Evolution over two decades of the tropical clouds in a subsidence area and their relation to large-scale environment

The interannual variability of cloud properties in a tropical subsidence area (South Atlantic Ocean) is examined using 23 years of ISCCP cloud fractions and optical depths, complemented with ISCCP/Meteosat visible reflectance and a four-years comparison with CALIPSO-GOCCP products. The mean seasonal cloud properties are examined in the area, as their interannual evolution. Circulation regimes (characterized with the SST and w500 from NCEP and ERA-Interim) that dominate summer and winter are also examined, and atmospheric situations are classified in five circulation regimes: ascending air masses, and moderate or strong subsidence with warm or cold SSTs. We examine the mean cloud cover, optical depth, and reflectance in each regime and their evolution in time over 23 years. Observational results (mean values and interannual variability) are compared with simulations from the IPSL and CNRM climate models (part of the CMIP5 experiment), using simulators to ensure that differences can be attributed to model defects. It results that regime occurrence strongly depends on the dataset (NCEP or ERA-Interim), as do their evolution in time along 23 years. The observed cloud cover is stable in time and weakly regime-dependent, whereas the cloud optical depth and reflectance are clearly regime-dependent. Some cloud properties trends actually do exist only in some particular regimes. Compared to observations, models underestimate cloud cover and overestimate cloud optical depth and reflectance. Climate models poorly reproduce regime occurrence and their evolution in time, as well as variations in cloud properties associated with regime change. It means that errors in the simulation of clouds from climate models are firstly due to errors in the simulation of the dynamic and thermodynamic environmental conditions.