The attribution of subseasonal extreme precipitation in arid and semi-arid regions based on extended convergent cross mapping

Extreme precipitation (EP) events might have intensified in arid and semi-arid regions under climate change. However, complex circulation and land–atmosphere interactions limit our understanding of subseasonal precipitation causality. This study provides a comprehensive analysis of EP in the arid and semi-arid region of Northwest China (ASRNC), combining spatiotemporal trend analysis of EP indices with diagnostic insights into thermodynamic and dynamic precursors of regional EP events. We innovatively apply Extended Convergent Cross Mapping (ECCM) to reveal nonlinear and lagged causality between precipitation and both large-scale circulation (external drivers) and land–atmosphere interactions (internal drivers) at the subseasonal scale. The results show a significant increase in EPs frequency, intensity and extent from 1961 to 2023, particularly in mountainous areas, driven by anomalous upper-level thermal and lower-level dynamic conditions. Low-level jet streams transport moisture from the Indian Ocean, North Atlantic and others, fueling EP events in various subregions of the ASRNC. ECCM reveals more stable robust nonlinear lagged causality than traditional methods. External drivers influence EP through multifactorial interactions and teleconnections. Geopotential height and winds triggering signals up to 15 days ahead, while specific humidity, vertical velocity and outgoing longwave radiation enhance short-term convective activity and low-level moisture transport. Internal drivers regulate local moisture cycling. Additionally, surface soil moisture together with vapor pressure deficit is shown to have a stronger and shorter-term impact, while impacts of subsurface and deep soil moisture are weaker. These findings provide valuable insights for improving early warning capacity and informing operational strategies to manage hydro-climatic risks in climate-sensitive regions.