Assimilation Of Cloud-free Radiances From Fengyun-3C Microwave Sounding Observations

The MWTS-? and MWHTS sounders are important satellite payloads of FY-3C which is a ploar-orbiting satellite of China.The observations of the two instruments are of good quality and have important implications for data assimilation and Numerical Weather Prediction(NWP).In order to improve the utilization of microwave vertical sounding observations over land and in the low-level atmosphere,we proposed the following new schemes,including the land surface emissivity denoising algorithm based on the wavelet analysis,dynamic retrievals of land surface emissivity constrained by the wavelet emissivity atlas,and the GPR(Gaussian Process Regression)cloud detecting algorithm which can intelligently learn the internal relationship between the two types of sounding observations.These new schemes solve the problems such as accurately calculating the emissivity for microwave observations over land surface,screening the cloud-contaminated observations and assimilating these vertical sounding observations.The complexity of land surface makes it difficult to have a perfect model to accurately calculate the surface emissivity,but the land surface emissivity retrieved by satellite observation is no limitation of massive input surface parameters,such as soil roughness,soil moisture,etc.The monthly mean emissivity atlas was proposed by assuming the surface emissivity changes slowly in the same place.However,there is error in the atlas and is an image matrix essentially,we applied the wavlet threshold denoising method in the field of digital image processing to filter the noise in the atlas,and developed a new wavlet atlas which is suitable to the FY-3C sounders.The results show that,the wavelet atlas can increase the observation numbers over land surface,improve the effect of brightness temperature simulation and the quality of analysis,accompanying with the reduction in forecast skill.Due to the fact that the low-level microwave channels are more sensitive to the land surface,the climate-averaged and daily-constant emissivity atlas is no longer applicable to these low-level observations.In this paper,the new assimilation system process was designed and the application of MWTS-? microwave low-level sounding channel 4 was realized,based on the dynamic emissivity retrieval scheme which was meanwhile constrained by the wavelet emissivity atlas.The assimilation results show that the departure of first guess field is significantly reduced in the microwave middleand lower-level channels,due to the improvements in land surface emissivity.Comparing to wavelet emissivity atlas,the number of land observations from the middle and lower microwave channels rises by about 6-8 percent,and the addition of numerous new land surface observations has not adversely affected the analysis field and the result of short-term forecast was improved.We developed a new GPR cloud detecting scheme based on the combination of 118.75 GHz and 50-60 GHz double oxygen bands on FY-3C sounders.The new method can improve the cloud detection capacity of about 20%,compared to the traditional linear regression method.Finally,the assimilation and forecasting system framework for MWHTS is designed,evaluated and validated the application value and the contribution ability of MWHTS sounding channels.