Evaluation of CO2 Greenhouse Gas Estimation Algorithms Based on GOSAT Satellite Data and Ground-based Observation Stations

In this report, we compare data products from three different algorithms with the reference data obtained by ground-based high-resolution Fourier Transform Spectrometers (g-b FTSs)  in the Total Carbon Column Observing Network (TCCON), with the 8 selected sites in five years(2011-2015). The algorithms evaluated are NIES, ACOS and SRFP algorithms. These algorithms are focused on retrieving the column abundance of the CO2 to take advantage of the molecular amounts of dry air carbon dioxide (XCO2). To evaluate the products of each algorithm with its equivalent ground observations, statistical indices such as Bias error, root mean square error (RMSE), absolute error (MAE), standard deviation (SD), and Pearson correlation coefficient (CR) were used. By examining the values presented by each algorithm and comparing it with the ground observation values, it can be concluded that the NIES, ACOS, and RemoTeC (SRFP) algorithms have the lowest RMSE, Bias and MAE error respectively. The best agreements with TCCON measurements in the most stations were detected for NIES 02.xx. The SRFP algorithm has a significant difference in estimating CO2 retrieving rates compared to the other two algorithms. So that the lowest correlation values belong to the SRFP algorithm and the highest correlation, values belong to the NIES algorithm.