Spatiotemporal variation analysis of global XCO 2 concentration during 2010-2020 based on DINEOF-BME framework and wavelet function.

Combining with Carbon dioxide column concentration (XCO2) remote sensing data, it is of great scientific significance to obtain XCO2 long time series data with high precision and high spatio-temporal coverage. In this study, the combination framework of DINEOF and BME were employed to integrate the XCO2 data of GOSAT, OCO-2 and OCO-3 satellites for generating global XCO2 data from January 2010 to December 2020, with the average monthly space coverage rate of more than 96 %. Through cross-validation and comparison of The Total Carbon Column Observing Network (TCCON) XCO2 data and DINEOF-BME interpolation XCO2 products, it is verified that DINEOF-BME method has better interpolation accuracy, and the coefficient of determination of interpolated XCO2 products and TCCON data is 0.920. The long time series of global XCO2 products showed a wave rising trend, with a total increase of ~23 ppm; obviously seasonal characteristics were also detected with the highest XCO2 value in spring and the lowest in autumn. According to the zonal integration analysis, the values of XCO2 in the northern hemisphere is higher than the southern hemisphere during January-May and October-December, while the values of XCO2 in the southern hemisphere is higher than the northern hemisphere during June-September, which accords with the seasonal law. Through EOF mapping, the first mode accounted for 88.93 % of the total variability, and its variation trend is consistent with that of XCO2 concentration, which verifies the variation rule of XCO2 from the time and space pattern. Through wavelet analysis, the time scale corresponding to the first main cycle of XCO2 change is 59-month, which has obvious regularity on the time scale. DINEOF-BME technology framework has good generality, while XCO2 long time series data products and the spatio-temporal variation of XCO2 revealed by the research provide a solid theoretical basis and data support for related research.