Deep-Time Climates as Stress Tests for Earth System Models

Atmospheric CO2 concentrations continue to rise, yet uncertainty in equilibrium climate sensitivity (ECS) remains large, with implications for future warming projections. In this talk, I use simulations of past climate states to assess Earth system model performance and to diagnose sources of model bias. By comparing multiple generations of the Community Earth System Model (CESM) against paleoclimate proxy constraints, I show that very high ECS values (as in CESM2) overestimate both Eocene warmth and glacial cooling, whereas a moderate, state-dependent ECS - one that increases with warming - better reconciles models and data. These results demonstrate how paleoclimate benchmarks can identify physical and parameterization issues (especially in cloud microphysics and turbulence schemes), refine model sensitivity, provide insights into the longstanding “equable climate problem”, and ultimately improve confidence in projections of future climate change.

Speaker: Chris Poulsen, University of Oregon