Developing a simple soil erosion model including spatial variability in headwater catchments

Abstract

Hillslopes and stream channels are tightly linked in headwater catchments, where soil erosion on slopes is transported to channels depending on topography, soil properties, and ground cover conditions. Because these environments are highly spatially variable, geographic information system (GIS)-based modeling approaches are effective for estimating soil erosion. This study therefore aimed to (1) examine the factors influencing ground cover conditions in relation to overstory structure and topography and (2) develop a simple model and test its applicability for estimating soil erosion in forested headwater catchments. The study was conducted in two catchments, No. 3 (7 ha) and No. 4 (5 ha), located in the Tanzawa Mountains of eastern Kanagawa prefecture, Japan. The Revised Universal Soil Loss Equation was applied using LiDAR-derived topographic data to estimate soil loss from 2010 to 2020. Estimated soil erosion in catchment No. 3 ranged from 0.32 to 3.47 ton ha⁻¹ year⁻¹ (mean 1.25), while values in catchment No. 4 ranged from 0.22 to 2.44 ton ha⁻¹ year⁻¹ (mean 0.85). The results showed that spatial variability, particularly in ground cover, is essential for parameterizing catchment-scale soil erosion models. The model achieved about 50% effectiveness during moderate rainfall and long monitoring periods but underestimated erosion by 31%–48% during short periods with intense rainfall. Overall, the study demonstrates that including spatially distributed understory conditions shaped by canopy structure and topography improves soil erosion modeling. The proposed model offers practical value for revising soil and water conservation guidelines. These findings highlight the importance of integrating fine-scale vegetation and terrain dynamics in future erosion assessments and monitoring.