On extreme rainfall scaling and stochastic storm transposition to enhance the design hydrographs definition and flash flood risk assessment

Flash floods can be considered among the worst natural hazards that usually impact small river basins, often ungauged, characterized by few hours or less of response time to short and high-intensity rainfall events. The events are characterized by sudden, small-scale heavy rains and unpredictable peak discharges. In this context, the definition of design hyetographs and hydrographs, at the basis of flood risk assessment, is a paramount task, because of the great uncertainty with concern with ungauged river basins and the short time series. The definition of synthetic hyetographs based on past observed storm events, scaled and transposed, in order to improve the flash flood risk assessment is presented here. Particularly, the spatial distribution of past observed storm events is analyzed and the Areal Reduction Factors (ARFs), to scale the observed hyetographs, for different river basins extensions, are considered. The hyetographs of past observed storm events, scaled with the ARF according the sizes of the investigated river basins,  are then applied on different locations with a  Stochastic Storm Transposition and used as input of hydrological models. The entire methodology is applied to a set of river basins in Northern Tuscany (central Italy).  The Stochastic Storm Transposition involves resampling and transposing storm events to generate synthetic events from a collection of realistic events. The synthetic hyetographs obtained by the past events are tested on river basins with a similar rainfall statistical distribution of the place in which the events occur, i.e. over catchments that can statistically have similar events, avoiding unreal comparison with the design hydrographs. The derivation of synthetic hyetographs is made by scaling recent short duration events occurred in the area to identify if the design hydrographs commonly used in the region underestimate the flash flood risk. It is shown that peak discharges of short-duration events are usually greater than the design floods, of the small river basins with an area generally less than 30 km².