CO2 is a primary greenhouse gas in the atmosphere and many scientific missions from all over the world are focused on space-based remote sensing of CO2 using shortwave infrared technique. We present an improved method to retrieve CO2 column abundance from space-based observations in a shortwave infrared band. We analyze the sensitivities of CO2 observations to aerosol scattering and atmospheric temperature profiling using the forward model. We also evaluate the reduced accuracy of CO2 retrieval caused by aerosol scattering and 1K random error of temperature profiles. We propose a Modified Damped Newton Method（ MDNM） for CO2 retrieval and present the retrieval samples to evaluate the performance of the proposed algorithm. Sensitivity studies show that the influence of aerosol scattering and temperature on CO2 remote sensing and scattering can cause a misestimate of up to 13. 2 ppm of CO2 concentration. We retrieve the CO2 column abundance from GOSAT L1 B data using MDNM method

and the results are compared with ground-based measurements from Total Carbon Column Observing Network（ TCCON）. The correlation coefficient R-square between the retrieved results and TCCON measurements is 0. 729. Sensitivity studies show that aerosol scattering can cause a significant error in CO2 retrieval. A comparison between satellite retrievals and ground-based measurements shows good agreement. The proposed MDNM method is preliminarily proven to accurate.