<p>Evacuation strategies play a crucial role in mitigating human casualties from geohazards. While evacuation simulations have been widely used to investigate crowd behavior in response to disasters such as fires and earthquakes, their application to investigating crowd behavior in response to geohazards in mountainous areas has been limited. In this study, a framework was developed for simulating and optimizing evacuation strategies in response to geohazards in mountainous areas that considers the behavioral characteristics of residents. First, a simulation scenario is constructed by analyzing satellite imagery of the region of interest to identify and classify various geographic features. Characteristic parameters are then embedded into a hybrid algorithm that combines the ant colony system algorithm with a social force model to simulate realistic evacuation scenarios that reflect crowd behavior during emergencies. Based on the results of numerical simulations, the existing configuration of shelter locations are optimized to address the chaos and congestion resulting from crowd behavior. As a case study, the proposed framework was applied to constructing geohazard scenarios for a community in the Longmen Mountains area of China and conducting numerical simulations to optimize the evacuation strategy. The results show that the optimized strategies helped facilitate the safe evacuation of residents. The proposed framework represents a multidisciplinary approach to developing evacuation strategies in response to geohazards in mountainous areas while considering crowd behavior. This research has practical implications for guiding public evacuations in mountain communities under the backdrop of geohazards and provides innovative solutions for crowd evacuations in similar scenarios.</p>