CO2 is one of the most important greenhouse gases due to its selective absorption of long wave radiation from the Earth’s surface. In this paper, we use the column average dry air mole fraction of CO2 (XCO2) data from the Japanese GOSAT satellite to conduct a comprehensive and systematic analysis of temporal and spatial distribution of XCO2. This includes: (1) analysis of seasonal change characteristics of XCO2 data; and (2) comparative analysis of the northern and southern hemispheres carbon dioxide concentration at different latitudes. The results show that (1) from 2010 to 2013, atmospheric XCO2 significantly increased each year. The southern hemisphere's annual averages of XCO2 from 2010 to 2012 were 385.2 ppm, 387.3 ppm, and 389.1 ppm, while the average annual values for the northern hemisphere from 2010 to 2012 were 387.8 ppm, 390.0 ppm, and 391.7 ppm. The annual XCO2 in northern and southern hemispheres exhibited growth rates of 1-2 ppm per year. (2) The results show seasonal change trends: winter months displayed higher XCO2. Regarding the global spatial distribution of XCO2, the results show that the total XCO2 in the northern hemisphere is higher than that in the southern hemisphere. (3) The growth of global XCO2 in 2011 and 2012 was 1.9 ppm/yr and 2.1 ppm/yr. These values are in accordance with the growth rates of 1.9 ppm/yr and 2.2 ppm/yr reported in the World Meteorological Organization's greenhouse gas bulletin.
Satellite observations and model simulations are of two important data sources to study atmospheric carbon dioxide concentration. For analyzing and evaluating the bias correction method of ACOS dry-air column averaged CO2 (Xco2) product, the GEOS-Chem Xco2 simulations are selected according to observing time and locations of the ACOS product. The GEOS-Chem simulations of CO2 profiles are transformed to Xco2 data by convolving with satellite averaging kernels and pressure weighting functions. The GEOS-Chem Xco2 data are then compared with both bias uncorrected and bias corrected satellite retrievals of ACOS. The comparisons show that the bias uncorrected ACOS retrievals are on average 1.12ppm higher than the model Xco2 data, while the corrected ACOS retrievals are only on average 0.06ppm lower than the model Xco2 data. By assuming consistency between model Xco2 simulations and true atmospheric Xco2, this study indicates that the bias can be obvious decreased through the bias correction method, and the correction is effective and necessary for satellite Xco2 retrievals.