Comparison of radiative feedbacks and climate sensitivities between CCSM3 and CCSM4

Equilibrium climate sensitivity in the CCSM4 is about a half degree higher than was cited for CCSM3 by Kiehl et al. (2006). Using the radiative kernel technique, we identify the radiative feedbacks that explain the increase. One of the main factors is a new formulation of the slab ocean model (SOM) itself. The older SOM in CCSM3 attempted to produce an observed SST and sea ice climatology, and the sea ice was motionless. In contrast, the new SOM in CCSM4 (also implemented and tested in CCSM3) produces an SST and sea ice climatology almost indistinguishable from the SST and sea ice in the fully coupled model (i.e., with the dynamic ocean general circulation model). The new SOM also has a dynamic sea ice model, as in the fully coupled model.

The radiative kernel technique indicates that the lapse rate feedback declines in magnitude (a warming effect) from CCSM3 to CCSM4. While the (positive) global average longwave cloud feedback decreases slightly, CCSM3 shows a larger regional variation in both positive and negative TOA energy budget contributions than CCSM4 does. Finally, the cloud shortwave feedback becomes more positive. In all cases, we see a monotonic change from CCSM3 with the old SOM to CCSM3 with the new SOM to CCSM4.

When we analyze the runs taking our averaging period later in the run and for a longer period of time than was used by Kiehl et al, we do not find a significant resolution dependence in CCSM3 with the old SOM. There is also no significant resolution dependence in either CCSM3 or CCSM4 with the new SOM. We are further investigating the influence of the averaging period on the corresponding feedback and sensitivity calculations as well as the importance of the years used to create the qflux data for the SOM.