COMPARISON OF METHODS FOR INDIVIDUAL TREE CROWN DELINEATION IN TROPICAL FORESTS USING VERY HIGH RESOLUTION SATELLITE MULTISPECTRAL AND AIRBORNE LASER SCANNING DATA

Individual tree crown (ITC) delineation is the first step to study forest biodiversity and carbon using very high resolution remote sensing data. In order to study large areas, automatic methods for ITC delineation are necessary. In this paper, we compared three methods to delineate ITCs over the Jamari National Forest, a closed-canopy tropical forest in Amazonia, using very high resolution satellite multispectral (MS) and airborne laser scanning data (ALS). The best methods for ITC delineation over this tropical forest site were the voronoi-based method for ALS data and the marker-controlled watershed (MCWS) method for MS data. Window sizes of 3x3 (ALS) and 5x5 (MS) m provided the appropriate scale for extracting tree crowns over this forest. The performance of ITC delineation from ALS and MS datasets was similar. Results indicate that tree crown delineations can be retrieved from satellite MS data for forest monitoring.