Quantify contribution of aerosol errors to cloud fraction biases in CMIP5 Atmospheric Model Intercomparison Project simulations

Identifying the error sources in total cloud fraction (CF) simulated by global climate models is essential for improving climate prediction. This study investigates if and how significant the aerosol simulation errors contribute to the model CF biases in the Atmosphere Model Inter-comparison Project (AMIP) simulations of the Coupled Model Inter-comparison Project Phase 5 (CMIP5) models. The aerosol optical depths (AODs) and CFs in 12 CMIP5/AMIP models have been examined and compared with 8-year moderate resolution imaging spectroradiometer (MODIS) satellite observations. The results show that the global-averaged multi-model ensemble mean AOD and CF, which are.14 and 56.2%, are 22.2 and 15.2% lower than those from MODIS, respectively. The simulated relationship between AOD and CF generally agrees with the observation on the global scale but differs on regional scale. Based on the “conditional sampling approach,” the AOD simulation errors that affect the CF biases of the models were separated from the model biases caused by the aerosol–CF errors that are related to dynamics, thermodynamics, and microphysics. It is found that the AOD errors barely contribute to the CF biases for most CMIP5/AMIP models on the global scale. Instead, simulated aerosol–CF errors are still the major contributors to the CF biases. However, we should note that AOD biases contribution in some regions, such as south Indian Ocean, Asia, Europe, and North Pacific Ocean, cannot be ignored. We also found that with increasing cloud liquid water path the CF does not increase with AOD as sensitively in the CMIP5/AMIP models as in the MODIS observations.