抄録
Soil erosion and land degradation are coupled, cross-scale processes arising from interactions among Earth-surface dynamics, environmental forcing, human activities, and ecological feedbacks. Despite extensive advances in erosion-process characterization, monitoring, modeling, and ecological management, a coherent framework for resolving multiprocess coupling, cross-scale transitions, driver attribution, and feedbacks within social–ecological systems remains underdeveloped. This review synthesizes recent progress in hydraulic, aeolian, gravitational, and compound erosion, together with advances and methodological limitations in field observations and tracer techniques, remote sensing and geographic information systems, empirical and process-based modeling, and artificial intelligence. Particular attention is given to how erosion processes propagate across spatial and temporal scales, interact with land degradation, and respond to global environmental change and human interventions. We further develop an integrated framework of “process observation–heterogeneity identification–mechanistic interpretation–risk prediction–management feedback,” which connects multisource observations, process-based models, data-driven approaches, and adaptive management through an “observation–interpretation–prediction–decision” chain. The framework identifies multiprocess interactions, cross-scale transitions, threshold responses, uncertainty propagation, and recovery capacity as key frontiers for advancing erosion–degradation research. The central contribution is to reposition soil erosion and land degradation from separately assessed processes toward an integrated, process-based social–ecological system, providing a conceptual and methodological basis for cross-scale prediction, resilience-oriented assessment, and adaptive land management.
