Estimating the scalar source/sink distribution of CO2 and its vertical fluxes within and above forest canopy continues to be a critical research problem in biosphere-atmosphere exchange processes and plant ecology. With broad-leaved Korean pine forest in Changbai Mountains as test object, and based on Raupach's localized near field theory, the source/sink and vertical flux distribution of CO2 within and above forest canopy were modeled through an inverse Lagrangian dispersion analysis. This model correctly predicted a strong positive CO2 source strength in the deeper layers of the canopy due to soil-plant respiration, and a strong CO2 sink in the upper layers of the canopy due to the assimilation by sunlit foliage. The foliage in the top layer of canopy changed from a CO2 source in the morning to a CO2 sink in the afternoon, while the soil constituted a strong CO2 source all the day. The simulation results accorded well with the eddy covariance CO2 flux measurements within and above the canopy, and the average precision was 89%. The CO2 exchange predicted by the analysis was averagely 15% higher than that of the eddy correlation, but exhibited identical temporal trend. Atmospheric stability remarkably affected the CO2 exchange between forest canopy and atmosphere.