Abstract
In French Guiana, forest areas managed for wood production represents 2.5 million hectare. Accurate and high-resolution data on resource quality and quantity are essential for planning and optimizing forest management operations. The Dendrolidar project (http://amap.cirad.fr/fr/edite_projet.php?projet_id=92) aims precisely to develop tools for predicting and geolocalizing wood resource using airborne laser scanning (ALS) coupled with THR photographies. Terrestrial lidar acquisitions (TLS) will provide a reference database for calibrating species-specific allometric models derived from airborne data. This poster outlines the main methodological issues when scanning and modelling woody structure using TLS and ALS in tropical dense forest. Wood volume is assessed from co-registered TLS points clouds after the critical steps of individual tree segmentation, wood-leaf points classification and tree reconstruction. Then these TLS-derived tree volume data are used to build species-specific models predicting tree volume from morphometric crown attributes extracted from ALS. This approach is applied to three major commercial timber species of French Guiana forest (Dicorynia guianensis, Qualea rosea, Eperua falcata).