Symmetric and antisymmetric components of polar-amplified warming

TL;DR

This study investigates the symmetric and antisymmetric components of polar-amplified warming in climate models, revealing that the symmetric component remains relatively stable over time, while the antisymmetric component diminishes, with implications for understanding polar climate responses.

Contribution

The paper introduces a combined analysis of symmetric and antisymmetric warming components using GCMs and models, highlighting the robustness of the symmetric component and its underlying mechanisms.

Findings

Symmetric warming component remains stable or slightly increases over time.

Antisymmetric component weakens and vanishes in some models.

Model simulations show modest changes in symmetric warming across different CO2 levels.

Abstract

CO$_2$-forced surface warming in general circulation models (GCMs) is initially polar-amplified in the Arctic but not Antarctic -- a largely hemispherically antisymmetric signal. Nevertheless, we show in CESM1 and eleven LongRunMIP GCMs that the hemispherically symmetric component of global-mean-normalized, zonal-mean warming ($T^*_\mathrm{sym}$) under 4\(\times\)CO$_2$ changes weakly or becomes moderately more polar-amplified from the first decade to near-equilibrium. Conversely, the antisymmetric warming component ($T^*_\mathrm{asym}$) weakens with time in all models, moderately in some including FAMOUS but effectively vanishing in others including CESM1. We explore mechanisms underlying the robust $T^*_\mathrm{sym}$ behavior with a diffusive moist energy balance model (MEBM), which given radiative feedback parameter ($\lambda$) and ocean heat uptake ($\mathcal{O}$) fields diagnosed from CESM1 adequately reproduces the CESM1 $T^*_\mathrm{sym}$ and $T^*_\mathrm{asym}$ fields. In further MEBM simulations perturbing $\lambda$ and $\mathcal{O}$, $T^*_\mathrm{sym}$ is sensitive to their symmetric components only, and more to that of $\lambda$. A three-box, two-timescale model fitted to FAMOUS and CESM1 reveals a curiously short Antarctic fast-response timescale in FAMOUS. In additional CESM1 simulations spanning a broader range of forcings, $T^*_\mathrm{sym}$ changes modestly across 2-16\(\times\)CO$_2$, and $T^*_\mathrm{sym}$ in a Pliocene-like simulation is more polar-amplified but likewise approximately time-invariant. Determining the real-world relevance of these behaviors -- which imply that a surprising amount of information about near-equilibrium polar amplification emerges within decades -- merits further study.