FORMALIZING EXPERT OPINION IN LANDSLIDE HAZARD ZONATION

C.J. van Westen (International Institute for Aerospace Surveys and Earth Sciences (ITC), Enschede, The Netherlands)

F. Mantovani (University of Ferrara, Italy) &

A.C. Seijmonsbergen (University of Amsterdam, The Netherlands)

The aim of this project is to evaluate the decision rules used by expert geomorphologists in the direct mapping of landslide hazard zones. Originally, before the widespread use of geographic information systems, landslide hazard mapping was done by means of direct mapping. The decision criteria used by the geomorphologists were not made clear to the user, and were often not formalized. During the last decade landslide hazard mapping is mostly done by data integration techniques in a GIS (using either bivariate or multivariate statistical techniques). Since most of these techniques are data-driven, the expert knowledge of the geomorphologist is not used sufficiently, often leading to incorrect results. This project aims in utilizing the expert knowledge on causal factors for landslides, within a GIS environment. This means that the geomorphologist who is preparing the hazard map should make clear why he considers a certain unit to be hazardous. One of the aims of the project is to develop a hazard database from which a user (for example a municipal planning authority) can consult these criteria why a certain unit has been classified as high, moderate or low hazard. He can consult this information together with other relevant information ( landcover, infrastructure and cadastral information). A method is presented using ILWIS 2.0 in which this consultation procedure becomes very simple.

A very detailed database was constructed within the new ILWIS 2.0 for Windows at a scale of 1:5.000 containing the following types of information: Topographical data, Infrastructural data, Landcover map, Geological map, Geomorphological maps, and three direct hazard maps made by three different groups. The direct hazard map was prepared and digitized, and was linked with the geomorphological information, so that for each of the approximately 1800 hazard polygons which were identified, information can be obtained on the hazard degree (high, medium and low), the hazard type (including rockfall, landslides, flowslides, flows, erosion, and flooding) and the decision criteria for defining the hazard.