9612531 Bras The prevalent approaches to estimating extreme precipitation and flood relationships are empirical curve fitting to historical data and numerical simulation also based on empirical models. Both methods become inaccurate if very rare events or the impact of climatic variations and human interventions on such extremes are estimated. The source of the inaccuracy is that, in the absence of a theoretical or physical basis, extrapolation far beyond the range of the historical data becomes highly uncertain. A large number of these empirical relations are of the power-type that reflect the scale-invariant properties of the observed hydrologic event. This project will use a scaling or fractal approach to the problem of extrapolation from limited data of hydrologic extremes. A new class of rainfall and width functions models will be developed which posses the simple- and multi-scaling properties observed in nature for these quantities and that will be validated through extensive data analysis. These models will combine two basic approaches, the multi-scaling and the pulse superposition. The final phase of the project will include the parameterization of rainfall and basin response in terms of climate and catchment characteristics, using them to assess the sensitivity of peak precipitation and flood events to both natural changes in the climate, basin characteristics, and human intervention. Improving the understanding of the rainfall-flood relationship is of great benefit to areas and people at risk and will advance the state-of-the-art. ***
