| According to the fifth assessment report by the Intergovernmental Panel on Climate Change in2013, CO2is considered to be the most important greenhouse gas since the21st century. Due to the limited observations of CO2, there are still a lot of uncertainty about the quantitative description of its emission, spatial distribution, and its feedback effects on the climate system. The traditional supplier of atmospheric CO2measurements is ground-based observation network which is too sparse to supply data with the coverage or resolution to map carbon sources and sinks. While space-based observations can address these shortcomings.In this paper, an eigenvector regression algorithm was adopted to retrieve CO2concentration from cross-track infrared Sounder CrIS. The main contributions of this dissertation contain:According to the multi-channel characteristics of CrIS sensor, we adopted information entropy iterative method to choose optimum channels. First, AIRS channels near15μm, which includes strong absorption lines of CO2, was found based on the atmospheric absorption spectrum. Then43channels were finally selected by using information entropy iteration method around the15μm. We analyzed the sensitivity of parameters of atmosphere and surface to the selected channels.By selecting different feature vectors and simulation training samples eigenvectors radiation values, statistical regression coefficients were investigated. Comparing the inversion results with the real value, the best number of feature vectors can be obtained.Distribution of monthly, seasonal average of CO2concentration retrieved from the CrIS data for the year of2013in China. And we analyzed the mean change of CO2and its spatial distribution. The results showed that the distribution of CO2emissions presents higher in north and lower in the south, changing with latitude. Judging from the year-round situation, the highest column concentration value of CO2appears in summer in China.The eigenvector regression algorithm was applied to Cross-track Infrared Sounder (CrIS) observations, validated by ground-based World Data Centre for Greenhouse Gases (WDCGG) observations and AIRS CO2products. Compared with WDCGG measurements, it showed a very well correlation:the correlation coefficient was0.5371, and the RMS relative errors were smaller than1.5%; compared with AIRS CO2product, the coefficient of determination for CO2concentration from eigenvector regression model was also well. |
