Ensemble Prediction of Flood Maps Under Uncertain Conditions (PAP014328)
Juan Pablo Rodríguez-Rincón, Adrián Pedrozo-Acuña, Ramón Domínguez-Mora
Floods in a changing climate
Flooding is the most common and damaging natural hazard faced by civilization, and flooding threats are likely to increase given current climate change predictions that suggest more intense hurricanes and precipitation. Flood events observed in developed and developing nations alike, highlight the necessity to generate a better understanding on what causes extreme flooding events and how we can better manage flood risk. Moreover, the process of flood risk evaluation and management contains a great deal of uncertainty, which in turn is ascribed to the limitations in the current body of knowledge. Thus, it is necessary to somehow consider and inform how current limitations in the knowledge affect a given prediction. Therefore, it is anticipated that quantification of these uncertainties and their propagation through to modelling process is of great importance. Having this in mind, the work in this paper aims to develop a cascade modelling approach for the generation of more reliable and useful flood maps. This involves the use of a Numerical Weather Model (NWP), a rainfall-runoff model and a standard 2D hydrodynamic model. Uncertainty is considered in both the meteorological and hydrological models through the estimation of ensemble precipitation scenarios and spaghetti plots, respectively. The characterisation of the runoff by multiple possibilities, opens the door to a probabilistic estimation of affected areas, which in turn allows the evaluation of uncertainty propagation to an estimated flood map. The methodology is enriched with the use of field data of high quality, comprised by a data from a field campaign, automatic gauging stations and LiDAR data for an accurate representation of topographic elevation. The presented approach is useful for both, the reduction of epistemic uncertainties and the generation of flood management strategies through probability flood maps. The approach is applied to a region of Mexico with the highest precipitation rate, expecting that generated results would be useful in the design of more effective flood management strategies.