YES-27-A-10: Understanding polar aerosols and improving their representation in climate models
Recent data reveals that the latest Earth System climate models — which will be used for the upcoming IPCC climate assessment — severely underestimate the number concentrations of cloud drops in remote, high-latitude regions. This bias points to critical missing sources and processes in our models. We cannot accurately predict future climate change without understanding the pre-industrial background state — and these pristine, remote regions provide our best window into that past. In this PhD you will explore the reasons for the bias in the UK Earth System Model (UKESM) climate model and test routes for fixing it. The PhD will equip you with high-demand skills in advanced climate modelling, satellite remote sensing, and machine learning techniques.
Cloud droplet number concentrations (CDNC) play a massive role in our climate by determining how much of the Sun’s energy is reflected back into space by clouds. Aerosol particles act as the “seeds” for these cloud droplets, and in remote polar regions, these aerosols are heavily influenced by biological activity. For instance, marine phytoplankton release sulphur-containing compounds that can trigger the formation of new aerosol particles. There is also evidence that melting sea-ice leads to flurries of new particle formation activity. However, the exact mechanisms behind this high-latitude particle formation remain highly uncertain and poorly represented in climate models. In this project, you will expand upon existing experimental particle formation schemes within the UKESM to test whether modifications can bring the model into alignment with real-world observations.
Alongside traditional model evaluation, you will utilize the UKESM as a “virtual laboratory” to pioneer new data-driven techniques. In the model world, we can run controlled experiments—such as switching specific aerosol processes on or off—to see exactly how the climate responds. You will apply machine learning to these controlled model fields, simulating what a satellite would “see” under different conditions. By training machine learning algorithms to correctly identify and isolate these aerosol signals in our controlled virtual world, you will build robust frameworks. These can then be confidently applied to real satellite and field observations, unlocking a powerful new way to detect and constrain missing aerosol processes in the real atmosphere.
Met Office partners will provide additional supervision, collaboration opportunities, and access to the latest cutting-edge UKESM Earth System Model and supercomputing facilities. The project includes a fully supported opportunity to work directly at the Met Office for at least three months, allowing you to collaborate with world-class climate scientists and experience research life outside of academia.
This PhD offers an exciting opportunity to help develop the next generation of climate models and directly reduce uncertainty in future climate predictions. In doing so, it will equip you with highly desirable, transferrable skills in machine learning, global climate modelling, and the analysis of massive datasets, field measurements, and satellite observations.
Further reading:
https://www.nature.com/articles/s41612-025-00990-5
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2024MS004683?af=R
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2017RG000593