Predicting tree heights for biomass estimates in tropical forests

The recent development of REDD+ mechanisms require reliable estimation of carbon stocks, especially in tropical forests that are particularly threatened by global changes. Even if tree height is a crucial variable to compute the above-ground forest biomass, tree heights are rarely measured in large-scale forest census because it requires consequent extra-effort. Tree height have thus to be predicted thanks to height models. Height and diameter of all trees above 10 cm of diameter were measured in thirty-three half-ha plots and nine one-ha plots throughout the northern French Guiana, an area with substantial climate and environmental gradients. We compared four different model shapes and found that the Michaelis-Menten shape was the most appropriate for the tree biomass prediction. Model parameters values were significantly different from one forest plot to another and neglecting these differences would lead to large errors in biomass estimates. Variables from the forest stand structure explained a sufficient part of the plot-to-plot variations of the height model parameters to affect the AGB predictions. In the forest stands dominated by small trees, the trees were found to have rapid height growth for small diameters. In forest stands dominated by larger trees, the trees were found to have the greatest heights for large diameters. The above-ground biomass estimation uncertainty of the forest plots was reduced by the use of the forest structure-based height model. It demonstrates the feasibility and the importance of height modeling in tropical forest for carbon mapping. Tree height is definitely an important variable for AGB estimations. When the tree heights are not measured in an inventory, they can be predicted with a height-diameter model. This model can account for plot-to plot variations in height-diameter relationship thank to variables describing the plots. The variables describing the stand structure of the plots are efficient for this. We found that variables describing the plot environment (rainfall, topography,...) do not improve the model much. (Resume d'auteur)