Entwicklung eines Expertensystems zur Hochwasserfrühwarnung unter Berücksichtigung der Vorhersageunsicherheit

The aim of the BMBF supported research project HORIX (project life 2006 - 2008) is the development of an effective early-warning system against floods in meso-scale catchments (An overview of all BMBF supported flood-related projects is given at: http://www.rimax-hochwasser.de/). This research project is a joint effort of several universities and consulting engineers to analyse the whole model chain of flood events, with the components: precipitation forecasting - rainfall-runoff-modelling - hydrodynamic modelling - identification of inundation areas. The individual contributions will then be integrated into a knowledge-based decision support system under the lead of the University of the Federal Armed Forces Munich (www.unibw.de/ifw/WWR/ forschung/horix). This paper introduces the modules rainfall-runoff-modelling, hydrodynamic modelling (with calibrating routines and uncertainty analysis), and rule-based expert system. The name "HORIX" stands for "Development of an Operational Expert System for Flood Risk Management Considering Prediction Uncertainty". Accordingly, one focus of the work is on the determination of the prediction uncertainties in the model chain of flood events. The expert system that should be easy to handle and robust is based on fuzzy-logic rules. These "if-then" rules are based on measured data and on a multitude of previously computed model simulations. Warnings to the population are issued when certain warning thresholds are exceeded, for instance in Bavaria those defined by the Flood-information Service (HND). For better visualization of the impending flood threat, the (dynamic) innovation areas with their uncertainty ranges are shown in the Internet. This is done by means of the Open-Source Software UMN Map Server (http://mapserver.gis.umn.edu/ gallery).