Accurate assessments of the CO2 fluxes between the terrestrial ecosystems and the atmosphere are pressingly needed for the climate change and carbon cycle studies. The Collatz et al. parameterization of leaf photosynthesis-stomatal conductance has been widely applied in land surface parameterization schemes for simulating the land surface CO2 fluxes. The study in this paper developed an analytical solution approach for the Collatz et al.'s parameterization for stable solution and computational efficiency. This analytical approach is then applied to the simplified biosphere model (SSiB), enhancing its capability of simulating land surface CO2 fluxes. The enhanced SSiB model is tested with field observation data sets from two Amazonian field experiments (ABRACOS missions and Manaus Eddy Covariance Study). Simulations of the land surface fluxes of latent heat, sensible heat and soil heat by the enhanced SSiB agree very well with observations with correlation coefficients being larger than 0.80. However, the correlation coefficient for the daily means Of CO2 fluxes is only 0.42 for the Manaus data set. A day-time, "square wave" in the simulated CO2 flux diurnal curves is found. The discrepancies between simulation and observation were found to be the results of incorrect parameter setup or improper leaf to canopy scaling strategy. A modification to the scaling strategy improves significantly the accuracy of the photosynthesis-stomatal conductance model. (C) 2002 Elsevier Science B.V. All rights reserved.