Initial equatorial Pacific cooling due to CO2 forcing shaped by internal variability

In this contribution, we argue that other processes beyond the ocean dynamical thermostat mechanism can be key for initial equatorial Pacific cooling from a quadrupling of CO₂ concentration, with those processes being influenced by internal climate variability. The thermostat mechanism, which is proposed to explain the cooling trend observed in recent decades, suggests that the cold upwelling in the eastern equatorial Pacific has delayed the warming of the equatorial Pacific from rising CO₂. We run a large-ensemble of 250 simulations of an abrupt CO₂ quadrupling with the Max Planck Institute - Earth System Model (MPI-ESM) climate model. While the ensemble mean shows weak initial warming in the equatorial Pacific in the first 2 years, compared to global changes, the individual ensemble members show a wide spread of responses, with 42 out of the 250 members simulating a cooling. We then separate between the 42 cooling members and the 46 members that warm the most and analyze the upper-ocean energy budget. The main driver distinguishing the two groups is the change in the meridional heat transport, with energy divergence driving cooling in the central-western Pacific and energy convergence driving warming in the eastern Pacific. This is mainly caused by the change in ocean meridional velocities and is amplified by the change in the meridional temperature gradient. Strengthened easterlies over the central and western Equatorial Pacific increase Ekman transport away from the Equator that drives cooling, while weakened easterlies decrease Ekman transport, warming the eastern equatorial Pacific. In contrast to the meridional heat transport, cooling due to the ocean dynamical thermostat mechanism from vertical heat transport is similar in all the cases and cannot explain the ensemble spread. Other contributions to the energy budget play a minor role, such as the shortwave surface radiation, or are a response to the temperature anomaly rather than a driver, such as the latent heat flux. Over longer, multidecadal timescales, both cooling and warming simulations converge to show amplified warming in the eastern equatorial Pacific, consistent with past studies. Our findings suggest that other mechanisms can be more important than the thermostat mechanisms for cooling the equatorial Pacific, with a large impact of internal variability. This highlights the need for large ensembles of simulations in studies of the initial response to increasing CO₂.