On the relative strength of radiative feedbacks under climate variability and change

Using the method of radiative ‘kernels’, an analysis is made of feedbacks in models participating in the World Climate Research Program Coupled Model Intercomparison Project phase 5. Feedbacks are calculated for RCP8.5 and abrupt4xCO2 experiments as well as for interannual and decadal variability from pre-industrial runs. Regressions across models are used to elicit relationships across experiments/timescales. Feedbacks between RCP8.5 and abrupt4xCO2 experiments show strong relationships, as expected from surface temperature response similarities arising from the two experiments. The analysis also reveals significant relationships between RCP8.5 and decadal and interannual lapse rate feedback, decadal water vapour and interannual total cloud—the latter confirming results elsewhere. To reveal the impact of warming pattern differences, ‘synthetic’ feedbacks are also generated, based on RCP8.5, whereby local feedbacks determined from that experiment are scaled by relative temperature changes (per degree of global warming) from the others. The synthetic feedbacks indicate that the (sometimes strongly) differing temperature response patterns themselves should not preclude strong correlations between variability and climate change feedbacks—indeed such correlations would be close if local feedbacks were a robust feature of the climate. Although such close correlations are not manifest, the synthetic feedbacks predict the interannual and decadal feedbacks to some extent (are correlated across models), and reveal the consistency, to a first approximation, of the mean model strength of variability feedbacks. Although cloud feedbacks at interannual timescales are correlated with those from RCP8.5, and show consistency with the strength of synthetic feedbacks, separate long and short wave components reveal very different, compensating, latitudinal patterns, suggesting the close correlation may be fortuitous.