Presenter/Author Information

O. Krol
T. Bernard

Keywords

fuzzy logic, landslides, mass movement processes, early warning systems

Start Date

1-7-2012 12:00 AM

Abstract

In many alpine and low mountain regions landslide events occur more frequently during the last years induced by a growing number of hazards. Since these events often cause high material damage with corresponding costs and/or even personal injury, there is an increasing demand of early warning systems in order to minimize the impact of the events. Currently static hazard index maps are used which indicate areas with a high disposition for landslide. These hazard index maps are derived from information about slope, land use, geotechnical structure, etc. Depending on crisp criteria (e.g. threshold parameters) several disposition classes are derived. Since these static disposition or susceptability maps do not consider up to date information and short-term conditions (e.g. precipitation, temperature), obviously the current danger of landslide events may differ considerably. Hence the key idea of ELDEWAS is to merge the static information with online data (e.g. from dynamic weather nowcasting) in order to detect the high danger landslide areas in real-time. The central challenge developing such an early warning system is to consider all available data (e.g. weather nowcasts, soil measurement data) and sources of knowledge (e.g. geological expert knowledge in terms of maps and heuristic expert knowledge; geological, hydrogeological and physical models). Obviously enhanced information fusion methods are required in order to deal with this very heterogenous sources of information in real-time. Especially the representation of the expert knowledge is crucial. In ELDEWAS the expert knowledge is represented by means of IF-THEN rules based on Fuzzy Logic. These rules in general can easily be formulated by the experts. Using Fuzzy Logic avoids crisp thresholds and binary decisions and hence provides an evaluation close to human decision makers. Static parameter maps represent the factors triggering landslide that can be merged in a case-dependent way and where the results correspond to current disposition for landslide/ mudflow. The weather nowcasting is computed by the software package INCA (Integrated Nowcasting through Comprehensive Analysis) that adopts forecasts from numerical weather models to a given topography by incorporating current measurements. This procedure allows a spatial and temporal refinement of the meteorological forecast/ nowcast. First results of the application of ELDEWAS to landslide detection for Lower Austria are presented.

Share

COinS
 
Jul 1st, 12:00 AM

ELDEWAS - Online early warning system for landslide detection by means of dynamic weather nowcasts and knowledge based assessment

In many alpine and low mountain regions landslide events occur more frequently during the last years induced by a growing number of hazards. Since these events often cause high material damage with corresponding costs and/or even personal injury, there is an increasing demand of early warning systems in order to minimize the impact of the events. Currently static hazard index maps are used which indicate areas with a high disposition for landslide. These hazard index maps are derived from information about slope, land use, geotechnical structure, etc. Depending on crisp criteria (e.g. threshold parameters) several disposition classes are derived. Since these static disposition or susceptability maps do not consider up to date information and short-term conditions (e.g. precipitation, temperature), obviously the current danger of landslide events may differ considerably. Hence the key idea of ELDEWAS is to merge the static information with online data (e.g. from dynamic weather nowcasting) in order to detect the high danger landslide areas in real-time. The central challenge developing such an early warning system is to consider all available data (e.g. weather nowcasts, soil measurement data) and sources of knowledge (e.g. geological expert knowledge in terms of maps and heuristic expert knowledge; geological, hydrogeological and physical models). Obviously enhanced information fusion methods are required in order to deal with this very heterogenous sources of information in real-time. Especially the representation of the expert knowledge is crucial. In ELDEWAS the expert knowledge is represented by means of IF-THEN rules based on Fuzzy Logic. These rules in general can easily be formulated by the experts. Using Fuzzy Logic avoids crisp thresholds and binary decisions and hence provides an evaluation close to human decision makers. Static parameter maps represent the factors triggering landslide that can be merged in a case-dependent way and where the results correspond to current disposition for landslide/ mudflow. The weather nowcasting is computed by the software package INCA (Integrated Nowcasting through Comprehensive Analysis) that adopts forecasts from numerical weather models to a given topography by incorporating current measurements. This procedure allows a spatial and temporal refinement of the meteorological forecast/ nowcast. First results of the application of ELDEWAS to landslide detection for Lower Austria are presented.