Comparing ensemble projections of flooding against flood estimation by continuous simulation

- Impacts of climate change on flooding in the Avon catchment in the UK are explored.
- Climate model ensemble utilized via a number of application techniques.
- Climate projections are compared with uncertainty under current conditions.
- Results indicate the need to incorporate multiple approaches in impact studies.
- Results also highlight the significance of uncertainty under current conditions.

SummaryClimate impact studies focused on the projection of changing flood risk are increasingly utilized to inform future flood risk policy. These studies typically use the output from global (GCMs) and regional climate models (RCMs). However the direct application of GCM/RCM output is controversial as often significant biases exist in predicted rainfall; instead a number of alternative 'correction' approaches have emerged. In this study an ensemble of RCMs from the ENSEMBLES and UKCP09 projects are applied, via a number of application techniques, to explore the possible impacts of climate change on flooding in the Avon catchment, in the UK. The analysis is conducted under a continuous simulation methodology, using a stochastic rainfall generator to drive the HBV-light rainfall run-off model under a parameter uncertainty framework. This permitted a comparison between the projections produced by differing application approaches, whilst also considering the uncertainty associated with flood risk projections under observed conditions.The results from each of the application approaches project an increase in annual maximum flows under the future (2061-2099) climate scenario. However the magnitude and spread of the projected changes varied significantly. These findings highlight the need to incorporate multiple approaches in climate impact studies focusing on flood risk. Additionally these results outline the significant uncertainties associated with return period estimates under current climate conditions, suggesting that uncertainty over this observed record already poses a challenge to develop robust risk management plans.