The Use of Microwave Satellite Data Affected by Cloud in Numerical Forecast of Typhoon

Satellite data is being used in numerical weather prediction and takes up the role of main data source among the observations used. The quality of initial condition and the accuracy of numerical weather forecast are greatly improved by using satellite data. It is extremely important to use the satellite data in the numerical forecast of typhoon for there are few regular observations on the sea. However, only cloud-cleared satellite data is used in most current data assimilation systems because of the complexity to cope with the radiant effect of cloud and rain particles in radiation transfer model. The cloudy and rainy reg ion is always the sensitive area.The observation in these areas implies much information to weather system and has great impacts on the numerical forecast. There is a crucial need to handle the satellite data affected by cloud and rain in numerical weather forecast, but with caution at the same time. At present, the satellite data affected by cloud is examined through the cloud examination method.The No. 0604 typhoon Bilis is taken as a research case. A set of experiments are designed to use the satellite microwave data in cloudy area based on the cloud examination scheme. Scattering index, precipitation probability and precipitation examination are taken for AMS U-A. For AMS U-B, the bias between simulated bright tempe rature and observation of AMS U-B channel 2 and Bennartz scattering index are utilized. Different satellite data affected by cloud is used in data assimilation system by different cloud examinations and channel selection schemes.The screening of satellite data affected by cloud, together with their influences on the numerical forecast of Bilis's three periods, corresponding to formation, maturation and landing, is examined respectively. The result shows that more AMSU-A data is screened by scattering index than precipitation probability and precipitation examination. The scattering index with threshold 15 is suitable for the use of AMSU-A data in regional model. For AMSU-B, the bias between simulated bright temperature and observation of AMSU-B channel 2 performs better than Bennartz scattering index.