云区星载微波观测直接变分同化的关键问题研究

受云和降水影响而不能被同化系统吸收的卫星资料占了全部卫星观测的绝大多数，然而，云和降水对天气预报及其预报准确率的提高尤为重要，因此，云、雨区卫星观测资料的同化应用是数值预报准确率继续获得改善的重要技术途径之一，是各国数值预报中心卫星资料应用的一个研究热点。本项目拟开展星载微波成像仪云雨区观测的直接变分同化研究：基于微波成像仪观测在对其进行无线电频率干扰识别和订正的基础上，采用一维变分算法反演云水、雨水、云冰等水成物垂直廓线，补充资料同化系统的输入背景场变量；在GSI三维变分同化系统中增加云水、雨水和云冰混合比作为控制变量，背景误差协方差矩阵中增加这3个水成物的量，实现云雨区资料的直接同化；用模拟和实测的资料进行同化效果检验。以期提高云、雨区卫星资料的使用率，进而提高数值预报模式初始场即同化分析场的质量，同时为我国FY-3微波资料在GRAPES系统中实现有云同化奠定基础。

In global or regional numerical weather prediction (NWP) systems, most satellite observations are discarded due to cloud and precipitation contamination and unknown surface emissivity. Improvements in cloud and precipitation assimilation are likely necessary for continuing significant gains in weather forecasting. The research on cloud- and precipitation- affected observations assimilation is the hot issue in current NWP data assimilation systems. Cloud- and precipitation- affected radiances of spaceborne microwave imager directly assimilated into 3D-Var data assimilation system will be studied in the proposal. First, the profiles of liquid and ice cloud, rain, snow, graupel and hail are retrieved by one dimensional variational algorithm based on AMSR-E or AMSR-2 measurments. These hydrometeors profiles are complemented as the input to the observation operators (radiative transform model), in which cloud absorption and scattering effect is considered. The measurements from different observation sources are used to validate the retrieval accuracy and calculate the statistical retrieval error. The radio-frequency interference (RFI) from X- and C-band channels need to be detected. The RFI effect on the above retrieval accuracy and RFI correction will be studied. Then, the scheme of assimilating cloud- and precititation- affected AMSR-E data into 3D-Var GSI data assimilation system will be developed and technical implemented. Retrieved hydrometeor parameters (mixing ratio of cloud water, rain and cloud ice) are included as control variables and the statistical hydrometeor retrieval errors are added to the background error covariance matrix. The higher data usages under cloud or precipitation conditions and the improved quality of NWP initial fields (namely analysis fields of data assimilation) are expected. The research will lay the foundation for our MWRI (microwave imager onboard FY-3 series satellites) cloudy data assimilation in GRAPES.