Assimilation of data from several measurements provides knowledge of the model's performance and uncertainties. In this work we investigate the complementary of Biomass data to net CO2 flux (NEE) and latent heat flux (LE) in optimising parameters of the biogeochemical model ORCHIDEE. Our optimisation method is a gradient based iterative method. We optimized the model at the French forest sites, European beech forest of Hesse (48 .67°N, 7.06°E) and maritime pine forest of Le Bray (44.72°N, 0.77°W). First we adapted the model to represent the past clearcut on these two sites in order to obtain a realistic age of the forest. The model-data improvement in terms of aboveground biomass will be discussed. We then used FluxNet and Biomass data, separately and altogether, in the optimization process to assess the potential and the complementarities of these two data stream. For biomass data optimization we added parameters linked to allocation to the optimization scheme. The results show a decrease in the uncertainty of the parameters after optimization and reveal some structural deficiencies in the model. In a second step, data from ecosystem manipulation experiment site Brandbjerg (55.88°N, 11.97°E), a Danish grassland site, were used for model optimisation. The different ecosystem experiments at this site include rain exclusion, warming, and increased CO2 concentration, and only biomass data were available and used in the optimization for the different treatments. We investigate the ability of the model to represent the biomass differences between manipulative experiments with a given set of parameters and highlight model deficiencies.