Prediction of initiation and failure of unsaturated loess slopes using a sliding-block model

Loess deposits are widely distributed across the Chinese Loess Plateau, which are highly susceptible to landslides induced by moisture changes such as rainfall and irrigation. While saturated loess landslides have been extensively studied, the failure mechanisms and predictive models for unsaturated loess slopes remain challenging. This study analyzed the mechanical behavior of unsaturated loess through consolidated undrained triaxial compression tests on specimens with varying water contents and confining pressures. Microstructural variations were examined using scanning electron microscopy (SEM) and image analysis. A sliding-block model within a stress state framework is applied to compute the factors of safety for the initiation and failure states of an unsaturated loess slope. The model includes two key parameters of pore water pressure ratio and degree of saturation. Results show that the stress ratio is independent of saturation, whereas effective stress decreases significantly with increasing saturation. Empirical relationships are established to describe the relationship between the degree of saturation and parameter variations. The proposed sliding-block model successfully distinguishes between the initiation and failure states of loess landslides, as validated by two field case studies. In addition, moisture loading alters the effective stress state and microstructure of loess, and pore water pressure develops sufficiently only when saturation exceeds 60%. The developed criterion enables slope stability prediction associated with water content and pore pressure monitoring. This research offers a practical procedure for predicting unsaturated loess landslides by replacing matric suction with the degree of saturation as a key indicator. Our findings show that the model reduces reliance on matric suction measurements and provides an easier pathway to predict landslides induced by moisture loading on unsaturated loess and other soil slopes.