Evaluation of spatial probability of landslides using bivariate and multivariate approaches in the Goriganga valley, Kumaun Himalaya, India

In the present study, landslide susceptibility mapping in the Goriganga valley, Kumaun Himalaya has been carried out using bivariate and multivariate approaches. Bivariate analysis includes Yule’s Coefficient, Frequency Ratio, Information Value, and Weight of Evidence, whereas multivariate analysis used is Artificial Neural Network. The input data used for this purpose include an inventory of 421 active landslides and twelve possible causative factors of landslides like lithology, slope angle, slope aspect, elevation, curvature-plan, curvature-profile, distance to drainage, road & thrusts, land use and land cover. Rainfall pattern and the peak ground acceleration (PGA) of area were also considered for the analysis. Using both the bivariate and multivariate methods, it has been observed that ~20–25% of the study area lies in the high and very high landslide susceptible zones, whereas ~50–63% of the study area lies in the low and very low susceptible zone. The high and very high landlslide susceptible zones are mainly confined in the Lesser Himalaya and along the Goriganga River, whereas low and very low susceptible zones are mainly located in the Higher Himalaya, Tethys Himalaya, and the higher elevation of the Lesser Himalaya. Further all the four bivariate methods indicate success rate between 84 and 86%, and the prediction rate between 80 and 86%, and increase with the application of the ANN.