Research On Multi-source Atmospheric CO₂ Data Fusio

CO₂is one of the most important greenhouse gases in the atmosphere.Since the industrial revolution,the concentration of CO₂in the atmosphere has been increasing,which has an important impact on global climate change.At present,the detection of atmospheric CO₂concentration has always been a hot topic of research at home and abroad.Satellite remote sensing has the advantages of wide observation range,but it can not be effectively observed in high latitude and other areas.At the same time,the atmospheric assimilation model can obtain the global CO₂concentration data,but it is excessively dependent on the ground-based site data.At present,there is no CO₂product that meets the high coverage and high spatial and temporal resolution required for carbon neutrality and global CO₂change research.Therefore,in this study,the two greenhouse gas satellites GOSAT and OCO-2currently in orbit and the Carbon Tracker atmospheric assimilation model are selected for fusion.The main conclusions of this paper are as follows:（1）Compared with the spatiotemporal differences among GOSAT,OCO-2 and Carbon Tracker,the XCO₂retrieved by them has the same change trend,and XCO₂has a relatively stable periodicity with the change of latitude.However,OCO-2 XCO₂and CT XCO₂values are greater than GOSAT XCO₂values.The world is divided into two parts:sea and land,and the difference of the three products XCO₂in the sea is smaller than that on the land.（2）In view of the lack of satellite data,taking into account the spatiotemporal variation characteristics of CO₂concentrations at different latitudes,the world is divided into six regions,and the regions without data are filled by Kriging interpolation.After Kriging interpolation,at 1°×1°grid,the average coverage of OCO-2 is 71.66%,the average coverage of GOSAT is 52.45%,and the coverage of Carbon Tracker is 100%.The coverage of OCO-2and GOSAT has increased by 53.65%and 48.5%respectively.（3）The uncertainty of the three products is evaluated by using the Triple Collocation algorithm,and the random errors of Carbon Tracker,OCO-2 and GOSAT are 0.278 ppm,0.403 ppm and 0.639 ppm respectively.The weights of the three products are determined as0.402,0.333 and 0.265 by the random errors obtained.The XCO₂data of the three products are fused according to the weights,and 1°×1°half-monthly average XCO₂fusion data set.Use TCCON site data to verify that the MAE of the fusion dataset and TCCON site data is0.627ppm,RMSE is 0.768ppm,and R²is 0.828.（4）According to the analysis of the spatiotemporal distribution characteristics of global XCO₂,XCO₂in the Northern Hemisphere has an obvious seasonal change cycle,and its seasonal change range is gradually increasing with the increase of latitude.The changes in the southern hemisphere are relatively gentle,and the seasonal variation amplitude of XCO₂is the most obvious in the region of 20°N-50°N north latitude.Compared with the northern hemisphere,the carbon dioxide concentration in the southern hemisphere is evenly distributed.The concentration of CO₂in the Northern Hemisphere is the highest in spring,followed by autumn and winter,and XCO₂in the Northern Hemisphere is lower in summer.