A Monte Carlo-based approach for reconstructing past flood events

This paper presents a new Monte Carlo-based inverse modelling approach allowing reconstruction of past flood events including the associated uncertainty. This is particularly useful for cases where no direct observations of peak flow are available, either because the event predates systematic gauging or because existing gauging equipment was overwhelmed by the event. Using a HEC-RAS 2D model, the Monte Carlo approach identifies subsets of input variables, specifically the hydrograph peak flow and Manning’s n values, that result in simulated water depths comparable to recorded flood level markers. In a case study of the River Chew in southwest England, reconstruction analysis of a large event gives a central 95% range of matched simulations for the reconstructed peak flow between 156 and 175 m3/s.