Background

Anthropogenic aerosols cause substantial cooling of climate that offsets some of the warming by greenhouse gases, but the uncertainty is very large. This is important because future global warming rates range between very rapid and quite modest depending on the aerosol forcing that is assumed. Despite extensive observations to test climate models, the uncertainty has not been reduced significantly. However, the latest research at Leeds using advanced uncertainty quantification (UQ) approaches points to several exciting ways forward. We can now quantify where, when and how we should make new aerosol observations to optimally reduce model uncertainty and we can use model-observation comparisons to detect model “structural deficiencies”, which are currently preventing further uncertainty reduction. With this knowledge, we can design a modelling and observation strategy with greater potential to reduce uncertainty in projections.

PhD Project

The project aims to use the latest techniques of model uncertainty quantification (UQ) to define optimal observation strategies for reducing climate model uncertainty.

You will work in collaboration with Silverlining to test the value of their new marine aerosol observations for model constraint (see https://www.silverlining.ngo/global-atmospheric-observations). Your results will also provide guidance to Silverlining on the optimum observation strategy.

The overall approach will be to use perturbed parameter ensembles of the UK climate model together with emulators that can generate millions of ‘surrogate models’ for rigorous model-observation comparison.

The research will address three major factors limiting uncertainty reduction:
1) Model structural deficiencies. We have recently shown that incomplete or deficient physical processes in models place a fundamental limit on how much model uncertainty can be constrained by observations. The research will involve extensive model-observation comparisons to detect these deficiencies for the first time using extensive marine observations.
2) Model “equifinality”. We know that multiple models can agree equally well with observations but still diverge in future projections, which is a hidden but major cause of projection uncertainty. We can now quantify and map this effect, so the aim of the research will be to determine how much an intensive marine observation programme could help to reduce the effect.
3) Representation error in observations, which is the uncertainty caused by comparing point observations of spatially and temporally variable aerosols with models. The research will assess whether repeated ship observations over extended periods could help to narrow this large source of uncertainty.

The project offers an exciting opportunity to work with state of the art modelling tools combined with new observations, with the potential to help shape future observation strategies with the collaborator Silverlining.

Applicant Profile

This project would be suitable for most STEM graduates. Students with a strong background in physics, engineering, mathematics, statistics or computing who want to apply these skills to improving our understanding of climate change.

Other Information

“Lead supervisor, Prof Ken Carslaw’s page: https://environment.leeds.ac.uk/see/staff/1196/professor-ken-carslaw-frs.

Co-supervisor, Dr Leighton Regayre’s page: https://environment.leeds.ac.uk/see/staff/1496/dr-leighton-regayre.

The Silverlining Scaled Observations for Atmospheric Resilience (SOAR)™ web page describes the ambitious observation plans: https://www.silverlining.ngo/global-atmospheric-observations.

A good summary of the latest science with these approaches is in this article: https://acp.copernicus.org/articles/26/4651/2026/acp-26-4651-2026.html

The “”equifinality problem”” is described here: https://www.pnas.org/doi/full/10.1073/pnas.1507050113

A recent PhD paper on detecting structural deficiencies in models: https://acp.copernicus.org/articles/26/2487/2026/