Title: Prediction of landslide susceptibility using rare events logistic regression: A case-study in the Flemish Ardennes (Belgium)
Authors: Van Den Eeckhaut, Miet ×
Vanwalleghem, Tom
Poesen, Jean
Govers, Gerard
Verstraeten, Gert
Vandekerckhove, Liesbeth #
Issue Date: Jun-2006
Publisher: Elsevier science bv
Series Title: Geomorphology vol:76 issue:3-4 pages:392-410
Abstract: In this article a statistical multivariate method, i.e., rare events logistic regression, is evaluated for the creation of a landslide susceptibility map in a 200 km 2 study area of the Flemish Ardennes (Belgium). The methodology is based on the hypothesis that future landslides will have the same causal factors as the landslides initiated in the past. The information on the past landslides comes from a landslide inventory map obtained by detailed field surveys and by the analysis of LIDAR (Light Detection and Ranging)-derived hillshade maps. Information on the causal factors (e.g., slope gradient, aspect, lithology, and soil drainage) was extracted from digital elevation models derived from LIDAR and from topographical, lithological and soil maps. In landslide-affected areas, however, we did not use the present-day hillslope gradient. In order to reflect the hillslope condition prior to landsliding, the pre-landslide hillslope was reconstructed and its gradient was used in the analysis. Because of their limited spatial occurrence, the landslides in the study area can be regarded as "rare events". Rare events logistic regression differs from ordinary logistic regression because it takes into account the low proportion of Is (landslides) to Os (no landslides) in the study area by incorporating three correction measures: the endogenous stratified sampling of the dataset, the prior correction of the intercept and the correction of the probabilities to include the estimation uncertainty. For the study area, significant model results were obtained, with pre-landslide hillslope gradient and three different clayey lithologies being important predictor variables. Receiver Operating Characteristic (ROC) curves and the Kappa index were used to validate the model. Both show a good agreement between the observed and predicted values of the validation dataset. Based on a qualified judgement, the created landslide susceptibility map was classified into four classes, i.e., very high, high, moderate and low susceptibility. If interpreted correctly, this classified susceptibility map is an important tool for the delineation of zones where prevention measures are needed and human interference should be limited in order to avoid property damage due to landslides. (c) 2005 Elsevier B.V. All rights reserved.
ISSN: 0169-555X
Publication status: published
KU Leuven publication type: IT
Appears in Collections:Division of Geography & Tourism
× corresponding author
# (joint) last author

Files in This Item:
File Status SizeFormat
VandeneeckhautEtAl2006Geom.pdf Published 1036KbAdobe PDFView/Open Request a copy

These files are only available to some KU Leuven Association staff members


All items in Lirias are protected by copyright, with all rights reserved.

© Web of science