Exploring the spatial distribution patterns of loess landslides and identifying the key driving factors in typical regions of the upper Yellow River, China

Loess landslides are severe regional disasters in the upper reaches of the Yellow River, significantly affecting the ecological environment, land planning, and human habitation. Gaolan County was selected for this study. SPSS was used for collinearity testing of evaluation factors, and Extra Trees (ET), Random Forest (RF), and Gradient Boosting (GB) models were applied to calculate landslide susceptibility and factor weights. The model’s accuracy was assessed using the receiver operating characteristic (ROC) curve. Fourteen factors were identified for susceptibility evaluation, and the ROC values for ET, RF, and GB were 0.910, 0.906, and 0.895, respectively. The areas of very high, high, moderate, low, and very low susceptibility zones were 6.56%, 7.63%, 16.72%, 32.26%, and 36.82%, respectively. High to very high susceptibility zones were primarily located in the southern part of the study area, closely linked to its proximity to Lanzhou. The main driving factors of landslides were distance to road, annual rainfall, and elevation, with contribution values of 17.80%, 14.62%, and 13.18%, respectively. As Lanzhou expands, the study area should focus on the impact of human activities like road construction and land reclamation on landslides. This research provides a basis for scientific decision-making and regional land use planning.