STATISTICAL-MODELS FOR SLOPE INSTABILITY CLASSIFICATION

Statistical models for reliable hazard assessment of landslides in a g iven area are presented. The models are based on information theory an d regression analysis. Information theory is a branch of probability i n which the most significant feature is unpredictability and regressio n analysis is based on simple statistical theory. Both methods of stat istical prediction have been applied to slope instability to provide c ross validation of the results. A FORTRAN 77 program for computer-aide d assessment of the landslide hazard has been developed. The program c onsiders the factors affecting the slope instability and history of pa st landslides of the area as its input. The program also calculates th e information and regression value in a given area and classifies the area into different grades of instability. A case study of landslide h azard assessment of a 66 km(2) area in parts of the AIkananda valley, the Gharwal Himalaya is presented. The factors considered in the analy sis are angle and height of the slope, rock type and geological struct ures such as faults and thrusts. The results obtained are illustrated in the form of landslide hazard maps. The predicted results are compar ed by the above methods.