Landslide susceptibility analysis using frequency ratio and weight of evidence approaches along the Lakhandehi Khola watershed in the Sarlahi District, southern Nepal

Abstract

Landslide susceptibility maps are considered as one of the most important keys to limiting and dodging potential landslide consequences worldwide. In the present study, landslide susceptibility maps are prepared using bivariate models: frequency ratio and weight of evidence approaches. At first, randomly selected 80% of landslides i.e.,one hundred eighty landslides are used as training data for the preparation of the model, and the rest 20% of landslides i.e.,forty-five landslides for its validation. Similarly,thematic layers of nine causative factors of landslides such as slope, aspect, curvature, stream density, TWI(Topographic Wetness Index), land use, geology, distance from river and distance from the road have been analyzed for the modeling in ArcGIS. Finally, prepared landslide susceptibility maps are classified into five classes from Very Low to Very High from both methods. The area of Low, Moderate, High, and Very High susceptible classes is also nearly equal. The success rate curve of FR (Frequency Ratio), and WOE (Weight of the Evidence), show accuracy of 71.09%, and 75.62% respectively. Likewise, the prediction rate curve shows 72.87% and 76.66% accuracy on FR and WOE methods respectively. Since the susceptibility maps prepared through both approaches show an accuracy of >70%, the result is deliberated as fair. These maps are useful to all the stakeholders for land use planning and developing mitigation strategies against the consequences of increasing landslides in the Siwalik Hills of Nepal.