Bayesian Network-Based Landslide Susceptibility Safe Route Assessment in the Face of Uncertain Knowledge and Various Information

Landslides are major natural hazards that pose significant threats to life, property, and economic stability, particularly in vulnerable regions such as China. Among various approaches for landslide prevention and control, landslide susceptibility assessment is the most commonly employed measure. However, this approach alone is insufficient. A critical issue remains: how to effectively and accurately use landslide susceptibility assessment results to identify safe evacuation or rescue routes in emergency situations when landslides occur. Unfortunately, little research has addressed this issue in depth. In this study, we propose an integrated framework that combines a Bayesian network (BN)-based landslide susceptibility map (LSM) with an improved A* algorithm for safe and efficient route planning. The BN model effectively integrates multi-source data and uncertainty knowledge to generate an accurate LSM, while the improved A* algorithm combines safety and efficiency considerations to optimize routes. The main contributions of this study include identifying the key factors influencing landslide hazards, developing a robust landslide susceptibility map (LSM), and improving the A* algorithm to ensure safe route planning in landslide-prone areas. Experimental results demonstrate that the improved A* algorithm significantly outperforms the original method by providing safer and more efficient routes. The proposed framework offers an integrated approach to landslide susceptibility assessment and emergency relief, providing valuable insights for strengthening early warning systems in high-risk areas and reducing economic losses and casualties.