航空機 LiDAR を用いた都市空間における樹木の形状情報の抽出精度

抄録

We examined the acquisition accuracy of high-resolution airborne LiDAR for capturing the external forms of individual trees in an urban district. The accuracy of the voxel-based reconstruction of the crown form was verified in anticipation of estimating the leaf area density (LAD) distribution. We calculated the decrease in the number of airborne laser beams incident on a crown in each height bin for 20 Japanese zelkova trees, and then, determined the target tree. Terrestrial LiDAR data of the target tree was used for the verification. We examined the spatial accuracy of the airborne LiDAR points on the crown surface before verifying the accuracy of the voxel-based reconstruction. The results confirmed that airborne LiDAR captured the crown periphery with the spatial accuracy of 0.5 m and 1 m in the upper and lower parts of the crown, respectively. Next, voxels including one or more points were extracted as the areas in which the foliage is distributed. We then verified the distribution accuracy of the extracted voxels, and we found that 60 % of the external voxels derived by the terrestrial LiDAR can be extracted. We also investigated the reasons for voxels not being extracted by the airborne LiDAR, using the LAD in voxels derived from the terrestrial LiDAR data. We observed that 96 % of the external voxels derived by the terrestrial LiDAR was comprised of voxels extracted by airborne LiDAR (the adjacent voxels of which can be extracted) and voxels that had insufficient foliage to trigger a return pulse. Our results quantitatively indicate that high-resolution airborne LiDAR is applicable to the voxel-based reconstruction of external forms of individual trees in urban spaces.