1950-control vs 1850-control: How do HighResMIP simulations relate to CMIP simulations?

A common set of simulation is important for intercomparison between different models. The ‘entry-card’ to participate in CMIP is to perform the baseline DECK simulations (1850-control, 1pctCO2, abrupt-4xCO2, historical-amip). However, performing the control and historical simulations from an 1850 baseline is prohibitively expensive for high-resolution, fully-coupled, general-circulation-models (GCMs). Therefore, HighResMIP chose to use a shorter experimental protocol based on 1950 conditions alongside a shorter spin-up length and simplified aerosols. Because of this difference in protocol, it is not clear exactly how the HighResMIP simulations relate to the other CMIP simulations. In this study we analyse the control and historical simulations of the HadGEM3-GC3.1 model, which performed control and historical simulations based on both 1950 and 1850 baselines. Our results show that the absolute temperature is sensitive to the different experimental protocol, but the anomalies are much more comparable. This opens an interesting discussion on whether climate change should be discussed in terms of absolute values or anomalies. The difference in the absolute value (and mean state) is largely due to the different aerosol scheme used in CMIP and HighResMIP for this particular model. The second phase of HighResMIP no longer require models to use Easy Aerosol, so modelling centres should use the same aerosol scheme if they would like their HighResMIP simulations to be comparable to CMIP simulations.