Choosing the Right Horizontal Resolution for Gully Erosion Susceptibility Mapping Using Machine Learning Algorithms: A Case in Highly Complex Terrain

Gully erosion susceptibility (GES) maps are essential for managing land resources and control. Choosing the optimal horizontal resolution in GES mapping is a challenge. In this study, complex loess hilly area on Chinese Loess Plateau was tested using two machine learning algorithms. Unmanned aerial vehicle (UAV) images with 9 cm GNSS RTK field-measured data were employed as base datasets, 11 factors used models. A series of resolutions, from 0.5–30 m, to determine which level how influenced mapping. The results showed that 2.5–5 m area, both random forest (RF) extreme gradient-boosting (XGBoost) algorithms study. High resolutions overestimated probability gully stable regions, it became difficult identify non-gully regions at too-coarse resolutions. variable importance changed varied among variables. Slope gradient, use, contributing were, general, three most critical factors. Land use remained an important factor all levels. slope gradient underestimated coarse (10–30 m), comparatively fine (0.5–1 m). This study provides reference selecting mapping, thus, offers support approaches attempting map gullies UAV.