Global Carbon Dioxide Flux Inferred From Atmospheric Observations and Its Response to Climate Variabilities

Atmospheric inversion has recently become an important tool in estimating CO2 sinks and sources albeit that the existing inversion results are often uncertain and differ considerably in terms of the spatiotemporal variations of the inverted carbon flux. More measurements combined with terrestrial ecosystem information are expected to improve the estimates of global surface carbon fluxes which are used to understand the relationships between variabilities of the terrestrial carbon cycle and anomalies of climatic factors.;Forest carbon dynamics is closely related to stand age. This useful terrestrial ecosystem information has been used as an additional constraint to the atmospheric inversion. The inverse estimates with this constraint over North America exhibit an improved correlation with carbon sink estimates derived from eddy-covariance measurements and remotely-sensed data, indicating that the use of age information can improve the accuracy of atmospheric inversions.;Terrestrial carbon uptake is found mainly in northern land, and a strong flux density is revealed in southeastern North America in an improved multi-year inversion from 2002 to 2007. The global interannual variability of the flux is dominated by terrestrial ecosystems. The interannual variabilities of regional terrestrial carbon cycles could be mostly explained by monthly anomalies of climatic conditions or short-time extreme meteorological events. Monthly anomalies of the inverted fluxes have been further analyzed against the monthly anomalies of temperature and precipitation to quantitatively assess the responses of the global terrestrial carbon cycle to climatic variabilities and to determine the dominant mechanisms controlling the variations of terrestrial carbon exchange.;Inversions using more observations have often been hampered by the intense diurnal variations of CO2 concentrations at continental sites. Diurnal variations of the surface flux are included with atmospheric boundary dynamics in order to improve the atmospheric inversion accuracy. Modeling experiments conducted in this study show that inverse estimates of the carbon flux are more sensitive to the variation of the atmospheric boundary layer dynamics than to the diurnal variation in the surface flux. It is however generally better to consider both diurnal variations in the inversion than to consider only either of them.