Study On Inversion Of Land Surface CO₂ Fluxes Based On The GEOS-Chem Model And A Data Assimilation Method

Atmospheric carbon dioxide（CO₂） inversion is an important way to improve estimated bioshperic carbon fluxes.This method was designed to continuous adjustement and optimization of priori CO₂ fluxes by using the Bayesian synthesis technique to minimize the differences between the observed and simulated CO₂ concentrations. As one of the most fashionable three-dimensional global chemistry transport model, GEOS-Chem has a broad application prospects in the field of modern top-down flux inversion.We developed a new data assimilation system jointly assimilating satellite and ground CO₂ concentration observations to estimate surface CO₂ fluxes over 218 geographical regions, using GEOS-Chem as forward model and referred to the framework of Carbon Tracker-China.We alse designed five experiments to analyse the key parameters’ s sensitiveness and verified the carbon flux inversion system with observations. Main research work and achievements are summarized as follows:（1）We evaluated the CO₂ concentration simulated by the posterior CO₂ fluxes from the Carbon Tracker-China.The results indicate that model driven by these fluxes both can accurately simulate spatial and temporal patterns of atmospheric CO₂ concentration, showing the average value of increasement CO₂ concentration was 2.081 ppm in global surface. However, there was a distinct seasonal difference between using these two CO₂ fluxes data sets. The outcome shows that model simulated CO₂ concentration driven by CT-China’s posterior CO₂ fluxs is better than thst driven by GEOS-Chem’s CO₂ fluxes in summer, while underestimated in winter and spring. Overall, GEOS-Chem can reasonably simulate global CO₂ concentration driven by Carbon Tracker-China’s posterior CO₂ fluxs.（2）Based on the research work above,we built a new carbon flux inversion system, then we discussed the method, key parameters and structure of the carbon flux inversion system.The results of key parameters’ sensitivity experiments show that the sensitivity of the geographical regions and observation values are relatively higher than that of the number of ensemble members and the assimilation window size. Among observation data, we found that GOSAT-ACOS3.3 had a great impact on the assimilation results, the uncertainty of posterior CO₂ fluxes estimated by ground-based atmospheric CO₂ observations is bigger due to sparse available local observations.（3）We also evaluated the performance of this new assimilation system by comparing with CO₂ concentration observations in various spatio-temporal scales. The results show that the carbon flux inversion system has high assimilation precision for both CO₂ fluxes and CO₂ concentrations（-0.147±2.428ppm）.The estimated global net ecosystem exchange is-3.893 Pg C?yr-1 from June-2009 to May-2010, which is very closed to Carbon Tracker-China’s posterior net ecosystem exchange（-4.397 Pg C?yr-1）. Net ecosystem exchange is-3.167 Pg C?yr-1 and the bioshpere carbon flux is-1.910 Pg C?yr-1 in the Northern Hemisphere from June-2009 to May-2010. Uncertainty analyses show that the system can reduce the uncertainty of posterior CO₂ fluxes signigicantly in European and North American temperate region with sufficient and high precise observational data, while reduce the uncertainty relative slightly in South American temperate regions. Moreover, the spatial and temporal distribution patterns of the bioshpere carbon fluxes in the Northern Hemisphere are consistent with the estimated gross primary production, except for Southeast Asia and Russia.