Abstract:A leaf-scale combination model has been successfully scaled up to canopy level and coupled with single source model (SSM) to estimate canopy-scale latent heat flux (λE) of winter wheat based on one year data. It is necessary to study the applicability of the combination model at longer time scale and for other ecosystems. In this study the combination model was used as the stomatal conductance sub-model in Shuttleworth-Wallace (S-W) model, which is a widely used dual source model (DSM). We applied the S-W model to estimate diurnal patterns of λE over a temperate mixed forest at Changbaishan (CBS) in Northeast China. Measured λE data from the eddy covariance (EC) system were used for model validation. Diurnal variation of estimated λE is in good agreement with that of measured λE. Simulated results of SSM and S-W model were compared and the S-W model exhibits better simulation than SSM. The slope and R2 of the line regression between the S-W model simulated and observed λE are 0.96 and 0.72 (n=8519), respectively. The results show that dominant factors controlling the variation in ecosystem evapotranspiration (ET) and vegetation transpiration (Tr) at CBS are different at seasonal and inter-annual time scales. Vapor press deficit (D) and photosynthetically active radiation (Qp) are the dominant factors for the seasonal variation of ET and Tr, while the differences in D and temperature determine their inter-annual variation.