Novel approaches to observationally constrain aerosol effects in climate models

Human emissions of aerosol particles have caused a large cooling effect on Earth’s climate that has partially offset global warming rates. The magnitude of the effect is highly uncertain and feeds through to estimates of how much Earth will continue to warm in future as greenhouse gases rise (Peace et al., 2020). Efforts to improve the representations of aerosol physical processes in climate models have not significantly constrained the magnitude of the effect, even though there are extensive measurements available to test models. In part, this is because these efforts to observationally constrain climate models do not separate the effects of structural model deficiencies (the way processes are represented in models) and uncertain process-related parameters in the model equations (parametric uncertainty). Comparison of the spread in multi-model datasets (e.g. Gliss et al., 2021) and in perturbed parameter ensembles (Carslaw et al., 2013, Regayre et al., 2018, Carslaw et al., 2018) shows that these two sources of uncertainty are of comparable magnitude.



Global mean temperature change relative to the 1850 to 1900 period for a future emission scenario (SSP2-RCP4.5). This figure displays an illustrative range in exceedance year of 1.5 °C warmer than early-industrial temperatures, and highlights how uncertainty in equilibrium climate sensitivity (ECS) and transient climate response (TCR) values (caused in part by uncertainty in aerosol forcing) limits our ability to predict future climate. This figure is adapted from Fig. 4 of Peace et al., 2020

The aim of this project is to develop and apply novel methods to reduce uncertainty in aerosol effects on climate in the Met Office climate models. Recent results from Leeds show that structural deficiencies severely limit the extent to which the overall model uncertainty can be reduced through model calibration against observations (e.g. Johnson et al., 2020, Regayre, et al. in prep.). Therefore, it is vital to be able to distinguish these two sources of uncertainty.

The aims of this project will be to:
1)         Develop and apply methods to distinguish between structural model errors and parametric uncertainty when comparing models to extensive observations
2)         Develop and apply methods to identify the potential causes of structural error and define approaches to improve them in models
3)         Define a ranked list of structural improvements that affect aerosol forcing to feed into Met Office model development activities
4)         Demonstrate that separation of these two sources of uncertainty enables a stronger observation-based constraint of model aerosol radiative forcing uncertainty

In this project, you will use large sets of climate model simulations (ensembles) combined with extensive observations of aerosols, clouds and radiation from surface sites, aircraft and satellite data. You will scale up output from the ensembles (using powerful techniques to create model emulators) to create essentially millions of variants of the climate model to analyse against the observations.