AUTOMATIC DETECTION AND VULNERABILITY ANALYSIS OF AREAS ENDANGERED BY HEAVY RAIN

摘要

In this paper we present a new method for fully automatic detection and derivation of areas endangered by heavy rainfall based onlyrnon digital elevation models. Tracking news show that the majority of occuring natural hazards are flood events. So already many floodrnprediction systems were developed. But most of these existing systems for deriving areas endangered by flooding events are based onlyrnon horizontal and vertical distances to existing rivers and lakes. Typically such systems take not into account dangers arising directlyrnfrom heavy rain events. In a study conducted by us together with a german insurance company a new approach for detection of areasrnendangered by heavy rain was proven to give a high correlation of the derived endangered areas and the losses claimed at the insurancerncompany. Here we describe three methods for classification of digital terrain models and analyze their usability for automatic detectionrnand vulnerability analysis for areas endangered by heavy rainfall and analyze the results using the available insurance data.