A First Comparison of Multiple Probability Hazard Outputs from Three Global Flood Models

Wednesday, 17 December 2014: 5:45 PM
Mark Trigg1, Paul D Bates1, Tim J Fewtrell2, Dai Yamazaki3, Florian Pappenberger4 and Hessel Winsemius5, (1)University of Bristol, School of Geography, Bristol, United Kingdom, (2)Willis Group, Willis Global Analytics, London, United Kingdom, (3)JAMSTEC Japan Agency for Marine-Earth Science and Technology, Kanagawa, Japan, (4)European Center for Medium-Range Weather Forecasts, Reading, United Kingdom, (5)Deltares, Delft, Netherlands
With research advances in algorithms, remote sensing data sets and computing power, global flood models are now a practical reality. There are a number of different research models currently available or in development, and as these models mature and output becomes available for use, there is great interest in how these different models compare and how useful they may be at different scales. At the kick-off meeting of the Global Flood Partnership (GFP) in March 2014, the need to compare these new global flood models was identified as a research priority, both for developers of the models and users of the output. The Global Flood Partnership (GFP) is an informal network of scientists and practitioners from public, private and international organisations providing or using global flood monitoring, modelling and forecasting. (http://portal.gdacs.org/Global-Flood-Partnership). On behalf of the GFP, The Willis Research Network is undertaking this comparison research and the work presented here is the result of the first phase of this comparison for three models; CaMa-Flood, GLOFRIS & ECMWF. The comparison analysis is undertaken for the entire African continent, identified by GFP members as the best location to facilitate data sharing by model teams and where there was the most interest from potential users of the model outputs. Initial analysis results include flooded area for a range of hazard return periods (25, 50, 100, 250, 500, 1000 years) and this is also compared against catchment sizes and climatic zone. Results will be discussed in the context of the different model structures and input data used, while also addressing scale issues and practicalities of use. Finally, plans for the validation of the models against microwave and optical remote sensing data will be outlined.