In a recent study (Luo et al. 2006, J Climate, in press), we have analyzed ten years of measurements of tropical upper-tropospheric water vapor (UTWV) by MOZAIC (Measurement of Ozone and Water Vapor by Airbus In-Service Aircraft) and described its climatology, variability, transport and relation to deep convection. The MOZAIC project funded by the European Union features more than 10 years of continuous measurements of water vapor and ozone around the globe (with the exception of the Pacific and Australia) with a quality-assured mean accuracy of 5% for upper-tropospheric humidity (UTH). The uniqueness of the MOZAIC data makes them a valuable source for evaluating other UTWV products, especially those from the analysis and reanalysis. In this study, we conducted a thorough comparison of the MOZAIC measurements with the ECMWF product over three regions – the tropical Atlantic, tropical Africa and Asian monsoon region.

The seasonal migration of the UTH in the ECMWF compares well with that of the MOZAIC, keeping close pace with the ITCZ. Regional differences that were identified in the MOZAIC (e.g. tropical Africa and Asian monsoon regions being moister than the tropical Atlantic) are also well represented in the ECMWF data. Some deficiencies, however, are also found. On average, tropical upper troposphere is about 20% drier (in relative humidity with respect to ice or RHi) in the ECMWF than the MOZAIC. The bimodal distribution of UTH seen rather ubiquitously in the MOZAIC is largely absent in ECMWF: the dry mode still remains but the moist mode is cut off right at 100% RHi with virtually no occurrence of supersaturation. The absence of supersaturation in the ECMWF explains the overall dry bias in the mean UTH. Another problem with the ECMWF is that UTH changes little with height, whereas the MOZAIC shows that UTH generally increase with height by 10 – 20% RHi from about 300 to 200 hPa. Other comparisons such as interannual variability and moisture fluxes will be presented.

These comparison results suggest that ECMWF model has good skills in simulating the general distribution and annual cycle of the tropical UTH probably because of the well-simulated large-scale circulation that controls these general features of UTH. Nevertheless, UTH in the ECMWF is also sensitive to the convection and cloud parameterizations used in the data assimilation system, just like in any general circulation and climate models. The absence of supersaturation is very likely linked to the details of these parameterizations. The lack of vertical dependence of UTH may also be attributable to this deficiency as we expect to see in the real world more supersaturation around the 200-hPa level than the 300-hPa level because the former level is closer to the climatological detrainment level of deep convection.