Effect of rainfall distributions on landside susceptibility mapping using frequency ratio method

Abstract

Landslide is one of the most common disasters in the world, especially in tropical monsoon climate regions like Vietnam. In this area, the effects of rainfall and its distribution to landslide are significant to occurrence of landslides. This paper is more likely to focus on evaluating the influences of rainfall and its distribution on landslide susceptibility by using the Frequency Ratio(FR) statistical model for a case study of six mountainous districts in Quang Ngai province. The rainfall distributions which include annual rainfall and antecedent rainfall were applied to achieve the research. The other landslide causative factors such as elevation, slope, aspect, soil types, and land use are also taken into consideration in creating landslide susceptibility maps. This FR model was trained and validated based on 445 landslide sites using Arc GIS software. The AUC (Area Under the Curve) and LD (Landslide Density) indexes were used for evaluating the performance of the FR model. Obtained results show that developed landslide susceptibility map based on 3 days antecedent rainfall data (P= 50%) get better AUC and LD indexes than using annual rainfall cases.