General Circulation Models evaluation at different time scales over tropical region using ESA-CCI satellite data records: a case study of water vapour and cloud cover

Abstract. Water vapour and cloud cover are two essential components of the earth's atmosphere. General circulation models (GCM) are used to study the long term evolution of the Earth's climate over past and future periods. The present work consists of assessing the representation of total column water vapor (TCWV) and total cloud cover (TCC) in the Atmospheric Model Intercomparison Project Phase 6 (AMIP6), the ERA5 reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF), and satellite data records from the European Space Agency Climate Change Initiative (ESA-CCI). ESA-CCI is used as the reference for the common observation period with AMIP6, spanning from July 2003 to December 2014, to calibrate the framework. For the period prior to the observational period, from January 1981 to June 2003, ERA5 serves as the reference. This study is carried out over the tropical region which has been splitted in two sub-regions: the tropical oceans and tropical lands. The assessment of TCWV and TCC at different time-frequency is performed using a mathematical tool called "multi-resolution analysis" (MRA). By applying the MRA decomposition, we found that the AMIP6 models produce consistent evolution of TCWV and TCC at seasonal to interannual scales (from 2 months to 5.6 years) in the tropical region, even if the representation of the amplitude of TCC remains sometimes challenging. The evaluation of ESA-CCI TCWV and TCC variability in AMIP6 models reveals that the models do not perform well at daily and subseasonal scales. At seasonal to interannual scales, the models reproduce more accurate variability of TCWV and TCC with respect to ESA-CCI. However, AMIP6 models do not capture the trend in the evolution of ESA-CCI TCWV and TCC. The co-variations between TCWV and TCC were analyzed in the Nino3.4 region, revealing a significant positive correlation at the subseasonal scale, with a value of 0.7 for ESA-CCI and 0.3 for AMIP6. At seasonal to annual scales, we found a strong positive correlation between TCWV and TCC, with the exception of the CanESM5 and IPSL models, which showed a negative but significant correlation around -0.5.