Uncertainty in Estimates of Extreme Flood Magnitudes for Large-Scale Flood Hazard Mapping

Abstract:

The vast worldwide impact of flood events, and the development of new remote sensing techniques and increasing computing power, has motivated the development and application of global flood hazard models (GFHM). These have become useful tools for risk assessment and management, especially in regions where there is little hazard information available. One of the key uncertainties associated with GFHMs is the estimation of extreme flood magnitudes (i.e. a 1-in-100 year flood event). Two key approaches are used for generating flood magnitudes: 1) use of continuous timeseries of simulated streamflow from a global land surface/hydrological model, or 2) statistical methods using observed gauged discharge and regional flood frequency analysis (RFFA). In this study, a 1-in-100 year flood magnitude was estimated using four global hydrological models and two global RFFA datasets for 1142 gauges (with upstream area >1000km²) across the conterminous US. Initial results have shown that there are clear spatial patterns of the biases associated with each of the methods. The global hydrological models mostly overestimate flood magnitudes in the west, and underpredict in the east. However, there are large differences in the spatial patterns when flood magnitudes are estimated using RFFA. This study will investigate what controls this spatial variability and assess whether ensemble predictions perform better than any individual prediction.