Two determination models of slope failure pattern based on the rainfall intensity–duration early warning threshold

This work attempts to consider the prediction of slope failure patterns and failure depth in the slope failure early warning system. For this goal, 4,752 scenarios of slope instability with 11 kinds of soil properties under 432 designed intensity–duration (I-D) conditions are simulated to discuss the influences of rainfall conditions and soil shear strength parameters on the slope failure pattern and failure depth. Based on the simulation results of the slope failure depths and failure modes of 198 scenarios of slope failure, it is found that the four slope failure modes (Mode I, Mode II, Mode III, and Mode IV) are not distributed in a disorderly manner, but are concentrated in the four intervals of the I-D early warning threshold. Moreover, with the increase in soil effective cohesion and internal friction angle, the I-D early warning threshold moves up and to the right in the I-D two-dimensional plane. Furthermore, for the non-cohesive soil, the slope failure depths are concentrated in three regions with the generation of two jump points (at near 25 mm/h and 15 mm/h in this study) between the three regions. At these two jump points, a slight increase in rainfall intensity will result in a sharp decrease in slope failure depth, while for the cohesive soils, with the increase in soil effective cohesion, the two jump points gradually disappeared. Finally, based on the results of parametric analysis, two determination models of slope failure pattern are proposed, i.e., the “Boot Model” for cohesive soils and the “Eggplant Model” for non-cohesive soils. The findings in this study will promote the prediction of slope failure patterns and failure depth under different rainfall conditions and the assessment of the impact scope of collapsed soil.