Variational analysis of humidity information from TOVS radiances

The impact of assimilating TOVS radiance data in the European Centre for Medium‐Range Weather Forecasts humidity analysis is evaluated. It has been found that the introduction of a one‐dimensional variational analysis scheme (1DVAR) applied to the TOVS radiances significantly improves the representation of many aspects of the hydrological cycle. The theoretical information content of the TOVS radiance data is discussed and found to be consistent with significant changes observed in the mid upper‐tropospheric moisture fields when the radiance data are assimilated. In particular, a tendency of the model (without radiance assimilation) to produce a tropical humidity structure that is far too dry is removed, and excessively moist conditions in the southern sub‐tropics are improved. It is argued that the latter problem originates from the use of operational NESDIS retrieved products in the analysis. The humidity adjustments caused by the assimilation of TOVS radiances are accompanied by significant changes in the model dynamics, especially the description of the tropical Hadley circulation. One such case is described in detail, where the moistening of the tropics and drying of the sub‐tropics has resulted in a stronger mean analysed meridional circulation in the Atlantic. The analysis changes are also shown to improve the medium‐range forecasting of humidity, together with some associated benefit in the prediction of cloud and precipitation.