The Lower Uncertainty Bound of Climate Sensitivity in Gcms: How Low Can We Go?...

Abstract:

The equilibrium climate sensitivity (ECS) is one of the most important metrics of climate change. As such, constraining the uncertainties of its magnitude, and the magnitude of its transient counterpart (TCR), is one of the primary goals of global climate science. General circulations models (GCMs) from modelling centres around the world have consistently failed to produce a model with a sensitivity of less than 2 degrees. However, as the CMIP5 multi-model ensemble is an ensemble of opportunity, it is unclear whether this fact is sufficient to rule out climate sensitivity of less than 2 degrees, or is the ensemble simply not diverse enough to sample low values of climate sensitivity? We present analysis based on the observed planetary energy budget and simple energy-balance models. When view in terms of the TCR:ECS ratio (RWF- the Realised Warming Fraction), we find a region of climate response space of low RWF and low TCR that is robust to the structure of the simple climate model and isn't sampled by the CMIP5 ensemble. We show that this region is better sampled by a perturbed physics ensemble of the HadCM3 GCM constrained solely on top of atmosphere radiative fluxes than the CMIP5 ensemble, raising the question of the physical plausibility of low climate sensitivity GCMs. Based on our results above, we have set out to systematically probe the ability to create GCMs with low climate sensitivity in the HadCM3 GCM. We train a statistical emulator on our perturbed physics ensemble and use it to identify regions of HadCM3 parameter space that are consistent with both a low climate sensitivity and a low RWF. We then run this “low sensitivity” ensemble to test our predictions and understand the combination of feedbacks needed to produce a sensible GCM with a sensitivity of less than 2 degrees. Here we hope to demonstrate our results from this systematic probing of the low climate sensitivity uncertainty bound and add further understanding to the physical plausibility of low climate sensitivity as simulated in GCMs.