Reconstructing a hydraulic model for historic flood levels in the city of Bath, United Kingdom

Assessing the risk of future flood events and the implications for flood risk in cities is an economically and socially costly problem. In this research, we assess the utility of documentary evidence of past flood events for contemporary flood risk assessments to reduce the uncertainty in flood frequency estimation due to the interpolation from short annual maximum series (AMS) records.

The historical city of Bath, United Kingdom, developed in close relation to the River Avon, and evidence of flooding in the city of Bath can be traced back to Roman occupation. For this research a particularly rich record of historical evidence was chosen occurring from the 19^th century onwards with flood marks on buildings through-out the city as well as documentary evidence in contemporary newspapers and technical reports. The earliest flood mark found in the city of Bath dates to 1823 with 15 more extreme floods after that marked as well. The extensive flooding in 1947 initiated work on what eventually became the present-day Bath flood protection scheme (BFS) which was implemented after the 1960 catalyst flood event.

Using an existing one-dimensional hydraulic model representing the current hydraulic system of the River Avon in Bath, a historical survey of how the river and its management has changed over time was conducted. The model was developed using historical evidence (e.g. maps, flood marks, photographs, newspaper articles etc), surveyed river cross sections, recorded and design hydrographs from National datasets.

The 1960 flood is reconstructed numerically using all available data, from flood marks to old surveyed river cross sections.  The resulting hydraulic model is used to investigate the effect of the Bath Flood Defence Scheme. Sensitivity studies with different values for the roughness coefficient are also presented in order to assess the uncertainty on water levels during extreme events. Finally, the numerically reconstructed historical peak flood discharge is compared with the results obtained using a simple Manning equation approach to assess the two methods. This paper demonstrates how hydraulic modelling can be applied to historical data and offers considerable potential to further investigations in the improvement of design flood flows.