Understanding Flood Probabilities at the Regional Scale

Research project P 14478 Understanding Flood Probabilities at the Regional Scale Günter BLÖSCHL 08.05.2000 Flood probabilities are often the result of a mix of flood generating mechanisms both in time (summer vs. winter, rainfalls vs. snowmelt, etc.), and also in space (different parts of the country, different climate, soils, vegetation and land use, topography etc.). At the regional scale (i.e. for regions of the order of the size of Austria containing a large number of catchments), however, further progress in understanding and estimating these flood probabilities has so far been hampered by the lack of adequate data for a complete process description. This project provides an exciting opportunity to come up with a solution based on the novel concepts of process classification and process mixing, and spatial mixing (Figure 2). In the proposed project we will classify flood producing processes in small to medium sized catchments across Austria into a number of process types by multiresponse data analyses; for each process type we will come up with a typical flood probability model, based on derived flood frequency concepts; for each catchment we will get an estimate of the relative weights that we allocate to each of the process types; we will then derive a composite flood frequency curve based on concepts of mixed distributions (process mixing); and finally we will apply space mixing to infer the spatial (regional) flood probability patterns. We are confident that this is possible as we address the data availability problem at the regional scale by combining different types of regional data (multiresponse data) (Table 1) in a novel way. This project will contribute to an improved understanding of flood probabilities at the regional scale and will lead to novel quantitative methods that account for a mix of flood generating mechanisms. These methods are likely to be superior to existing simple statistical methods for those cases where a significant amount of extrapolation beyond the conditions of a single flood peak sample is needed, such as transposition to ungauged catchments, to changed land use and/or climatic regimes and to rare events.