A Probabilistic Approach for Dam-break Flood Hazard Assessment

The storage of water provided by dams offers numerous benefits associated with the different uses of the resource (irrigation, energy production, water supply, etc.). However, in the event of a dam-break, a large amount of water may be released with catastrophic consequences for the population and the economic and environmental assets located downstream of the dam. For this reason, flood risk assessment is fundamental to planning emergency responses and developing mitigation measures. A key step of the flood risk evaluation is flood hazard assessment, which consists in the identification of the area prone to flooding and in mapping hydraulic parameters (e.g. maximum water depth, maximum flow velocity, flood arrival time, etc., or a suitable combination of these) representative of flood intensity. Usually, dam-break flood hazard is assessed deterministically considering only one dam-break scenario. In particular, for concrete dams, the collapse is considered total and instantaneous with the water level behind the dam at the spillway crest level. However, this approach does not take into account uncertainties in dam-break parameters and their effect on flood hazard estimation in the downstream area. To this end, we propose a probabilistic approach based on a set of dam-break scenarios characterized by different reservoir levels and breach widths. These parameters are actually the most important in defining the breach outflow hydrograph and the resulting inundation extent. Each scenario is characterized by a probability conditional to the dam-break event. For each scenario, the dam-break wave propagation is simulated via a finite volume hydrodynamic model based on the 2D shallow water equations. The flood inundation maps calculated for each scenario, along with the associated probabilities, are then combined to produce probabilistic inundation maps: a probabilistic flood extent map; probabilistic hazard level maps (with reference to pre-selected flood hazard classes); a probability-averaged flood hazard index map and a probability-weighted dam-break wave arrival time map, both coupled with a map representing the spatial distribution of the associated uncertainty. The probabilistic method is applied to the case study of the Mignano concrete gravity dam in northern Italy.