如果卫星遥感反演大气CO2 精度要求达到1%,则可以通过全球连续高精度观测弥补地面观测的不足。在实际反 演中,反演精度不仅受到卷云和气溶胶的影响,还受到其它诸多因素(地表温度,地表压力,地表反照率,大气 湿度等)的影响。讨论分析了这些因子的影响:在以 US1976美国标准大气为背景、观测高度是100 km条件下,运 用最优非线性反演算法,使用逐线积分辐射传输模型LBLRTM模拟计算,模拟分析出地表温度、地表压力、地表 反射率、大气湿度各增加1%和减小1%的情况下,对CO2 反演结果的影响,以及各典型气溶胶对反演 结果的影响。结果显示:海洋型、沙漠型、城市型、乡村型四种典型气溶胶对CO2 反演结果的影响 都比较大,影响最大的为海洋型气溶胶,能见度为5 km,影响百分比相对于CO2 初始值(380 ppm)降 低了27.18%,影响最小的为乡村型气溶胶,能见度为23 km,影响百分比相对于CO2 初始 值(380 ppm)减小了7.78%;地表反射率、地表温度、地表压力对CO2 反演结果也有很大的影响,而大 气湿度对反演结果的影响很小,基本可忽略不计。

If retrieval accuracy for CO2 using the satellite remote sensing reached 1%, the deficiency of ground observation can be made up through the global continuous high precision observation. In the retrieval algorithm, the retrieval accuracy is not only affected by the cirrus clouds and aerosols, but also by effects of other factors (surface temperature, surface pressure, surface albedo and atmospheric humidity). The influence on the retrieval results for CO2 , as well as the influence of typical aerosols on the retrieval results were analyzed when surface temperature, surface pressure, surface albedo, atmospheric humidity respectively increased by 1% and reduce by 1%, under the background of US1976 America Standard Atmosphere and the observation height of 100 km, based on the optimum nonlinear retrieval algorithm, using a line by line integral radiation transfer model (LBLRTM) simulation. The results showed that effects of the four kinds of typical aerosol (marine, desert, urban, rural) on the retrieval results are relatively large. The maximal impact corresponded to the marine aerosol, if the visibility is 5 km, the influence percentage relative to the initial value (380 ppm) decreased by 27.18%. The minimal impact corresponded to rural aerosol, if the visibility is 23 km, the influence percentage relative to the initial value (380 ppm) decreased by 7.78%. The surface reflectance, surface temperature, and surface pressure have great influences on the retrieval results for CO2 , but have little influence of atmospheric humidity on the retrieval results, which can be ignored.