Effective sensitivity and heat capacity in the response of climate models to greenhouse gas and aerosol forcings

Simulations of greenhouse-gas-induced climate change from the pre-industrial period to 2100 by the CSIRO (Mark 2, Mark 3), Hadley Centre (HadCM2) and Canadian (CGCM1) coupled atmosphere-ocean general-circulation models are compared with those from mixed-layer ocean (MLO) versions of the models. The effect of additional transient aerosol and ozone forcings in Mark 2 is examined. Some of the variation in the global-mean surface warming of the coupled models can be attributed to the rate of heat uptake by the subsurface ocean. Based on the effective heat-capacity diagnostic, the HadCM2 ocean retards the warming more effectively than the ocean in the other models. Much of the variation between both the coupled and MLO models is due to the differing effective sensitivity (to doubled CO2) of the models. In particular, the rapid warming after 2000 from CGCM1 relates to increasing sensitivity that is largely associated with rising feedback from the snow-free or low latitudes, according to a regional feedback analysis. The Hadley Centre model has a lower global sensitivity than Mark 2, related to the high-latitude feedback. The lowest warming, in both model versions, is from Mark 3, which has relatively low feedbacks. The addition of aerosol forcing is shown to raise the net effective sensitivity in Mark 2. A simple two-region system including horizontal heat transport based on Mark 2 is used to explore the effect of variations in forcings and feedbacks on surface warming at equilibrium. Effective sensitivity can vary by -20% to +70% compared with the standard doubled CO2 case as forcing is confined to either the low- or high-latitude regions, Choosing extreme feedbacks from the other models leads to variation from 2.7 K to 7.2 K. A two-parameter transport model, which allows poleward net flux changes consistent with HadCM2 and Mark 3, demonstrates additional variation in the global sensitivity depending on transport characteristics.