Spatiotemporal evolution of groundwater drought and multi-type drought propagation responses in the Yellow River source region

Existing studies often identify drought propagation time using correlation analysis or fixed-lag methods, which makes it difficult to reveal directional predictive relationships among drought types and to assess the stability and uncertainty of lagged response windows. To address these gaps, this study focuses on the Yellow River source region. Based on the spatiotemporal evolution of groundwater drought, directional prediction tests and Bootstrap-based lag window identification were used to examine the propagation direction and response time among Standardized Precipitation Index (SPI), Standardized Runoff Index (SRI), and Drought Severity Index (DSI), with Pearson lagged correlation used for supplementary validation. The results show that: (1) the estimated groundwater storage anomalies (GWSA) agrees well with in-situ groundwater observations, and groundwater drought in the Yellow River source region showed clear stage-wise and spatial differences during 2003–2022, shifting from nearly drought-free conditions to more frequent and severe droughts; (2) clear directional predictive relationships and distinct lagged responses exist among different drought types, and historical groundwater drought information can significantly improve the prediction of surface water drought; and (3) meteorological and surface water droughts show a relatively synchronous lagged response, whereas surface water and groundwater droughts exhibit clear asynchronous changes during the study period.