Addressing non-linearities when estimating radiative feedbacks associated with different historical forcing agents

Understanding how radiative feedbacks respond to different historical forcing agents (e.g. aerosols or greenhouse gases) improves our ability to relate historical changes (1850-2014) to future projections. This is often investigated using historical single forcing experiments, where only one forcing agent is allowed to vary, to decompose the total (all-forcing) response. However, using a 45-member ensemble, we demonstrate that there are non-linearities in this decomposition which challenge its utility in HadGEM3-GC31-LL. Specifically, strong warming in the Southern Ocean and sea ice loss are seen in the aerosol single forcing experiment despite global cooling, which is found to be a feature that does not combine linearly with other climate drivers. Instead, we calculate the aerosol response as the difference between the all-forcing experiment and an “all-but-aerosol” experiment, where all forcing agents apart from aerosols are included. This method does not show strong Southern Ocean warming and sea ice loss in response to anthropogenic aerosols. We instead see a positive surface albedo feedback in this region, which is more consistent with the feedbacks seen in the all-forcing response. This allows for a more accurate comparison between feedbacks caused by different forcing agents.