针对风云三号C星微波湿温探测仪（FY-3C/MWHTS）的陆地晴空观测资料，建立了一维变分反演系统，对大气的温湿廓线进行反演。为了更好地描述温湿廓线的相关性，同时减小温度和湿度在反演过程中相互之间的误差传递，提出了使用背景协方差矩阵的联合矩阵和单独矩阵进行组合反演的方法。对于MWHTS模拟亮温和观测亮温之间的偏差，使用逐扫描点的统计回归方法进行校正。选择中国部分陆地区域的晴空观测亮温进行温湿廓线的反演，并利用欧洲中期天气预报中心（ECMWF）再分析数据、美国国家环境预报中心（NCEP）分析数据以及无线电探空观测（RAOB）数据对反演结果进行验证，温湿廓线的反演结果与ECMWF再分析数据验证的最大均方根误差分别是2.59 K和11.87%，与NCEP分析数据验证的最大均方根误差分别是1.88 K和21.50%，与RAOB数据验证的最大均方根误差分别是3.43 K和25.48%，验证结果表明了反演结果的可靠性。另外与国外同类载荷AMSU观测亮温的物理方法和统计方法反演精度进行了对比，结果表明：MWHTS具有较强的湿度廓线以及高空温度廓线的探测能力，且针对MWHTS的观测亮温建立的一维变分反演系统具有较高的反演精度。与NCEP 6小时预报廓线的验证结果表明：反演的湿度廓线可以提高预报廓线的精度。

Land surface emissivity has a significant effect on the atmospheric temperature and humidity sounding from space. Given the complicated land surface, the calculated emissivity is generally with low accuracy. A one-dimensional variational retrieval system was built using measurements of microwave humidity and temperature sounder onboard the Fengyun-3C satellite (FY-3C/MWHTS) to improve retrieval accuracy and to reduce the computation complexity in retrieving atmospheric temperature and humidity profiles over land under clear sky.
By analyzing the a priori information affecting the accuracy of inversion, a hybrid retrieval approach based on united and individual matrices of background covariance is proposed. The method established a better correlation relationship between temperature and humidity profiles, reduced the error propagation of the retrieval temperature and humidity, and prevented the complicated land emissivity calculations according to different surface types. From the correlation between the observed values from FY-3C/MWHTS and those simulated by the forward radiative transfer model, a statistical regression method was also adapted in pixel-by-pixel correction procedure to correct the bias between observed and simulated values.
This retrieval system obtains temperature and humidity profiles over a part of China's land under clear sky and validates the retrieval results with respect to the European Centre for Medium-Range Weather Forecasts(ECMWF) reanalysis data, National Centers for Environmental Prediction(NCEP) analysis data, and Radiosonde Observation(RAOB) data. With respect to the ECMWF reanalyzed data, the maximum root mean square errors of the resulting temperature and relative humidity are 2.59 K and 11.87%, respectively. With respect to the NCEP analyzed data, the maximum root mean square errors of the resulting temperature and relative humidity are 1.88 K and 21.50%, respectively. With respect to the RAOB data, the maximum root mean square errors of the resulting temperature and relative humidity are 3.43 K and 25.48%, respectively. The comparison of the retrieval results with those measured by AMSU using the physical and statistical retrieval methods shows that the MWHTS has higher accuracy. The comparison of the retrieval results with the NCEP 6 h forecast profiles shows that the retrieval humidity profiles can improve the accuracy of the forecast profiles, particularly in the upper atmosphere.
The proposed hybrid approach using the united and individual matrices of background covariance can provide satisfactory retrieval accuracy, although land emissivity is calculated without classifying the surface types in the retrieval system built in this study. The retrieval system, whose retrieval results are evaluated by root mean square errors with respect to the three data sources, has high accuracy and high reliability. The comparison of the retrieval results with the measurements of AMSU indicates that MWHTS has a greater ability to probe the temperature in the upper atmosphere and humidity in the entire atmosphere. The comparison of the retrieval results with the NCEP 6 h forecast profiles indicates that MWHTS can obtain high-quality data for sounding atmosphere and is of significance to numerical weather prediction radiance assimilation.