Abstract:
The objective of this study is to produce a landslide susceptibility map of the Kitchakut range, Chanthaburi using a combination of rare events logistic regression and geographic information system as analyzing tools. The dependent variable is binary of historical landslide occurrence variables are the related factors extracted from digital elevation model and from geology, forest, landuse and soil maps.The landslides in the study area can be regarded as "rare events" because the landslide occurrences in the study area are twelve times fewer than non-occurrences. Rare events logistic regression, different from ordinary logistic regression, incorporates three correction measures : the endogenous stratified sampling of the dataset, the prior correction of the intercept and the correction of probabilities to include the estimation uncertainty.After ecxluding non-significant independent variables, the analytical results show that elevation, rock type, aspect, slope and fault are the most important predictive variables to a landslide model in the study area, respectively. Minus twice the log of the likelihood (-2LL) and Pseudo were used to validate the model. Both show a good agreement between the observed and predicted values of the validation dataset. The resulted landslide susceptibility map was classified into four classes: very high, high, moderate and low susceptibility, covering 2%, 19%, 19% and 60% of the total study area, respectively.