多模型耦合的流域水土流失监测方法研究

濑溪河流域存在严重的水力侵蚀,同时伴随着水土流失分布点多线长面广的特点。而现有模型几乎只考虑了流域尺度上的坡面和河流侵蚀过程,或者是坡面和流域尺度的基本结合,导致对土壤侵蚀全过程的物理模拟产生高度不确定。为此提出不同尺度模型相互耦合的水土流失监测体系,从三级顺序"坡-沟-河"结构,全面反演流域水沙的时空动态迁移过程。基于多模型耦合体系的流域应用表明:①单一模型在濑溪河流域都表现出良好的适用性和准确性。RUSLE模型预测结果与实际侵蚀规律高度吻合,SWAT模型参数率定的纳什效率系数(NSE)和决定性系数(R²)均达到0.6以上,模拟预测结果鲁棒性较好。②不同侵蚀模型的特征反演相互关联程度都达到极显著相关水平。WEPP和SWAT模型间R²为0.96,RUSLE和WEPP模型间R²为0.77,RUSLE和SWAT模型间R²为0.58,分析表明坡面、细沟和河道尺度下的侵蚀物理过程是紧密耦合的。③多模型耦合的水土流失全过程风险评价较单一模型更为全面。通过对土壤侵蚀物理过程的空间耦合,实现流域土壤侵蚀、细沟冲刷和河道沉积等相关侵蚀过程的整体性评价,利于从不同尺度全面反演流域水沙输移的时空动态迁移过程和预测侵蚀风险发生规律。

The mechanisms and assessing methods for time lags in the responses of soil loss to best management practices

In order to study the multidimensional dynamic evolution of soil erosion under the influence of multi-factors and spatial variability of the subsurface, to realize high-precision simulation of water and sand erosion processes in the watershed, to comprehensively analyze the spatial and temporal distribution characteristics of water and sand migration, and to predict the occurrence of erosion risk. Taking the existence of serious hydraulic erosion in the Laixi watershed as an example, the basin was accompanied by a wide distribution of soil erosion with multiple points, long lines, and extensive coverage. The existing models almost exclusively considered slope and fluvial erosion processes at the watershed scale or a basic combination of hillslope and watershed scales. It led to high uncertainty in the physical simulation of the soil erosion process. This paper proposes a coupled soil erosion monitoring system with different scale models, which comprehensively reflects the spatial and temporal dynamic migration process of the watershed and sediment in the watershed from the subsurface to the fine channel on the slope. Then it migrates with the watershed to the river channel for deposition from the surface, line, and point scales, following a hierarchical structure of surface-slope-river. Based on the watershed application of the multi-model coupling system, the results showed that, firstly, a single model demonstrated good applicability and accuracy in the Laixi watershed. The RUSLE model predicted erosion patterns that closely matched actual observations. The Nash-Sutcliffe Efficiency (NSE) and the coefficient of determination (R²) obtained from parameter calibration of the SWAT model exceeded 0.6, indicating good robustness of the simulated results. Secondly, intercorrelations between the feature inversions of the different erosion models all reached highly significant levels of correlation. The R² between the WEPP and SWAT models was 0.96, between the RUSLE and WEPP models was 0.77, and between the RUSLE and SWAT models was 0.58, and the analyses indicated that the physical processes of erosion at the surface, slope, and channel scales were tightly coupled. Finally, the whole process risk evaluation of soil erosion with multi-model coupling was more comprehensive than a single model. It facilitated the study of the multidimensional dynamic evolution of soil erosion under the influence of multiple factors and spatial variability of the subsurface. This enabled high-precision simulation of water and sediment erosion processes in the watershed, comprehensive analysis of the spatiotemporal distribution characteristics of water and sediment transport, and prediction of the regular occurrence of erosion risk.