University of Leeds: Earth, Environment & Sustainability
Supervisor: Léa Prévost eelp @ leeds.ac.uk
Aerosols play a major role in climate, but they remain one of the largest sources of uncertainty in climate models. One way we study this uncertainty is through a perturbed parameter ensemble (PPE), where we run the same model many times with different values for uncertain aerosol processes such as emissions, chemistry and deposition. Doing so creates a range of plausible atmospheres and helps us identify which processes drive model uncertainty and how well the model matches real world observations.
Until now, our research has focused on monthly averages, which smooth out the day to day variability of aerosols. Daily data can look very different because aerosols are short lived and influenced by weather. In this project, you will analyse the daily dataset from a new PPE, giving our first detailed view of day to day variability. You will assess how much conditions change from one day to the next, how well the model captures daily observations, and whether the key uncertain processes differ at finer timescales. You will also test whether grouping days into categories such as wet and dry, or high and low aerosol days, reveals patterns hidden by monthly averages.
A student who enjoys data analysis and problem-solving will be a great fit. You will gain hands on experience with climate data, present your results to our research group and an external team, and have the opportunity to attend the UKCA Science Meeting to hear about current modelling priorities and ongoing challenges in the model you will be working with.
The project will follow a structured six week plan:
- Week 1: Set up on the HPC, get familiar with the ensemble, and explore daily vs monthly behaviour.
- Week 2: Bring in daily observations and compare model–observation distributions.
- Week 3: Run daily scale sensitivity analyses to see whether the dominant uncertain parameters change from day to day.
- Weeks 4–5: Test alternative ways of grouping the daily data, for example by meteorological conditions (such as wet versus dry days) or by aerosol state (such as high AOD versus low AOD days) and examine how model biases and parameter sensitivities differ between these groups.
- Week 6: Summarise what daily data reveals that monthly averages hide and clearly demonstrate how daily variability and grouped day analysis change our understanding of the PPE compared to the existing monthly mean perspective.
Interactions: The supervisor will check in with the student daily to discuss goals and progress, provide feedback on figures and interpretation, and guide next steps. The student will also have opportunities to present emerging results to the research group and receive supportive, constructive input from the wider team.
Skills gained: The student will gain an understanding of how climate models work and where uncertainty comes from, develop familiarity with aerosol processes and how parameter choices influence model output, and build experience handling large datasets on an HPC. They will also strengthen their data analysis and coding skills and learn to communicate their findings clearly to an audience of researchers.
The student will be fully embedded in an active research environment.
They will:
– Participate in weekly Aerosol, Clouds & Climate group meetings, a vibrant research group that runs skill building sessions (for example on designing posters or making the most of conferences) as well as regular science talks from researchers at a range of career stages.
– Join our smaller PPE subgroup to hear about current PPE research and the open questions/challenges about modelling uncertainty.
– Present their results to the external project “Towards maximum feasible reduction in aerosol radiative forcing”, a collaboration with the University of Sheffield. The project uses the same PPE tools to investigate how aerosol uncertainty influences radiative forcing, and the insights from daily scale analysis will provide an important contribution that is expected to attract strong interest from the team.
– Attend the UKCA Science Meeting, where they’ll hear about current modelling priorities and challenges specific to the model they’ll be working with.
