Application of LAND-SUITE for Landslide Susceptibility Modelling Using Different Mapping Units: A Case Study in Croatia

Abstract

LAND-SUITE software was applied to local-scale landslide susceptibility zonation in a study area (20 km2) located in NW Croatia, which is characterised by small and shallow landslides in engineering soils and hard soils-soft rocks. Landslide susceptibility models were prepared using a LiDAR-based landslide inventory, and causal factors derived and/or modified using high-resolution remote sensing data. The landslide susceptibility analysis was carried out using different statistical methods and mapping units. As a result, we obtained ten susceptibility maps, classified into five zones based on modelled landslide probability. The main objective of this paper is to analyse the use of LAND-SUITE for deriving susceptibility zonations, useful to support applications in a spatial planning system on a local scale. One of the most valuable advantage of LAND-SUITE is that the selection of the appropriate causal factor maps can be performed using simple decisions based on multiple analyses, including multicollinearity tables and correlograms. The evaluation of the susceptibility models allowed to conclude that the pixel-based map represents the best option for application in a local-level spatial planning system, although it requires “post-processing” of the susceptibility zones to produce clustered and homogeneous information. LAND-SUITE has proven to be a valuable tool for testing numerous landslide susceptibility hypotheses.