Prediction and mapping of soil thickness in alpine canyon regions based on whale optimization algorithm optimized random forest: A case study of Baihetan Reservoir area in China

Abstract

Accurate measurements of soil thickness are crucial for assessing landslide susceptibility, slope stability, and soil conservation. However, there is a relative scarcity of research on the spatial distribution of soil thickness in areas with complex terrains, such as alpine canyon regions. Given this research gap, the aim of this study is to develop a reliable method for predicting soil thickness in these regions. In this study, the Baihetan Reservoir area (China), characterized by typical alpine canyon regions, was selected as the research site. The slope index (SI) and slope (S) factor, in addition to other factors, were used to predict soil thickness. Subsequently, the random forest (RF) model and its version based on the whale optimization algorithm (WOA) were used to model soil thickness. The results showed that compared to the other models, the WOA-RF model, which considers the slope index factor, performed best in 100 tests, achieving the highest coefficient of determination (R² = 0.93) and the lowest root mean square error (RMSE = 5.6 m). Furthermore, the soil thickness data from the WOA-RF (SI) model displayed the highest congruence with the soil thickness data obtained from environmental noise measurements. Therefore, predicting soil thickness in alpine canyon regions by comprehensively considering environmental variables and using the WOA-RF model is feasible. The resulting soil thickness maps can serve as key fundamental inputs for further analysis.