Assessing the performance of climate change simulation results from BESM-OA2.5 compared with a CMIP5 model ensemble

Abstract

Abstract. The main features of climate change patterns, as simulated by the coupled ocean–atmosphere version 2.5 of the Brazilian Earth System Model (BESM), are compared with those of 25 other CMIP5 models, focusing on temperature, precipitation, atmospheric circulation, and radiative feedbacks. The climate sensitivity to quadrupling the atmospheric CO2 concentration was investigated via two methods: linear regression (Gregory et al., 2004) and radiative kernels (Soden and Held, 2006; Soden et al., 2008). Radiative kernels from both the National Center for Atmospheric Research (NCAR) and the Geophysical Fluid Dynamics Laboratory (GFDL) were used to decompose the climate feedback responses of the CMIP5 models and BESM into different processes. By applying the linear regression method for equilibrium climate sensitivity (ECS) estimation, we obtained a BESM value close to the ensemble mean value. This study reveals that the BESM simulations yield zonally average feedbacks, as estimated from radiative kernels, that lie within the ensemble standard deviation. Exceptions were found in the high latitudes of the Northern Hemisphere and over the ocean near Antarctica, where BESM showed values for lapse rate, humidity feedback, and albedo that were marginally outside the standard deviation of the values from the CMIP5 multi-model ensemble. For those areas, BESM also featured a strong positive cloud feedback that appeared as an outlier compared with all analyzed models. However, BESM showed physically consistent changes in the temperature, precipitation, and atmospheric circulation patterns relative to the CMIP5 ensemble mean.