[H-TT16] Geographic Information Systems and Cartography

[HTT16-02] Combining aerial and terrestrial LiDAR on the Unmanned Aerial System (UAS) and the Backpack system for estimating forest attributes with a full 3D scan of the forest environment

*Kotaro Iizuka^1,2, Takashi Matsubara³, Masayuki Itoh^4,5, Takashi Oguchi¹ (1.Center for Spatial Information Science, University of Tokyo, 2.Research Institute for Humanity and Nature, 3.Technical Research Institute, Obayashi Corporation, 4.School of Human Science and Environment, University of Hyogo, 5.Center for Southeast Asian Studies, Kyoto University)

Keywords:UAV, LiDAR, Forest, Remote Sensing, Biophysical Parameter

Observing the characteristics of forests and estimating their resources have been essential tasks for forest ecologists and forest managers. Despite the advancements in remote sensing technologies and their various approaches, a robust methodology is still under development. Small, weighted unmanned aerial systems (UASs) are increasing interest to researchers not limited to forestry, and the development of the UAS platform within this decade has enabled onboard sensors such as LiDAR (Light Detection and Ranging). The advantage of LiDAR in investigating the forest environment is that it can reconstruct the 3D structure of forests with high precision, which is much more robust than the Structure from Motion (SfM) approach that is conventionally focused when utilizing UAS’s optical sensors. Therefore, works utilizing UASs with LiDAR systems are increasing; however, there are limitations in low-cost LiDAR systems to sense the full view of the dense forest environment. One issue is that the dense and mature forest environment, especially in Japan, limits a LiDAR system to sense sufficient returning-point information in the understory and forest floor. An entire forest scan (full 3D scan) needs to be performed to estimate an accurate forest structure.
This work focuses on reconstructing a full 3D model of forests by complementing the UAS LiDAR-based 3D model with the Backpack LiDAR system. The study site is a private plantation site for coniferous trees in Hannou City, Saitama Prefecture, Japan. The elevation of the site is approximately 260-330 m m.s.l. with extremely rugged topography that limits activities on the ground. The LiAir-50 LiDAR system is mounted on a Matrice 600 UAS, and the point cloud information is obtained from the air, while the LiBackpack D50 system is utilized to collect the lacking-point information in the lower layer of the forests from the ground. The two obtained point clouds are merged to complete a full 3D model of the forests and the automated extraction of forest attributes. The forest resources are estimated by plotting a 10 by 10 m square grid, and the total volume of the trees was computed for each unit area. Further discussions will be made on the advantages of the proposed methodology, its limitations, and future challenges.