A study on estimation of bio-physical parameters of individual tree stands derived from LIDAR data and digital matrix camera image

摘要

Among the aerial remote sensing topics in forest area, the new age technologies LiDAR (Light Detection And Ranging) and DMC (Digital Matrix Camera) are now available to monitor and to estimate several biophysical parameters in forest. LiDAR makes it possible to acquire the ground surface information and DMC has high resolution specification. In this study, tree species classification and individual tree stands extraction were conducted by using the simultaneously acquired LiDAR and DMC data. One part of specific planted forest was selected as study site, where the density of tree stands are high and Japanese ceder and Japanese red pine are mixed. Tree species classification was applied by following 3 steps: 1) create true orthophoto image from DMC and DSM (Digital Surface Model) derived from LiDAR. 2) apply unsupervised classification using green band luminance of the orthophoto. 3) for the each part of the classified area, fit the each type of quadratic surfaces model to DSM to calculate the tree top likelihood areas. Then the tree stands were extracted and the optimum quadratic surface shapes were estimated by changing the surfaces at the characterized ranges of each tree species. The results of the extraction were 73 stands per 89 for Japanese ceder, 26 stands per 29 for Japanese red pine, but the miss extraction which means that pines were recognized as ceders was 14 stands. Additionally, upper part of each crown shapes were acquired as optimum quadratic surfaces. These results indicate the efficiency and the probability of this study toward the estimation of biophysical parameters by aerial remote sensing.