Landslide susceptibility assessment using the frequency ratio model in the Mae Chan River watershed, northern Thailand

Abstract

A landslide is a significant geological hazard that impacts society, the environment, and local infrastructures. The Mae Chan River watershed, a watershed that is surrounded by high erodible mountains, is particularly vulnerable to landslides. This research aimed to assess the susceptibility of landslides in the unique watershed using the frequency ratio approach. Landslide inventory data were 173 landslide scars in the mountainous region of the watershed. The data were extracted from satellite images that captured historical landslide scars on its surface. Landslide causative factors were selected based on causative elements observed in the watershed that potentially caused the previous landslide occurrences. These inventory data and causative factors were combined to create a landslide susceptibility index and classes. The analysis indicated that around 36 % of the entire watershed was highly prone to landslides, especially in the northwestern and southern high mountains. The remaining 43% and 21% of the watershed's area were classified as moderate and low landslide susceptibility classes, respectively. The landslide susceptibility data's accuracy, reliability, and predictability were verified using the area under the receiver operating characteristic (ROC) curve (AUC) analysis. The AUC values represented the success and prediction rates curve of 0.738 and 0.712, respectively, suggesting that the model performed reasonably well in identifying and predicting landslide susceptibility classes. The study highlights the qualification of landslide susceptibility mapping in a watershed in Thailand to other large-scale landslide hazard research. • One-third of the Mae Chan River watershed is highly prone to landslide occurrences. • High mountains in the northwestern and southern zones experience high landslide susceptibility levels. • The major causative factors include geology, geomorphology, and land modification. • The success and prediction rate curves suggest that the model is well-performing in identifying and predicting landslides. • Landslide susceptibility assessment is a reliable result for future landslide research.

Topics & Concepts