The Fast CO2 Retrieval Using Semi-physical Statistical Model for the High-resolution Spectrometer on FengYun-3 Satellite

China’s FengYun-3D launched in December 2016 carries the high-resolution Greenhouse gases Absorption Spectrometer (GAS) aimed at providing global observations of carbon dioxide. GAS is one of the few instruments to measure CO2 from the near infrared spectrum. On orbit, oxygen A band suffers the disturbance and the signal to noise ratio is lower than nominal specification. This leads to difficulties of retrieval of surface pressure, furthermore, retrieval of XCO2 if full physics retrieval algorithm is used. To overcome this deficiency, a fast CO2 inverse method, named semi-physical statistical algorithm, has been developed. The instrument characteristics, the semi-physical statistical algorithm and the results of comparison with ground-based measurements over land have been introduced in this paper. XCO2 can be obtained from the three bands, i.e., oxygen A band, weak CO2 band and strong CO2 band with the compensating from the MERSI-2 products, the ERA-5 re-analysis data and the TCCON data. The techniques of eigenvectors of covariance matrices and the least square fits are used to derive retrieval coefficients and yield cloud-free solutions. In addition to GAS radiance, some key factors learned from the full physics retrieval algorithm are also taken as input information, for instance, air mass, surface pressure and a priori XCO2 that are necessary for accurate estimations of XCO2. Filters are also essential to exclude the poor observation spectra. The global GAS XCO2 restricted over land are compared against the simultaneously, collocated observations from TCCON. We find that this retrieval algorithm can solve the problem of the oxygen A band to some extent. Overall, by the site-by-site comparison, GAS XCO2 shows good agreements with average precision (1σ) of 1.52 ppm and biases of -0.007 ppm. The variations trends of GAS XCO2 caused by season changes can be clearly seen at TCCON sites in the time scale of a year