Warm mid-Pliocene conditions without high climate sensitivity: the CCSM4-Utrecht (CESM 1.0.5) contribution to the PlioMIP2

Abstract

We present the Utrecht contribution to the Pliocene Model Intercomparison Project Phase 2 (PlioMIP2), using the Community Earth System Model version 1.0.5 (CCSM4-Utr). Using a standard pre-industrial configuration and the enhanced PlioMIP2 set of boundary conditions, we perform a set of simulations at various levels of atmospheric pCO2 (280, 400, and 560 ppm). This allows us to make an assessment of the mid-Pliocene reference (Eoi400) climate versus available proxy records and a pre-industrial control (E280), as well as determine the sensitivity to different external forcing mechanisms. We find that our simulated Pliocene climate is considerably warmer than the pre-industrial reference, even under the same levels of atmospheric pCO2. Compared to the E280 case, our simulated Eoi400 climate is on average almost 5 °C warmer at the surface. Our Eoi400 case is among the warmest within the PlioMIP2 ensemble and only comparable to the results of models with a much higher climate sensitivity (i.e. CESM2, EC-Earth3.3, and HadGEM3). This is accompanied by a considerable polar amplification factor, increased globally averaged precipitation, and greatly reduced sea ice cover with respect to the pre-industrial reference. In addition to radiative feedbacks (mainly surface albedo, CO2, and water vapour), a major contribution to the enhanced Pliocene warmth in these simulations is the warm model initialisation followed by a long spin-up, as opposed to starting from pre-industrial or present-day conditions. Added warmth in the deep ocean is partly the result of using an altered vertical mixing parameterisation in the Pliocene simulations, but this has a negligible effect at the surface. We find a stronger and deeper Atlantic meridional overturning circulation (AMOC) in the Eoi400 case, but the associated meridional heat transport is mostly unaffected. In addition to the mean state, we find significant shifts in the behaviour of the dominant modes of variability at annual to decadal timescales. The Eoi400 El Niño-Southern Oscillation (ENSO) amplitude is greatly reduced (-68 %) versus the E280 one, while the AMOC becomes more variable. There is also a strong coupling between AMOC strength and North Atlantic sea surface temperature (SST) variability in the Eoi400, while North Pacific SST anomalies seem to have a reduced global influence with respect to the E280 through the weakened ENSO.