Application of Geographical information Systems (GIS) and discriminant analysis in modelling slope failure incidence in Pulau Pinang, Malaysia

Abstract

Slope failure is a phenomenon due to frequent rainfall that occurs in tropical areas such as Malaysia. This paper describes the use of Geographical Information Systems (GIS) and Discriminant Analysis to model the physical features of slope failure and the statistical association between slope failure events with physical parameters that contribute to the incidence of slope failure in Pulau Pinang. Discriminant analysis is an analysis method that can be used to discriminate against a set of slope failures based on certain criteria. The main purpose of this analysis were to understand the factors that affect the difference between the group of slope failure and subsequently making predictions about a possible slope failure. Therefore, a linear combination of independent variables has been formed and used as a basis for classifying certain slope failure cases. The study used ten variables: distance to the road, average annual rainfall, lithology, topography height, slope gradient, soil series, slope aspect, distance to river, landuse type and lineament. The resulting model was able to predict 92.5% of actual slope failure events. The model was validated using 30% of the actual incident samples and found 91.24% accuracy.