In the satellite remote sensing of carbon dioxide (CO2), atmospheric environmental factor is the important factor affecting the inversion accuracy. The inversion condition is usually limited to the situation of the aerosol optical thickness less than 0.3. The higher requirement of the atmospheric conditions will seriously affect the application ability of our country’s CO2 satellite remote sensing data. For this kind of situation, based on principal component analysis (PCA), atmospheric CO2 of Beijing and Tianjin areas of China is inversed of high aerosol optical thickness, the CO2 column concentration obtained is compared with the product of GOSAT-Level2 in 2013, 2014. The root mean square error is 0.65% and 0.46%, respectively, and the correlation is 0.77 and 0.93, respectively.