Abstract

The global ocean modulates the climate’s temperature response to forcing. Ocean turbulent mixing processes mediate this relationship, and global climate models (GCMs) often struggle to simulate this mixing, but most research has ignored ocean processes as an avenue for constraining predicted warming. Here we show that regional mixed layer depth (MLD) strongly constrains climate sensitivity because of its relationship to oceanic heat capacity, heat uptake, and the depth of the Atlantic Meridional Overturning Circulation. We fit a two-layer energy balance model (EBM) to each GCM in a 25-member CMIP6 ensemble, and correlate the parameters of the EBMs with average pre-forcing mixed layer depths in the northern (55N–75N), tropical (26S–26N), and southern oceans (65S–45S). Using these correlations and observations from the Argo float network, we revise the ensemble mean and narrow the 66% range of equilibrium climate sensitivity (ECS) for the particular CMIP6 model collection from 4.51 (3.13–5.71)C, to 4.66 (3.88–5.43)C, amounting to a 40% reduction in the span of the uncertainty range. Such a reduction in uncertainty rivals the impact of other critical processes, e.g., clouds, in their effect on overall surface warming projections