Bài báo khoa học

Landslides are a serious natural disaster that greatly affects people's lives and the natural environment. This study surveyed the current status and assessed and zonated landslide risks in 4 mountainous districts in the West of Nghe An province: Ky Son, Tuong Duong, Que Phong and Quy Chau. This study integrated remote sensing and GIS in machine learning methods, quickly and accurately updating data on landslide factors for the study area. Although the SVM, RF and LR models showed relative effectiveness in landslide susceptibility of the study area, RF was the optimal algorithm in this case study with the best OA and Kappa values. The assessment and survey results showed the severity of landslides in this area. The density of landslides was relatively high, the scale of landslides was very large, the length of the landslide block foot was up to more than 150 m, seriously affecting property, life and the ecological environment. The results of landslide risk zoning provide management agencies with a database to plan residential and production areas and have solutions to respond to this natural disaster during the rainy season.