Scale-Dependent Transition in Soil Moisture Memory and Its Environmental Controls in Complex Mountain Terrain

Abstract. Soil moisture memory (SMM) defines the antecedent wetness states that modulate catchment responses to meteorological triggers, serving as a critical determinant of background hydraulic susceptibility. However, its multi-scale characteristics and environmental drivers remain poorly understood in complex terrain. This study characterizes SMM dynamics across daily-to-interannual scales using daily data (2003–2022) from three hazard-prone watersheds in southwestern China (Dali River, Anning River, and Jiangjia Ravine). By integrating Power Spectrum Analysis, Detrended Fluctuation Analysis (DFA-2), and a spatial attribution modeling framework, we identify a distinct scale-dependent transition in SMM persistence and its controls. Results revealed that while memory intensity generally weakened with scale, humid catchments exhibited a robust "inherent persistence" regime extending to multi-year scales. Crucially, feature importance analysis uncovered a structural transition at approximately the 5-year scale: atmospheric variables and vegetation dominated short-term variability, whereas soil properties and topography governed the system's long-term capacity to integrate low-frequency signals. Mechanistically, this marks a shift from event-driven hydraulic responses to background storage trends regulated by deep soil buffering. These findings provide a basis for distinguishing event-scale hydraulic preconditioning from long-term background susceptibility, offering a conceptual framework for incorporating operational persistence horizons into hierarchical hazard assessment strategies.