Landslide susceptibility zonation mapping using frequency ratio and fuzzy gamma operator models in part of NH-39, Manipur, India

The hilly region of Manipur especially along the NH-39 road, which is the lifeline of the State, is prone to landslides every year particularly during the monsoon season. Anthropological factors, such as excessive deforestation, unsystematic changes in land use and land cover pattern and slope cultivation, etc. are indirectly initiate the process of landslides. In the present study, landslide susceptibility mapping was carried out using frequency ratio and fuzzy gamma operator models with the help of geomatics techniques. The landslide susceptibility mapping was prepared using landslide inventory data and nine landslide causative factors, i.e. lithology, land use and land cover, geomorphology, drainage density, lineament density, slope gradient, slope aspect, curvature, and elevation. These causative factors were prepared with the help of toposheet, high resolution IRS P6 LISS IV satellite imagery, cartosat DEM data and extensive field work. The landslide susceptibility maps were prepared by calculating the relationship between the landslide causative parameters with landslide areas using a frequency ratio model. To get the fuzzy membership values, the frequency ratio values were normalized between the ranges of 0 and 1. The landslide susceptibility maps were compared and prediction accuracy of both the models was derived using the area under curve (AUC) method. The success rate curves were obtained using both training and all landslide inventory dataset. For training landslide inventory dataset, the AUC value of the success rate curve for the frequency ratio model was found to be 0.8056, whereas for the fuzzy gamma operator (using γ = 0.99) model, it was calculated as 0.9150. In the case of all landslide inventory dataset, the AUC value of the success rate curve for the frequency ratio model and the fuzzy gamma operator model were 0.7921 and 0.8188, respectively. The landslide susceptibility index was also compared with the landslide validation inventory dataset to obtain the prediction rate curves. The AUC value of the prediction rate curve for the frequency ratio model was 0.5681, whereas in the case of the fuzzy gamma operator model, it was 0.6721.