Ecosystems in arid regions are experiencing increasing water stress. In existing studies, it is difficult to describe ecological drought evolution and its nonlinear drivers under spatial autocorrelation. In this study, we established the standardized ecological water deficit index (SEWDI). Using a CatBoost–GeoShapley framework, we analysed ecological drought in northern China from 1982 to 2022. The results revealed a clear overall mitigation trend at a rate of 0.0211/a. A west‒east dipole pattern was observed. Notably, a wetting trend prevailed in the west, whereas localized aridification trends emerged in the east. The temporal signal was dominated by significant high-frequency oscillations with periods of 1–2 years. Attribution analysis suggested that climate factors constituted the primary drivers, although their influence varied by region. In West China (WC), precipitation climatology was the primary constraint, and its influence exhibited marginal saturation effects. In contrast, warming served as the main drought-enhancing factor in southern Northeast China (SNEC). Anthropogenic impacts were twofold. Land use changes helped reduce drought in Central China (CT). Conversely, intense irrigation in East China (EC) signaled concentrated water stress. By explicitly considering spatial effects, this study offers practical insights for ecological restoration and adaptive water resource management.
Dong et al. (Tue,) studied this question.