Space-time extreme value models for heatwaves under different climatic conditions

Background

The 2025 summer was the hottest on record for the UK, including several prolonged and extensive hot periods with cumulatively notable impacts. Climate projections show hot conditions getting worse, but the development of suitable adaptation strategies is hampered by uncertainty in the severity, duration and extent of future heatwaves. Physical climate models are too expensive to generate sufficient data for this complex hazard, but statistical models capturing spatio-temporal extremes have their own challenges. This is compounded by different circulation regimes having distinct characteristics for the spatio-temporal behaviour of heatwaves. This project will develop space-time statistical models for heatwaves that capture climate change and account for different circulation regimes. This can augment future physical climate model simulations, through large event set simulation, reducing uncertainty in the future heatwaves hazard and informing suitable adaptation strategies.

PhD Opportunity

This project will develop statistical models for UK heatwaves. We will use extreme value theory as a basis to extrapolate into the upper tail of temperatures and will incorporate the conditional approach to extreme value modelling (Heffernan & Tawn, 2004) to capture features of heatwaves in space and time; see Simpson & Wadsworth (2021). The project will also aim to extend the conditional extremes methodology to capture the relationship between heatwaves and different circulation regimes.

The first challenge will be to identify and define appropriate covariates that can represent these circulation regimes and be used to characterise heterogeneity in the marginal distribution of extreme temperatures. This will combine methods related to generalised additive models, Gaussian Markov random fields, and machine learning. The marginal model will allow probabilistic risk estimates across the UK given different circulation regimes, which can be projected into the future to capture a changing climate. The project will also focus on event simulation, to bring better understanding of the spatio-temporal behaviour of heatwaves and inform adaptation strategies. Conditional simulations, given physical climate model simulations of different circulation regimes, will bring a stratified sampling approach that allows simulations to give a much fuller picture of extreme heatwaves. In particular, the project will explore heatwaves coinciding with unusual circulation regimes and regimes expected within a changing climate. This can significantly expand upon the limited existing knowledge, such as from current methods based on regional climates models. The project will also involve the development of user-friendly software to make the research accessible to end-users.

The project will involve supervisors at UCL, the University of Exeter and the Met Office. Collaboration with the Met Office can include a placement at the Hadley Centre.

Applicant Profile

Students with a strong background in mathematics and/or statistics, who are interested in developing statistical methodology with applications in climate science.
Experience with R coding will also be useful.

Other information

Heffernan, J. E. and Tawn, J. A. (2004). A conditional approach for multivariate extreme values (with discussion). JRSSB, 66(3):497-546. https://doi.org/10.1111/j.1467-9868.2004.02050.x

Simpson, E. S. and Wadsworth, J. L. (2021). Conditional modelling of spatio-temporal extremes for Red Sea surface temperatures. Spatial Statistics, 41:100482. https://doi.org/10.1016/j.spasta.2020.100482

Youngman, B. D. (2019). Generalized additive models for exceedances of high thresholds with an application to return level estimation for US wind gusts. Journal of the American Statistical Association 114(528):1865–1879. https://doi.org/10.1080/01621459.2018.1529596