Reconstructing Historical Flood Events: A Monte Carlo-Based Uncertainty Approach

Reconstructing historical flood events can offer critical insight into past hydrological responses to extreme weather, informing contemporary flood risk management and infrastructure design. This study employs reverse engineering, based on historical data such as recorded rainfall, flood marks, visual records, and eyewitness accounts to reconstruct a flood event. Historical data was collected by the team during a workshop with the local community. The approach involves hydrological (HEC-HMS) and hydraulic (HEC-RAS) models to simulate the flood event. The July 1968 UK storm, remarkable for record rainfall reaching 175 millimetres within 18 hours, caused extensive devastation in south-west England. This study focuses on reconstructing the 1968 flash flood on the River Chew, notably the peak hydrograph in the village of Pensford. A Monte Carlo simulation approach is used in conjunction with the HEC-HMS and HEC-RAS models to produce a range of potential input hydrographs with uncertainty input parameters (primarily event rainfall and Manning’s roughness) that match the historical evidence.  In particular, the Monte Carlo approach is implemented using a series of Python scripts enabling multiple HEC-RAS simulations to be conducted and the results synthesised in the form of an uncertainty analysis of key parameters such as peak flow.