The interaction mechanism between the dynamically changing environment and the subsequent natural hazards in the years following an earthquake has been studied based on a multi-temporal remote sensing dataset and continually measured data. However, how to quantitatively model this interaction mechanism has not been fully addressed. In this study, we first modeled the interaction mechanism between the ‘rainfall-landslide-flash flood’ disaster chain and the dynamically changing environment in mountainous areas by incorporating the SLope-Infiltration-Distributed Equilibrium (SLIDE) model within the landscape evolution model (CAESAR-Lisflood) and then applied the integrated CAESAR-Lisflood model in a Wenchuan earthquake-stricken area. The results demonstrated that the landslide susceptibility under extreme rainfall can be predicted effectively based on the integrated CAESAR-Lisflood model and that new/enlarged landslides occur more easily in mountain valleys, near the valley outlet, and in the main steep gullies. Most of the high landslide susceptibility areas are not located in coseismic landslide areas with high vegetation recovery. The landslide legacy effects had a significant influence on the landscape erosion and deposition processes and a great effect on the spatial distribution pattern of the material redistribution in the basin, which in turn affected the subsequent disaster occurrence. The integrated CAESAR-Lisflood model compensates for the effect of the “rainfall-landslide-flash flood” disaster chain on the process of erosion and deposition, improves the model’s applicability in earthquake-stricken areas, and provides scientific information for regional disaster management and reduction.