Retrieval of large-scale wind divergences from infrared Meteosat-5 brightness temperatures over the Indian Ocean

Over the tropics the atmospheric general circulation models usually fail in predicting horizontal wind divergence, which is closely related to atmospheric heating and to the vertical exchanges associated with convection. With the aim of forcing atmospheric models we present here a reconstruction of wind divergences based on the links between infrared brightness temperatures, convective activity, and large-scale divergence. In practice, wind divergences are reconstructed from brightness temperatures using correlations obtained from numerical simulations performed with a general circulation model. When building those correlations a distinction must be made between the brightness temperatures of opaque clouds and those of semitransparent clouds, only the former being directly associated with convection. In order to filter out semitransparent clouds we use radiative thresholds in the water vapor channel in addition to the window channel. We apply our approach to Meteosat-5 data over the Indian Ocean. Comparison with wind divergences reconstructed independently from Meteosat water vapor winds partially validates our retrieval. Comparison with European Center for Medium-Range Weather Forecasts analyses indicates that much can be gained by adding information on the wind divergence in the tropics to force an atmospheric model.