Synergy of observations from various satellites for the fast retrieval of atmospheric carbon dioxide amounts

We present an algorithm for the rapid retrieval of the carbon dioxide total column amounts (XCO2) using short wave infrared (SWIR) spectra of the reflected sunlight measured from space. The algorithm takes advantage of the combined processing of observational data from two different satellite missions. For the algorithm implementation we adopted the previously developed EOF (Empirical Orthogonal Functions)-based approach that exploits regression relations of the principal components of the measured spectra with target XCO2 values. In the original algorithm version the regression coefficients were derived by using training sets of collocated satellite and ground-based observations (ground-based observations were treated as “true values”). In this paper we implemented similar approach in which training set for one satellite mission is created using collocated observations of the another “reference” space mission simultaneously on-orbit (in this case XCO2 retrievals of the “reference” mission were treated as “true values”). This approach enables rapid data processing of the new satellite missions omitting expensive and time consuming stage of retrieval algorithm development. The feasibility of the approach was tested by joint processing of GOSAT and OCO-2 observation data. For the analysis of the algorithm precision/accuracy characteristics we used the collocated observations from the Total Carbon Column Observing Network (TCCON).