Improving the understanding of the future hydrological drought variability in cascade reservoirs area is indispensable for water resource management, especially in the context of changing climate. This study aims to investigate the hydrological drought variations in the upper Yellow River Basin (UYRB) during 2021–2100 employing the latest Coupled Model Intercomparison Project Phase 6 (CMIP6) datasets. The study area is divided into three sub-regions based on the location of the two large reservoirs, i.e., Region 1 (R1) and Region 2 (R2) have reservoirs while Region 3 (R3) has none. The meteorological data from 10 CMIP6 models are downscaled by utilizing the bias-corrected spatial disaggregation (BCSD) method and then used to force the improved abcd hydrological model for future monthly runoff simulations. The changes in severity, duration, and frequency of drought events in three sub-regions under two future climate scenarios (SSP2-4.5 and SSP5-8.5) relative to the historical period are identified. Then, the copula function is applied to explore the changes in probabilities of spatially concurrent drought for different cases in the future. The most likely combination of drought intensity for R1 and R2 corresponding to 10-years conditional return periods is estimated to demonstrate changes in drought risks across the reservoir areas. Results show that the drought frequency of R2 and R3 is expected to increase in spring and summer, and the drought events are projected to have a shorter duration and lower severity overall under two scenarios. Additionally, the spring probabilities of concurrent drought for different cases are expected to increase in the near future (increase by 1.9%–19.3%), but only increase in R2&R3 case in the far future. It is noted that although the drought probabilities will decrease in winter, R1 and R2 are expected to experience more severe drought when drought occurs in R3. The findings are helpful to improve the understanding of future hydrological drought risk in the UYRB and provide guidelines for making adaptation strategies to climate change.

提高对梯级水库区未来水文干旱变率的认识对于水资源管理是必不可少的，尤其是在气候变化的背景下。本研究旨在利用最新的耦合模型比对项目第 6 期 (CMIP6) 数据集调查 2021-2100 年黄河流域上游 (UYRB) 的水文干旱变化。研究区根据两个大型储层的位置划分为三个子区域，即区域1（R1）和区域2（R2）有储层，而区域3（R3）没有储层。来自 10 个 CMIP6 模型的气象数据通过使用偏差校正空间分解 (BCSD) 方法进行缩减，然后用于强制改进的 abcd 水文模型用于未来的每月径流模拟。严重程度、持续时间的变化，确定了相对于历史时期的两种未来气候情景（SSP2-4.5 和 SSP5-8.5）下三个子区域的干旱事件和频率。然后，应用 copula 函数来探索未来不同情况下空间并发干旱概率的变化。估计对应于 10 年有条件重现期的 R1 和 R2 干旱强度的最可能组合表明了整个水库区干旱风险的变化。结果表明，R2和R3的干旱频率在春季和夏季预计会增加，并且在两种情景下预计干旱事件的持续时间更短，整体严重程度更低。此外，预计不同情况下春季同时发生干旱的概率在不久的将来会增加（增加 1. 9%–19.3%），但在遥远的未来只会增加 R2&R3 案例。值得注意的是，虽然冬季干旱概率会降低，但当 R3 发生干旱时，预计 R1 和 R2 将经历更严重的干旱。研究结果有助于提高对 UYRB 未来水文干旱风险的认识，并为制定气候变化适应战略提供指导。