Study on geological disaster susceptibility assessment based on k-means clustering algorithm

With the acceleration of urbanization, the threat of geological disasters to urban safety has become increasingly severe, and precise susceptibility assessment has become crucial for territorial spatial planning and risk control. This study proposes a geological disaster susceptibility evaluation method based on the K-means clustering algorithm, constructing an assessment system with 12 indicators covering topographic features, geological structures, hydrometeorology, and human engineering activities. Taking Gui'an New Area as a case study, multi-source data (terrain slope, lithology distribution, annual rainfall, population density, etc.) are integrated for clustering analysis. By optimizing the selection of the cluster number K (determined as 5 using the silhouette coefficient method), the study area is divided into five susceptibility zones: extremely high, high, moderate, low, and extremely low. The results reveal that extremely high-susceptibility areas, accounting for 28.6% of the study area, are concentrated in karst-intensive regions such as Gaofeng Town and Machang Town, with an 89.3% coincidence rate with historical disaster points. This research provides a data-driven technical approach for the refined management of urban geological disasters, facilitating the dual control of "hidden hazard points and risk zones" in typical karst areas like Gui'an New Area, and promoting the collaborative optimization of geological safety and territorial spatial planning.