Scientific issues and perspective of assimilation of meteorolo gical satellite data

The main features of the observations by meteorological satellites and related s cientific issues in terms of data assimilation in numerical weather prediction a re analyzed in comparison with the in situ data. The raw observations of the sat elliteborne instruments are electromagnetic radiances which may not be digeste d directly by the NWP(Numerical Weather Prediction) models. The measured radianc es are affected by the state of the whole column of the atmosphere rather than a n individual level, so that the sensitive height and the equivalent vertical res olution of the satel lite observations are crucial. Different from most conventional observations, the measurements from the satellite are not biasfree. In order to assimilate the satellite data, it is fundamental to convert the model predictands to the elect romagnetic radiances of specific channels sensed by the satelliteborne instrum ents or conversely. The former, which is so called direct assimilation, needs co mplicated observational operators, but in the latter, known as indirect assimila tion, the retrieval must be included as one step of the data preprocessing. The choice of direct or indirect strategies depends on the capability of the data as similation system in dealing with the complicated observations and is crucial fo r the effects of the satellite data usage in NWP. Following the general discussi ons, the achievements in assimilation of five main categories of satellite data, i.e. the satellite soundings, satellite derived atmospheric motion vectors, spa ce based scatterometors, precipitation affected radiances and GPS radio occultat ion data are reviewed with the stress on the discussion of the further improveme nts. The progresses of satellite data assimilation in China are also reviewed fo llowed by concerns for the near future developments.