Preliminary landslide susceptibility zonation using GIS-based fuzzy logic in Vitoria, Brazil

This paper presents the application of geographic information system (GIS)-based fuzzy logic to infer landslide susceptibility for a preliminary mapping of Vitoria, Espirito Santo, Brazil. Input spatial layers derived from a digital elevation model, roads, land-use, and geotechnical type were classified, and previous geological accident occurrences were summarized for each class. The fuzzy membership functions and fuzzy rules were extracted by determining the landslide frequency ratio for each class of conditioning factors of elevation, slope, profile and plan curvature, aspect, distance to drainage lines, and distance to roads. The analytic hierarchy process was applied to rank classes of land-use and geotechnical type based on expert knowledge. The spatial layers were combined by fuzzy overlay procedures in a GIS and results were compared with a weighted mean method. The results from the fuzzy and weighted mean methods were also compared with previous geological-geotechnical risk maps, providing statistics to assess the best options from different viewpoints. One advantage of this method is that it enables spatial data layers with discrete and continuous data to be used as input in a fuzzy inference system, taking into account both knowledge of previous events and expert knowledge.