Kevin Wilson: Probability elicitation for a future flooding scenario in the CReDo digital twin

In a previous seminar in the series, Jim Smith discussed structural elicitation conducted to assess the vulnerability of different infrastructure assets (water, power, telecoms) to a future extreme flooding event, as part of the UK's Climate Resilience Demonstrator (CReDo) digital twin. One purpose of the digital twin was to demonstrate the potential impact of an extreme flooding event over a particular geographical area, in terms of the likelihood of failure of individual assets and the cascade of these failures through the network. In this talk I will take as my starting point the Bayesian Networks developed in the structural elicitation, which provide a comprehensive picture of the path to failure (or not) of individual assets during a flood. I will then describe the probability elicitation we conducted for one particular asset, a water pumping station, for one of the flooding scenarios considered in CReDo. The elicitation was conducted virtually, using a light version of the Sheffield elicitation framework. As part of the first phase of CReDo a simple "rule of thumb" was created to provide the probability of failure of individual assets based on the flood depth at the asset. The probabilities elicited provided a small sense check of this rule of thumb.