Spatial Zonation of Landslide Prone Area Using Information Value in the Geologically Fragile Region of Samdrup Jongkhar-Tashigang National Highway in Bhutan

Abstract

Samdrup Jongkhar-Tashigang National Highway (SJ-TG NH) in Bhutan experiences several landslides every year. However, there are no studies on the landslides which will assist in highway realignment. This study developed the landslide susceptibility mapping (LSM) using the information value (IV) and check the reliability of the IV. The workflow consists of landslide inventory, factor preparation, LSM development, and its validation. During the landslide inventory, a total of 130 landslides were identified from satellite image interpretation, google earth image, and field investigation. The landslide inventory was divided into a training dataset (70%) and a validation dataset (30%). Then, nine factors were used to construct a spatial database. The accuracy was conducted using the area under curve (AUC) and the reliability of the model was performed using the kappa index. The AUC for the success rate (0.7700) falls under a good category and the prediction rate (0.6798) falls under the moderate category. The kappa index (0.3407) for the IV falls under the fair reliability category. The LSM was classified into very safe (16.42%), safe (30.64%), moderately (27.67%), risky (16.18%), and high risky zones (9.09%) based on the natural break. The LSM will guide decision-makers in the realignment of the road.

Keywords

• information value
• kappa index
• area under the curve