Abstract
The effectiveness of using vegetation to reinforce slopes is influenced by the soil and vegetation characteristics. Hence, this study pioneers the construction of an extensive soil database using random forest machine learning and ordinary kriging methods, focusing on the influence of plant roots on the saturated and unsaturated properties of residual soils. Soil organic content, which includes contributions from both soil organisms and roots, functions as a key factor in estimating soil hydraulic and mechanical properties influenced by vegetation roots. This innovative approach of using organic content to estimate soil properties performs well when applied to machine learning models for soil database development. The results reveal that organic content markedly affects the hydraulic properties of soils, more than their mechanical properties. The finding illustrates the importance of exploring the hydraulic effects of vegetation on slope stability in addition to the traditional emphasis on mechanical reinforcement. This rooted soil database has practical applications in GIS-based analyses for mapping regional slope stability, incorporating the role of plant roots. A case study demonstrated the database's utility, showcasing that vegetation effectively limited rainwater infiltration and improved slope stability. Therefore, this research offers a valuable approach to improving slope stability through informed vegetation strategies.