Shortwave Flux Trends in the CMIP6 UKESM1 Climate Model Are Too Strong With Implications for Climate Sensitivity.

Abstract:

Changes in how clouds respond to temperature and atmospheric changes have been identified as the main cause of an increase in climate sensitivity between CMIP5 and CMIP6. Uncertainty in the radiative forcing (RF) from aerosols also leads to large uncertainties in model predictions of future climate change with aerosol RF being the largest of the climate RF uncertainties over the industrial period. Recently the DEEP-C long timeseries (1985-2019) dataset of shortwave (SW) top of the atmosphere (TOA) fluxes and surface fluxes based on satellite data and atmospheric reanalysis data has become available. This long observational record provides opportunities for evaluating the performance of models in terms of long-term trends in SW fluxes. Significant regional trends in SSTs and aerosols have occurred over the period of the Deep-C record and hence the evaluation of the model SW TOA trends and timeseries provides insight into the realism of the model SW TOA response to these changes.

Here we examine changes in SW TOA fluxes over recent years for the UKESM1 historical coupled ocean ensemble and find that observed and modelled SWTOA trends are negative due to a combination of increasing SSTs and decreasing aerosol emissions. However, all of the members produce negative SWTOA trends that are too strong compared to observations in almost all regions of the globe. This suggests that either the ensemble is not representing a realistic spread of real-world variability, or that there are errors in the model physics. Either would be cause for concern in terms of using that model to make predictions about the future. The AMIP model (where the observed SSTs and sea-ice concentrations are prescribed) does a better job, matching the observed distribution of trends well over the oceans. The better agreement here compared to the coupled model might give confidence in the physics of the model in terms of cloud feedbacks and adjustments to aerosols and suggests issues with the ocean component of the coupled model. Many regions still show discrepancies, though, particularly in the Atlantic Ocean region where the trends are too negative in the northern hemisphere, but too positive in the southern hemisphere. The lack of agreement may be due to atmospheric variability and is currently being explored using nudged models with realistic meteorology.