Postdoctoral Research Assistant in Uncertainty Quantification for Earth-system Prediction

Postdoctoral Research Assistant in Uncertainty Quantification for Earth-system Prediction

University of Oxford - Department of Physics

Oxford, United Kingdom

GBP 39,424 - 47,779 per year

Weather and climate prediction are undergoing a profound transformation. Alongside traditional physics-based forecast systems, machine-learned (ML) weather prediction models, hybrid ML-physics models, and climate emulators are rapidly becoming an integral part of the prediction landscape. While these approaches can achieve remarkable predictive skill, fundamental questions remain regarding the quantification and representation of uncertainty within and between these models. Addressing these questions is essential if their predictions are to be used for scientific understanding and societal decision-making.

The successful candidate will join Principal Investigator Christensen in addressing the long-standing challenge of delivering reliable probabilistic weather and climate predictions. The project will develop innovative approaches to understand, quantify, and ultimately reduce model uncertainty in next-generation Earth-system prediction systems.

The research adopts a seamless perspective in two key ways. First, it will use insights gained from short-range weather forecasting to improve our understanding of longer-term climate behaviour, and vice versa. Second, it will exploit advances in machine learning (ML) weather and climate models to improve physics-based models, while using physical understanding to enhance ML approaches. By integrating ideas from numerical weather prediction, climate modelling, and modern machine learning, the project aims to develop physically grounded methods for uncertainty quantification that are applicable across the full spectrum of Earth-system prediction. The post-holder will also have the opportunity to contribute to teaching.

Applicants should hold, or be close to completing, a PhD/DPhil in physics, climate science, computer science, or a closely related discipline. They should have strong computational skills, including experience with UNIX/Linux and programming in Fortran, Python, or other high-level languages. Candidates should also demonstrate the ability to conduct original research of international standing.