The method of using ground-based hyper-spectral measurements of the solar beam to improve satellite detection ability of CO2 concentration is a new trend from the past years. However, the understanding about satellite remote sensing and ground-based verification is still limited. Based on the spectrometers from Chinese carbon satellite ground observation network, two algorithms for remote sensing of CO2 from ground-based hyper-spectral measurements of the direct solar beam in shortwave infrared (SWIR) band are developed, one as DOAS (Differential Optical Absorption Spectroscopy) method and the other as optimum estimation method, with the algorithms both applied to observation spectral. Moreover, the spatial and temporal variations of XCO2 from SCIAMACHY and GOSAT are presented. The main results are summarized as following:
1. Based on LBLRTM and DISORT, a complete forward model for remote sensing of CO2 is established according to index parameters of ultra-high resolution spectrometers OSA and FTS125M.
2. Based on the forward model, sensitivity tests for aerosol, surface pressure, temperature, spectral resolution, spectral offset, signal noise ratio (SNR), solar spectrum and so on, are analyzed. The results show that: The higher the spectral resolution, the larger the effect of spectral offset, the inversion errors caused by 0.005 cm-1 spectral offset are respectively about -0.5 and -1.5 ppm for spectral resolution of 0.2 and 0.02 cm-1. The signal noise ratio (SNR) is low if the spectral resolution is too high, while the factors interfering CO2 inversion are not easy to separate if the spectral resolution is too low. The effect of solar spectrum lies in Fraunhofer lines and that of water vapor lies in water vapor absorption lines, thus the inversion errors caused by solar spectrum and water vapor can be eliminated by channel selection. And the inversion errors caused by surface albedo and aerosol are less than 0.1 ppm. The instrument line shape
(ILS) has significant influence on inversion, so accurate ILS is necessary. The inversion error caused by 1 hPa surface pressure is about 0.25 ppm, therefore the error of surface pressure in the inversion should be less than 1 hPa. The simulated spectrum errors caused by 1 K temperature profile are larger than that caused by 1 hPa surface pressure, but the impact of temperature can be decreased by channel selection. The inversion error caused by CO...