It was commonly believed that Carbon Dioxide (CO2) is the important greenhouse gas, which is connected with the global climate change. Scientists have researched that the total amount of CO2 in the atmosphere had improved 30% since the industry age, and it is being increasing now. But there are not enough in-situ observation stations for monitoring CO2, which could be used to study the temporal-spatial variable laws of CO2. Remote sensing technique has been an important tool to monitoring the CO2 of the atmosphere, which is based on the CO2 absorption features in many wavelengths of electronic-magnetic spectrum of the sunlight radiation. The absorption wavelengths are relatively stable, that can be used in the wavelength calibration for hyper-spectra remote sensing sensors on-orbit. Curves of Radiation Spectra varied with different amount of CO2 were calculated with MODTRAN, and the depth of the absorption trough calculated by the difference value of the radiation at 2036nm and 2005nm were calculated, there are good correlations between the depth and the amount of CO2 in the atmosphere. Hyperion hyper-spectral data also was used to further study the impacts on the CO2 of different land use types. The preliminary study showed that the vegetation was good to reduce the amount of CO2 in the near surface. This study could be used in the further study on the carbon emission of human activity.