The radiative feedback pattern effect remains a large source of uncertainty for both projections of future climate change, as well as interpretations and attribution of warming trends over the last 40 years. In particular, it remains one of the major sources of poorly quantified and poorly constrained uncertainty for climate sensitivity. Most of the pattern effect, as well as the attendant uncertainty, is attributable to low cloud radiative feedbacks.

Here we use low cloud radiative kernels and cloud controlling factors to disentangle the drivers of the low cloud radiative effect (CRE) and its uncertainty in recent and future time periods, using both AMIP historical and abrupt quadrupling simulations.

We find that over the last forty years (i.e. the satellite record) changes in Estimated Inversion Strength (EIS) is the main driver low cloud CRE, while the large majority of the inter-model spread can be traced to model-spread in the sensitivity of low-clouds to EIS. Conversely, we find that the more uniform Sea Surface Temperature (SST) warming pattern in abrupt4xCO2 simulations leads to SSTs being the primary driver of the CRE, and the model-spread being related to the sensitivity of low clouds to SST changes.

The fact that drivers of the net feedback is expected to change in the future should lead to increased caution when trying to use historical global-mean changes in Earth’s energy budget to place constraints in future warming.