A21E-0105: Drivers of nonlinear behavior in the evolution of transient climates
Authors: Elisabeth J Moyer1, David J McInerney1, 2
Author Institutions: 1. Dept. of the Geophysical Sciences, University of Chicago, Chicago, IL, USA; 2. School of Civil, Environmental and Mining Engineering, University of Adelaide, Adelaide, SA, Australia
Numerous studies in recent years have shown robustly that over long timescales, heat uptake in general circulation models (GCMs) is not linear with global mean temperature. Instead, a rapid decline in heat uptake during initial warming is followed by long tail of gradual warming while heat uptake reduces more slowly relative to temperature. One implication is that extrapolating initial climate behavior would lead to underpredicting climate sensitivity and eventual equilibrium temperature. We use multi-millenial GCM runs to identify causes of this nonlinear behavior, driving the NCAR CCSM3 model (T31 resolution) with increasing CO2 and observing transient climate responses for > 5000 years after stabilization. Model results suggest that representation of long-term climate evolution as a changing ocean heat uptake efficacy does not capture the underlying physical processes. The evolution is better represented as a nonlinear response in shortwave cloud forcing that sets in only after several hundred years. We show further that these temporal changes do not primarily reflect nonlinearity in climate processes but instead are largely an artifact of inhomogeneous warming. Because local cloud feedbacks differ in magnitude and sign across the Earth’s surface, differential rates of warming, and especially the long delay in warming of the Southern Ocean, produce apparent nonlinear feedbacks even in the case of linear physics. Warming rates themselves are controlled by ocean turnover times, as is demonstrated by the tight relationship between nonlinear behavior in radiative forcing and transient precipitation response, though these signals derive from different physical mechanisms in different locations. Results are robust across forcing magnitudes and forcing agents. Though confirmation in other models would be important, we expect that these conclusions are qualitatively robust across models as well. The general principle of local control of cloud forcing combined with latitudinally differential warming should produce similar behavior even if specific cloud responses differ between models.