Uncovering nonlinear causal relationships and propagation dynamics of drought types in Xinjiang using convergent...

Drought is one of the most destructive natural disasters globally. Understanding its propagation mechanisms and the causal relationships among different drought types is crucial for effective monitoring and mitigation. Using meteorological (SPI), hydrological (SRI), and agricultural (SSMI) drought indices from 1983 to 2023 in Xinjiang, this study employs the Convergent Cross Mapping (CCM) method to systematically quantify nonlinear causal relationships among the three drought types, revealing their temporal lag characteristics, spatial heterogeneity, and multiscale dynamics. The results show that: (1) The drought transmission chain in Xinjiang is stable, following a unidirectional hierarchical sequence of "SPI→SRI→SSMI", with clear master-slave coupling characteristics. (2) Drought propagation exhibits significant lag effects that vary by timescale. The SPI→SRI pathway shows the longest delay (up to 12 months), while the SPI→SSMI pathway demonstrates shorter, phased, and more sensitive responses. (3) The spatial distribution of causal strength is markedly heterogeneous, generally exhibiting a "low in the north, high in the south" pattern. High-intensity areas are concentrated in the Tarim Basin, while lower intensities are observed in the Tianshan and Altai Mountains. (4) The strength of causal links in drought propagation is jointly influenced by natural and anthropogenic drivers. Elevation, snow cover, temperature, NDVI, and LUCC play synergistic roles, reflecting a transition from natural-dominant to coupled natural-human driving mechanisms. These findings provide a theoretical foundation for improving drought risk assessment and understanding transmission mechanisms in arid regions.