Stepping into the breach: understanding the impacts of defence breach in a UK context

Flood defences help mitigate the impacts of flooding on people, properties, and infrastructure. In JBA’s catastrophe flood model for the UK, defences are included to help estimate flood losses for re/insurers. It is important that these defences operate properly, although there is some uncertainty whether defences will continue to offer protection to properties if their condition deteriorates or a flood event reaches a loading greater than a defence was designed to mitigate against. The breach of a defence drastically increases losses to properties local to the point of breach, and representing this in a catastrophe model can help re/insurers better understand the impact defences can have on losses.  

The Environment Agency publish fragility curve data for different types of defence structures, condition grades, and breach mechanisms which estimate the probability of a flood defence failing. Using these data in the catastrophe flood model as part of a probabilistic defence breach algorithm shows that the uncalibrated fragility curves vastly over-estimate the likelihood of defences breaching in the UK, where the condition of defences is generally good and investment is strong. Calibration of the fragility curves is required to accurately reflect the observed rate of breaches given the maintenance regime in the UK. Catastrophe models which have not been calibrated to match observed breach rates could over-estimate loss, this finding is supported by previous research funded by the JBA Trust which concluded the breach rate was being overstated in the NaFRA. We investigate the effect of defence breach in our catastrophe model. 

JBA has developed an ultra-flexible approach to catastrophe modelling, which enables “what if” type questions to be easily answered. Models are configured, built and executed, all at run-time. The modularity inherent in our catastrophe model means that the development and integration of a probabilistic defence breach algorithm is quick and straightforward. The parameterisation of the model makes it simple to select from interchangeable methods, swap input data sources and to set the value of variables. This paradigm is ideally suited for calibrating defence breach parameters and for running different defence breach scenarios effectively. 

By comparing losses modelled with uncalibrated and calibrated defence breach probabilities, we quantify the potential over-estimation of loss when using the input fragility curve data in its published form. In the UK we have relative wealth of information to work with, however defence fragility and breach rates around the world are not well understood and emphasis should be put on the benefits of collecting data to help accurately quantify the effect of proper maintenance and to achieve accurate loss estimates.