Mesoscale modelling of water vapour in the tropical UTLS: Two case studies from the HIBISCUS campaign

Abstract

In this study, we evaluate the ability of the BRAMS (Brazilian Regional Atmospheric Modeling System) mesoscale model compared to ECMWF global analysis to simulate the observed vertical variations of water vapour in the tropical upper troposphere and lower stratosphere (UTLS). The observations are balloon-borne measurements of water vapour mixing ratio and temperature from microSDLA (Tunable Diode Laser Spectrometer) instrument. Data from two balloon flights performed during the 2004 HIBISCUS field campaign are used to compare with the mesoscale simulations and to the ECMWF analysis. The observations exhibit fine scale vertical structures of water vapour of a few hundred meters height. The ECMWF vertical resolution (∼1 km) is too coarse to capture these vertical structures in the UTLS. With a vertical resolution similar to ECMWF, the mesoscale model performs better than ECMWF analysis for water vapour in the upper troposphere and similarly or slightly worse for temperature. The BRAMS model with 250 m vertical resolution is able to capture more of the observed fine scale vertical variations of water vapour compared to runs with a coarser vertical resolution. This is mainly related to: (i) the enhanced vertical resolution in the UTLS and (ii) to the more detailed microphysical parameterization providing ice supersaturations as in the observations. In near saturated or supersaturated layers, the mesoscale model predicted relative humidity with respect to ice saturation is close to observations provided that the temperature profile is realistic. For temperature, the ECMWF analysis gives good results partly attributed to data assimilation. The analysis of the mesoscale model results showed that the vertical variations of the water vapour profile depends on the dynamics in unsaturated layer while the microphysical processes play a major role in saturated/supersaturated layers. In the lower stratosphere, the ECMWF model and the BRAMS model give very similar water vapour profiles that are significantly drier than micro-SDLA measurements. This similarity comes from the fact that BRAMS is initialised using ECMWF analysis and that no mesoscale process acts in the stratosphere leading to no modification of the BRAMS results with respect to ECMWF analysis.