Abstract:

Abstract This paper presents the application of a multi-criteria analysis (MCA) tool for landslide susceptibility assessment in Porto Alegre municipality, southern Brazil. A knowledge driven approach was used, aiming to ensure an optimal use of the available information. The landslide conditioning factors considered were slope, lithology, flow accumulation and distance from lineaments. Standardization of these factors was done through fuzzy membership functions, and evaluation of their relative importance for landslide predisposition was supported by the analytic hierarchy process (AHP), based on local expert knowledge. Finally, factors were integrated in a GIS environment using the weighted linear combination (WLC) method. For validation, an inventory, including 107 landslide points recorded between 2007 and 2013 was used. Results indicated that 8.2% (39.40 km²) of the study area are highly and very highly susceptible to landslides. An overall accuracy of 95% was found, with an area under the receiver operating characteristic (ROC) curve of 0.960. Therefore, the resulting map can be regarded as useful for monitoring landslide-prone areas. Based on the findings, it is concluded that the proposed method is effective for susceptibility assessment since it yielded meaningful results and does not require extensive input data.

Keywords: