Weak dependence of future global mean warming on the background climate state

It is debated whether the global mean transient temperature response to forcing is largely independent of biases in the simulated base state climate. To test this, we run the Community Earth System Model (CESM1) into new quasi-equilibria by altering the solar constant by ± 25 Wm −2 and initialize idealized CO 2 climate change experiments from these climate states. The simulations branched off from the warm and control states, with increased and unaltered solar constant, respectively, show similar global mean feedback parameters, effective CO 2 radiative forcings (ERF) and ocean heat uptake efficiencies and therefore simulate indiscernibly different amounts of global warming. The experiments starting from the cold climate state behave differently: The CO 2 ERF is lower mostly because of rapid adjustments in the clouds, and the climate sensitivity is enhanced, mainly due to a large surface albedo feedback caused by the increased sea ice and snow coverage that extends to lower latitudes. The global heat uptake efficiency is reduced in the cold state. While taking up more heat in the Southern Ocean than the experiments from the warmer states much less heat is taken up by the North Atlantic. The less stabilizing net climate feedback parameter and the decreased ocean heat uptake efficiency dominate over the reduced CO 2 ERF, and the simulations initialized from the cold state show ~ 10% more global warming compared to those from the control state. This enhanced warming mainly occurs at higher-latitudes over sea ice and snow-covered areas and the North Atlantic. While the local climate response may depend strongly on the base state, the future global mean temperature increase simulated by CESM1 is remarkably independent on the initial climate state.