Landslides Susceptibility Analysis in Blumenau, Southern Brazil: a Probabilistic Approach

抄録

The translational landslides result from the combination of favorable conditions when one assumes the existence of functional relations between the spatial distribution of determining factors and processes. Thus, it is possible calculating the probability of occurring these landslides, as well as generating organized maps of susceptibility classes based on the inventory of slip scars and on the variables able to influence their occurrence. Accordingly, the aim of the present study is to analyze the spatial relations between the occurred landslides and a set of potential determining factors, as well as to propose and test susceptibility models based on this analysis. The application of Bayes' theorem of conditional probability using the weights-of-evidence technique allowed analyzing the spatial relations between a set of potential determining factors (geological, geomorphological and soil use and cover factors) and 294 translational landslides in 2008 in the study area. Fifteen different integration schemes of the weighted values, able to reflect the spatial association of each factor in relation to the landslides, were elaborated. The adjustment concerning the spatial independence between the themes and the classification and prediction efficiency of each scheme were analyzed. Three models have met the conditional independence requirements. It was possible identifying that the high-to-very-high susceptibility conditions have varied from 15% (in the model integrating the values of themes such as slope, plan curvature and distance to lineament) to 28% (in the model integrating values of themes such as slope and profile curvature) in Blumenau County－Santa Catarina State, based on the selected models.