Novel statistical AI approaches for modelling and evaluating extreme windstorm risk

Extra-tropical cyclones are fascinating storms that have devastating impacts on society. For example, in winter 2013/14, more than 10 extreme storms passed

over Europe, leading to total insured losses of more than $3.3 billion. Losses due to such events are expected to rise rapidly due to climate change.

Insurers make decisions about such losses by using stochastic event simulations from catastrophe models. Such models have many simplifying assumptions and so it is useful to have independent ways to evaluate such simulations.

This project will build on existing work by advancing current approaches using modern statistical AI methods (e.g. Bayesian hierarchical spatial extreme models) for characterising extreme wind speeds from windstorm spatial footprint datasets for Europe, Japan, and other regions.

As well as providing new insight into storm behaviour, this project will help insurers develop a more reliable view of storm risk, resulting in better protection for society.

This project will address one or more of the following questions:

a) What is the distribution of extreme wind gust speeds at any given location?

b) How do these distributions depend on local climate modes and climate change?

c) How does the maximum upper wind speed vary over space and in time as the climate evolves?

This will involve extending current statistical methods for analysing windstorm footprint datasets and will widen their impact to a global scale. The approaches will enable better validation and adjustment of catastrophe model estimates of extreme wind speeds and how maximal wind speeds might change due to climate change.

The project will be supervised by Prof. Stephenson, Prof. Economou, Prof. Scaife, and Dr Priestley at the U. of Exeter who have had extensive experience of successfully collaborating with natural catastrophe insurers over many years. The project will also involve working closely with Willis Towers Watson (WTW) catastrophe modellers in London. The WTW Research Network is an award-winning collaboration supporting and influencing science to improve the understanding and quantification of risk. Linking more than 60 organisations in science, academia, think tanks, and the private sector, the WTW Research Network forms innovative partnerships with the risk management and insurance industries to confront the full spectrum of risk modelling challenges. The project will also benefit from close collaboration with the nearby Met Office via Prof. Scaife who is Head of Monthly to Decadal Prediction.