Nonlinear and seasonal response mechanisms of hydrological drought in coupled human and natural systems

Hydrological drought evolution in heavily disturbed basins is jointly driven by climatic forcing, basin characteristics, and human activities. However, the nonlinear, seasonal mechanisms underlying hydrological drought responses in coupled human and natural systems remain poorly understood. In this study, the Wei River Basin (WRB) was used to examine hydrological drought evolution under coupled climatic and anthropogenic disturbances. A Soil and Water Assessment Tool (SWAT)-based streamflow reconstruction framework, explainable artificial intelligence, nonlinear response analysis, and structural equation modelling were integrated to identify the multiple controls on hydrological drought evolution. The results indicated that hydrological drought progressively diverged from meteorological drought after the 1990s, while human-induced drought intensification became particularly evident after 2000. Explainable machine learning analysis revealed marked seasonal differences in the dominant drought drivers. Climatic factors remained the primary controls across all seasons, whereas human activities exerted greater influences in spring and ecohydrological buffering effects became increasingly important in summer and autumn. Lagged climatic effects indicated a greater dependence of winter hydrological drought on antecedent hydroclimatic conditions. Soil moisture served as a key intermediate variable linking meteorological anomalies to hydrological responses. Domestic water withdrawal generally intensified hydrological drought, with greater impacts in arid and urbanised regions, whereas grassland expansion partially mitigated drought impacts. Overall, hydrological drought in the WRB exhibited marked nonlinear, environmentally dependent, and seasonally divergent characteristics under coupled climatic and anthropogenic disturbances. These findings provide novel insights into hydrological drought evolution in coupled human–natural systems and inform integrated drought management in heavily disturbed basins.