Climate sensitivity, the pattern effect, and cloud parametrisation

Climate sensitivity changes over time in numerical global climate models (GCMs) due to a so-called “pattern effect”. That is, surface-warming patterns evolve over time to favour different geographical regions giving rise to different climate feedbacks, thus changing climate sensitivity over time. One of the most important climate feedbacks is the cloud feedback and it has been shown that the pattern effect may strongly impact the strength of this feedback in GCMs. Here we perform slab-ocean model simulations with different versions of the Community Earth System Model (CESM). Different patterns of ocean heat transport convergence (Q-flux) are prescribed, inducing different patterns of surface warming. Notably, the prescribed Q-flux changes average to zero in the global mean, thus introducing no net forcing. We show that (1) net-zero forcing Q-flux changes can have surprisingly large effects on the climate, (2) that the impact strongly depends on the geographic pattern of the Q-flux change and, (3) that different cloud parametrisations may imply different impacts of the same patterns. While these results may have important implications for the quantification of the pattern effect and climate sensitivity in climate models, we caution against overinterpretation, as preliminary experiments with fully coupled models indicate a weaker sensitivity to similar pattern changes.