Abstract: The physical essence of Soil Hydrological Memory (SHM) was systematically reviewed, elucidating its role as an antecedent state that nonlinearly modulates high-frequency meteorological forcings, and the methodological paradigm shift was traced from traditional linear statistics to multi-source data assimilation and physics-informed machine learning. . Focusing on the increasing compound drought-heatwave extreme event, we highlight the core driving and spatial propagation mechanisms of SHM. The authors particularly emphasize the "master switch" role of antecedent root-zone hydrological memory in determining the state-dependent mechanistic shift of ecosystems across the Budyko framework (from water-limited to energy-driven regimes) and the emergent dual decoupling phenomenon. Furthermore, we analyze the structural bottlenecks of current Earth system models (ESMs), specifically the memoryless static water stress functions ( \beta factor), the absence of deep root-zone processes and the lack of two-waycoupling for anthropogenic water use (e.g., irrigation). Guided by classical theories, future research must prioritize developing memory-aware dynamic parameterizations, reconstructing global 3D root-zone hydrological memory networks, and quantifying the cascading effects of hydrological memory reshaped by human activities. These efforts will enhance the reliability of extreme climate prediction, carbon sink evaluation,and regional water resource risk management.