Evaluation of subtropical marine low stratiform clouds in CMIP5 AMIP experiments using ISCCP observations and simulator outputs

Abstract:

Subtropical marine low stratiform clouds (LSCs), which are quite important for the earth's climate in terms of cloud radiative effect, simulated in AMIP experiments by COSP ISCCP simulators implemented in nine CMIP5 models are evaluated by comparing with ISCCP observational data. In regions with large amounts of higher level clouds, low clouds observed from satellites can be obscured by them, especially by using passive sensors like ISCCP; however, high cloud amounts are climatologically small in the subtropical regions, corresponding to the descending branches of the Hadley circulation. Although middle cloud amounts are smaller in lower latitudes, they are relatively large in the eastern subtropical oceans such as off Peruvian coast. These clouds are not middle but indeed low clouds; this is because of the misdetection due to their cold cloud top temperatures. In addition, optically thick low clouds correspond to LSCs, while thin to shallow cumuliform clouds, generally in the ISCCP classification. Therefore, to facilitate robust evaluation, only ocean areas between 40N and 40S where climatological high cloud amounts are less than 20% are considered, and middle+low cloud amounts, corrected with the random-overlap assumption, with optical thickness >9.4 are defined as "the LSC amount" in this study. From these definitions, five LSC regions (Californian, Canarian, Peruvian, Namibian, and Australian) are detected in the ISCCP observations. We also derived the LSC amounts from the ISCCP simulator outputs of each CMIP5 model. CCSM4, HadGEM2-A, and MRI-CGCM3 are relatively well simulated, whereas IPSL-CM5B-LR and bcc-csm1-1-m produce similar patterns but with greater LSC amounts. CanAM4 and MPI-ESM-LR show noisy spatial distributions. The LSC amounts in CNRM-CM5 and MIROC5 are much less; they are smaller in regions closer to land. Comparing with the satellite observations and reanalysis data, the relationships with meteorological processes will be discussed.