Temperature-Dependent Feedbacks and Large Increases in Climate Sensitivity

Abstract:

The strength of the climate feedbacks by which the Earth responds to a given radiative forcing depend themselves on the climate state of the Earth. If the sum of these feedbacks gets more positive as the Earth warms, the warming response can become highly nonlinear, and sensitivity can greatly increase. Such increases are difficult to model properly with general circulation models (GCMs), as these increases can push GCMs into regimes outside of their configurations, causing numerical and physical blow-up. Understanding these cases is vital to understanding the risks posed by anthropogenic forcings.

In this talk, we present results from two approaches to studying feedback temperature dependence. The first explores our ability to model highly nonlinear behavior using a perturbed physics ensemble in which parameters controlling convection and cloud formation are varied in several versions of the CAM atmospheric models. The feedback temperature dependence of the resulting ensemble members is estimated. We test our hypothesis that existing GCMs lie close to (or past) the threshold of extreme nonlinear behavior. We also test this hypothesis by analyzing the RCP8.5 runs in the CMIP5 archive.

The second approach uses observations and paleoclimate proxies to estimate how sensitivity changes under warming. An established literature uses estimates of the temperature changes and forcings from the instrumental record to estimate modern day sensitivity. We combine these estimates with temperature proxies and δ13C records of the Paleocene-Eocene Thermal Maximum to estimate the sensitivity of the Paleocene, as well as the temperature dependence of the overall climate feedback. Consideration is given to other sources of nonlinear sensitivity, such as changes in boundary conditions. We find a substantial increase in sensitivity under warmer temperatures.