大气温度、 水汽、 地表温度和地表发射率是大气和地表的本征信息量。 利用卫星红外资料精确反演大气温湿廓线有利于准确预报天气和研究气候变化, 同时地表温度和地表发射率光谱的反演为研究植物生长与作物产量、 地表水分蒸发与循环、 能量平衡、 地表成分及物理性质、 气候变迁与全球环境提供重要参数指标。 把大气和地面作为一个整体系统来考虑, 建立一种能同步反演大气温度廓线、 大气水汽廓线、 地表温度和地表发射率的反演方法, 利用超光谱红外卫星资料（atmospheric infrared sounder, AIRS）, 针对我国新疆地区沙漠和雪地两种典型发射率地表同步反演大气温度廓线、 水汽廓线、 地表温度和地表发射率。 反演方法首先线性化地球-大气系统红外辐射传输方程, 提出通过经验正交函数构建大气廓线和地表发射率光谱, 有效减少反演变量数, 建立同步物理反演模式, 然后以美国国家环境预报中心（National Centers for Environmental Prediction, NCEP）的预报结果（初始大气温度、 水汽廓线以及地表参数）作为初始值, 最后通过牛顿迭代得到最优化解。 反演观测区域覆盖我国新疆塔克拉玛干沙漠和准噶尔盆地, 分别选择位于塔克拉玛干沙漠腹地的塔中探测站（纬度38.98°, 经度83.64°）和准噶尔盆地的阜康荒漠生态系统国家野外科学观测研究站（纬度44.2°, 经度87.9° ）为反演地面验证点。 反演结果表明, 塔克拉玛干沙漠地表温度明显高于准噶尔盆地地表温度, 与实际情况相一致; 根据反演的8.6和13.4 μm处的地表发射率分布情况, 可以看出在8.6 μm处沙漠地表发射率明显低于雪地发射率, 在6～15 μm范围内, 反演的沙漠地区（塔中站）地表发射率和雪地地区（阜康站）地表发射率与美国喷气推进实验室测量的沙漠发射率光谱和雪地发射率光谱相一致。 研究表明, 把大气和地面作为一个整体系统来考虑, 把地表发射率加入到反演中, 通过比较和分析沙漠地区（塔中）和雪地地区（阜康）的大气廓线反演结果与当地气象探空值和传统反演方法反演值, 改进了大气温度廓线和水汽廓线反演精度, 特别是边界层温度和水汽改进尤为明显; 同时分析表明在发射率光谱变化较大的沙漠地区, 大气廓线反演精度的改进比雪地要高, 这是由于地表发射率光谱在沙漠、 戈壁地区变化较大, 而雪地的发射率光谱变化不大。 用该方法针对地表发射率光谱变化较大的地区（沙漠）同步反演大气廓线、 地表温度和地表发射率, 可以更有效的提高大气温度廓线、 水汽廓线的反演精度。 该研究结果可以为数值天气预报和我国未来超光谱红外卫星应用提供服务和有力支持, 具有十分重要的意义。Hyperspectral Infrared Satellite Data

Atmospheric temperature, water vapor, surface skin temperature and surface emissivity are the intrinsic information of the atmosphere and surface of earth. Retrieval of atmospheric temperature profile and water vapour profile is important for accurate weather forecasting and climate change research by using satellite infrared data, at the same time, the retrieved surface skin temperature and surface emissivity spectra were used to study the growth of plant and crop yield, evaporation and circulation of surface water, energy balance, surface composition and physical properties, climate change and global environment. In this paper, considering the atmosphere and the ground as a whole system, the retrieval method for simultaneous retrieval of atmospheric temperature profiles, water vapour profiles, surface emissivity, and surface skin temperature was established. simultaneous retrieval were performed by using hyperspectral infrared satellite Atmospheric Infrared Sounder data (AIRS) in China’s Xinjiang region for two typical desert and snow features. Firstly, infrared radiation transmission equation of earth-atmosphere system was linearized. Then, it was proposed that atmospheric profile and surface emissivity can be structured by Empirical Orthogonal Functions (EOF) to effectively reduce the inversion variables. The physical simultaneous retrieval algorithm could be developed finally. In the retrieval process, the first guess values were obtained through National Centers for Environmental Prediction (NCEP), and the optimum solution could be obtained by Newton iteration method. The observation area covered the Taklamakan desert and Junggar basin in Xinjiang, China. The latitude of the Tazhong observation station is 38.98 degrees and the longitude is 83.64 degrees, which is located in the hinterland of the Taklimakan desert in central tarim basin. The latitude of National field science observation station of fukang desert ecosystem is 44.2 degrees and the longitude is 87.9 degrees, which is located in Junggar basin. The Tazhong observation station and National field science observation station of Fukang desert ecosystem were selected to be the retrieval of the ground verification point. These stations were selected to be the retrieval of the ground verification point. The results showed that the surface temperature in the Taklamakan desert is significantly higher than that in Junggar basin, which is consistent with the actual situation. According to the retrieval of the surface emissivity distribution at 8.6 and 13.4 μm, it could be seen that the desert surface emissivity is significantly lower than the emissivity of snow at 8.6 μm, and retrieval of the two kinds of ground infrared emissivity spectrum is consistent with the laboratory measurement of emissivity spectra by comparing the retrieved surface emissivity and the jet propulsion laboratory measurement of desert and snow emissivity data between 6～15 μm. The atmosphere and ground were considered as a whole system, the surface emissivity was added in the retrieval in this paper. Through comparison and analysis of retrieved two kinds of ground atmospheric profile with the local meteorological sounding values and traditional method retrieval, the research showed that the retrieval accuracy of atmospheric temperature and water vapour profile is improved, especially the improvement is obvious in the boundary layer. At the same time, the analysis showed that the improvement of the retrieval precision of the atmospheric profile in the desert region is higher than that in the snow region. Because the surface emissivity changes within the spectrum is larger in the desert area, while the surface emissivity changes within the spectrum is smaller in the snow area. It is important that using the proposed approach can simultaneously retrieve atmospheric temperature profiles, water vapour profiles, surface emissivity and surface skin temperature. The retrieval precision of the atmospheric temperature profile and water vapour profile in the desert area can be improved more effectively compared with snow region. This paper can provide the service and support for the numerical weather forecast and the future hyperspectral infrared satellite application in China, which is of great significance.