One of the major challenges for the current and future climate diagnostics is the occurrence of very rare, but extreme intense precipitation events. Those record-shattering events lead to the Valencia-Flood in October 2024 in Spain or the Ahr-floods in July 2021 in Germany with billions of Euro of financial losses and unfortunately significant numbers of fatalities (cf. e.g., Leckebusch & McCarthy (2026) and references therein).
While the principal underlying physical mechanism of precipitation formation and the principal mechanisms of necessary precursor conditions are well understood, the exact amount of extreme precipitation and when and where those events may occur are less well understood. The same holds true for the principal rate of occurrence, which is difficult to estimate as those events may occur only once or twice per century in the current climate. Even more difficult is an assessment of those events for future climate conditions, altered by human activities. We can already diagnose that those events are more likely to occur under the influence of a warming climate, but the exact changes in those statistics of their occurrence remains a challenge. One of the reasons for this is the inability of current climate models to fully simulate those rare and extreme precipitation events in their full characteristics, from very local scale processes to the wider statistical occurrences for e.g., 30-years climatic time spans.
This project will develop and test a novel approach to estimate those rare events. It will investigate whether a combination of AI- and Extreme Value Theory (EVT)-approaches can produce reliable estimates for future climate states. Thus, can AI–EVT models trained on reanalyses and large hindcast ensembles estimate regional extreme-rainfall distributions that remain reliable when transferred between global warming levels and circulation regimes?
This project would best suit a student with a Computer Science background, or otherwise a numerical background for example Meteorology, Physics, or Maths, with experience coding in Python or similar (e.g. R, Matlab), and preferably with some experience in AI approaches.
This project does not offer a CASE studentship
Each host has a slightly different application process.
Find out how to apply for this studentship.
All applications must include the CENTA application form.
Choose your application route
In the project the combined set-up between specific AI-modelling approaches and the implementation in the context of EVT to estimate tail events in the precipitation event distribution will be explored. The likely methodological work will include:
DRs will be awarded CENTA Training Credits (CTCs) for participation in CENTA-provided and ‘free choice’ external training. One CTC can be earned per 3 hours training, and DRs must accrue 100 CTCs across the three and a half years of their PhD.
The student will build specialist knowledge in AI methodologies and drivers of European precipitation extremes, and experience in analysing climate datasets. The student will enhance their computing skills and experience, in particular AI techniques, HPC usage and coding and visualising data with Python, with the opportunity to build further transferrable skills, e.g. version control (Git). They will develop a deep practical understanding of the use of GenAI, the role of explainable and responsible AI, and how to design and create systems to communicate complex problems to end users. Opportunities for specialist training via external courses, or internal courses (e.g. Deep Learning) offered through our HPC centre, BEAR.
This PhD would provide a strong foundation for an academic career in climate AI, or a career in the private sector e.g. the growing climate intelligence industry.
This project will be embedded in our University of Birmingham’s strategic partnership with the UK Met Office (MOAP; https://www.birmingham.ac.uk/research/centres-institutes/met-office-academic-partnership), the Institute for Data and AI (IDAI) (https://www.birmingham.ac.uk/research/centres-institutes/data-and-ai), and the Birmingham Institute for Sustainability and Climate Action (BISCA; https://www.birmingham.ac.uk/research/centres-institutes/birmingham-institute-for-sustainability-and-climate-action-bisca).
Year 1: Selection of predictors and predictands, and initial evaluation of both AI and conventional approaches to test statistical relationships between these, including testing on out-of-distribution samples (i.e. data from warmer periods or different modes of interdecadal variability within the observational record).
Year 2: Design of AI-EVT approaches to estimate future climate states. Evaluate approaches in comparison with those investigated in Year 1.
Year 3: Design of AI-EVT approaches to improve robustness of physics, and applicability of frameworks for use in CMIP6 or 7 models. Evaluate approaches in comparison with those investigated in Years 1 and 2.
Angus, M., and G.C. Leckebusch, 2020: On the dependency of Atlantic Hurricane and European Windstorm Hazard. Geophysical Research Letters, 47(23), https://doi.org/10.1029/2020GL090446
Angus, M., M. Widmann, A. Orr, R. Ashrit, G.C. Leckebusch, and A. Mitra, 2024: A comparison of two statistical postprocessing methods for heavy precipitation forecasts over India during the summer monsoon. QJR Meteorol. Soc. 2024; 1–19, http://doi.org/10.1002/qj.4677
Beale, R., R. Geen, G.C. Leckebusch, M. Widmann et alii, 2025: AI4Climate Policy brief: AI will not replace climate science — but it can make it fit for an age of extremes. https://www.birmingham.ac.uk/news/2025/ai-will-not-replace-climate-science-but-it-can-make-it-fit-for-an-age-of-extremes.
Lam, R., et al., 2023: Learning Skillful Medium-Range Global Weather Forecasting. Science. https://doi.org/10.1126/science.adi2336.
Leckebusch, G.C. & M. McCarthy, 2026: The State of the UK Climate. In: The Fourth Climate Change Risk Assessment – Independent Assessment (CCRA4-IA) Technical Report [Lowe, J.A., Harrison, M. and Perks, R.J.]; https://www.ukclimaterisk.org/publications/type/technical-reports/.
Ng, K.S., Leckebusch, G.C. & Hodges, K.I., 2022: A Causality-guided Statistical Approach for Modeling Extreme Mei-yu Rainfall Based on Known Large-scale Modes—A Pilot Study. Adv. Atmos. Sci. https://doi.org/10.1007/s00376-022-1348-3.
Ng, K.S., G.C. Leckebusch, and K. Hodges, 2024: Improvement of decadal predictions of monthly extreme Mei-yu rainfall via a causality guided approach. Environ. Res.: Climate (ERCL) 3 041001; https://doi.org/10.1088/2752-5295/ad6631.
Walz, M.A., and G.C. Leckebusch, 2019: Loss potentials based on an ensemble forecast: How likely are winter windstorm losses similar to 1990? Atmospheric Science Letters, 20 (4), UNSP e891. https://doi.org/10.1002/asl.891
For any enquiries related to this project please contact:
Lead supervisor:
Prof. Dr. Gregor C. Leckebusch, Chair of Meteorology and Climatology and UK Met Office Joint Chair
School of Geography, Earth and Environmental Sciences, University of Birmingham
W: http://www.birmingham.ac.uk/staff/profiles/gees/leckebusch-gregor.aspx
To apply to this project:
Applications must be submitted by 23:59 GMT on Wednesday 6th January 2027.