Simulator (BEPS) and light response curve were combined to separate the eddy covariance GPP data into sunlit GPP and shaded GPP, due to the lack of measured sunlit GPP. The reliability of Vcmax was validated through the correlation analysis between the normalized difference vegetation index (NDVI) and Vcmax. To illustrate the advantages of the proposed scheme in our study, BEPS simulated GPP using the derived Vcmax from sunlit GPP in our study, the derived Vcmax from canopy GPP and the reference Vcmax as constants were respectively compared with corresponding eddy covariance GPP measurements. Results showed that (1) the proposed assimilation scheme based on sunlit GPP can be used to derive reliable temporally continuous Vcmax. The Vcmax obtained in our study was significantly correlated with NDVI (R2 = 0.77); (2) the seasonal variations in Vcmax could significantly improve the accuracy of photosynthesis simulation. The GPP difference between eddy covariance GPP and BEPS simulated GPP using derived Vcmax from sunlit GPP (mean = 0.12, std = 0.92) were significantly smaller than those between eddy covariance GPP and BEPS simulated GPP using reference Vcmax as constants (mean = 1.37, std = 1.37); (3) the sunlit GPP is able to obtain the relatively accurate Vcmax that can improve the simulation of photosynthesis. GPP difference between eddy covariance GPP and BEPS simulated GPP using derived Vcmax from sunlit GPP (mean = 0.12, std = 0.92) showed lower mean values and standard deviations than those between eddy covariance GPP and BEPS simulated GPP using derived Vcmax from canopy GPP (mean = 0.67, std = 1.00). Our study illustrated the reliability that sunlit GPP can be used to derive relatively accurate temporally continuous Vcmax, and the assimilation scheme gives new potential to improve the simulation of photosynthesis through terrestrial biosphere models.