Uncertainty Quantification of Equilibrium Climate Sensitivity

Significant uncertainties exist in the temperature response of the climate system to changes in the levels of atmospheric carbon dioxide. We report progress to quantify the uncertainties of equilibrium climate sensitivity using perturbed parameter ensembles of the Community Earth System Model (CESM). Through a strategic initiative at the Lawrence Livermore National Laboratory, we have been developing uncertainty quantification (UQ) methods and incorporating them into a software framework called the UQ Pipeline. We have applied this framework to generate a large number of ensemble simulations using Latin Hypercube and other schemes to sample up to three dozen uncertain parameters in the atmospheric (CAM) and sea ice (CICE) model components of CESM. The parameters sampled are related to many highly uncertain processes, including deep and shallow convection, boundary layer turbulence, cloud optical and microphysical properties, and sea ice albedo. An extensive ensemble database comprised of more than 46,000 simulated climate-model-years of recent climate conditions has been assembled. This database is being used to train surrogate models of CESM responses and to perform statistical calibrations of the CAM and CICE models given observational data constraints. The calibrated models serve as a basis for propagating uncertainties forward through climate change simulations using a slab ocean model configuration of CESM. This procedure is being used to quantify the probability density function of equilibrium climate sensitivity accounting for uncertainties in climate model processes. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 and was funded by the Uncertainty Quantification Strategic Initiative Laboratory Directed Research and Development Project at LLNL under project tracking code 10-SI-013. (LLNL-ABS-491765)