a weak upward trend, with rates of 0.0064/10w, 0.0080/10w, and 0.011/10w, respectively, possible enhancement of regional water regulation capacity. (3) Human activities exhibited contrasting effects on hydrological drought predictability across the Yellow River Basin: reservoir operation buffered the decline in the mainstream, while irrigation and land use change aggravated the decline in some tributaries. This study advanced the scientific basis for adaptive early warning of extreme hydrological drought and water resource zoning regulation in a changing environment.