ABSTRACT
A landslide is a very dangerous phenomenon which increases the chance of loss of property and livelihood. In a hilly region, landslide occurrence is the main barrier to human civilization in the sense of development. A mountainous region like East Sikkim also faces this type of acute problem, which destroys natural resources as well as property. Different types of causative factors are associated with this type of slope instability. In recent studies, researchers tried to investigate this environmental problem by using a suitable model. The development of the model and selection from it as to which one is most optimal is the main thrust area of research. In this study, hybrid computational approaches like hybrid biogeography based optimization (BBO) techniques are considered for landslide susceptibility modeling by considering the different causal factors. A landslide inventory map has been prepared with the basis of the landslide locations obtained from high-resolution satellite data and primary field observation. Non-landslide locations were also identified in the same manner for this model. All the information regarding the landslide and non-landslide are randomly split into a 70:30 ratio. The landslide susceptibility map from the hybrid BBO is associated with high accuracy, and that is determined by the area under the curve (AUC) values from a receiver operating characteristic curve. The AUC values in this outcome for training and validation data are 92.37% and 89.79% respectively.
