Climate Model Sensitivity to Moist Convection Parameter Perturbations

Thursday, 18 December 2014: 5:30 PM
Diana N Bernstein and J David Neelin, University of California Los Angeles, Los Angeles, CA, United States
The perturbed physics ensemble in this study examines the impact of poorly constrained parameters in representations of subgrid-scale processes affecting moist convection using the National Center for Atmospheric Research fully coupled ocean-atmosphere Community Earth System Model. An ensemble of historical period simulations and an ensemble of end-of-the-century simulations under the Representative Concentration Pathway 8.5 scenario for global warming quantify some of the implications of parameter uncertainty for simulation of precipitation processes in current climate and in projections of climate change. Regional precipitation patterns prove highly sensitive to the parameter perturbations, especially in the tropics. In the historical period, nonlinear parameter response with local changes exceeding 3 mm/day is noted. Over the full range of parameters, the error with respect to observations is within the range typical of different climate models used in the Coupled Model Intercomparison Project phase 5 (CMIP5). For parameter perturbations within this range, differences in the end of century projections for global warming precipitation change also regionally exceed 3 mm/day, also comparable to differences in global warming predictions among the CMIP5 models. These results suggest that improving constraints within moist convective parameterized processes based on better assessment against observations in the historical period will be required to reduce the range of uncertainty in regional projections of precipitation change.