Cyril Morcrette

[Funding covers fees plus a stipend per annum for 2026/7; stipend amount stated as £21,805.] University of Exeter offers a PhD studentship in Physics-Aware Graph Neural Networks for Robust Short-Term Visibility Forecasting at the Penryn Campus in the United Kingdom. This project addresses the challenge of forecasting low-visibility events for safe and efficient transport, especially aviation. The research will develop physics-aware spatio-temporal graph neural networks (STGNNs) for 0–6 hour visibility prediction, with an initial focus on low-visibility events below 1000 m in southeast England. The project combines machine learning, atmospheric science, and operational meteorology to improve the reliability and interpretability of short-term forecasts. The PhD is structured around six linked work packages: characterising low-visibility events and benchmark methods; developing graph-based forecasting models; linking internal ML representations with atmospheric physics; building physics-aware multi-source learning approaches; improving robustness under missing or degraded observations; and evaluating the approach with Met Office scientists for operational translation. Supervision is led by Dr Zhou Zhou with contributions from Prof. Georgios Efstathiou and Prof. Cyril Morcrette , alongside operational input from Steven Ramsdale at the Met Office. The student will work across Exeter’s data-centric engineering and atmospheric-science communities and may, subject to availability, undertake a Met Office placement or secondment. Funding is available and includes fees plus a stipend for 2026/7. The stipend amount is stated as £21,805 . The application deadline is 8 December 2026 . Applicants should use the University of Exeter application link for full details and submission instructions.