In current Environmental Impact Assessment (EIA) practices, the estimation of damaged tree volume primarily relies on quadrat-based sampling methods. However, such approaches have inherent limitations in terms of accessibility, safety, and representativeness, particularly in steep slopes and densely vegetated forest areas. This study aims to quantitatively estimate damaged tree volume using LiDAR point cloud data and the Cloth Simulation Filter (CSF) algorithm, and to propose an analytical workflow applicable to EIA practices.
The study area is a forest site subject to small-scale EIA located in Yangji-ri 527, Onam-eup, Namyangju-si, Gyeonggi-do, South Korea. High-density point cloud data were acquired using the SATLAB SL9 system. The workflow consisted of point cloud preprocessing, CSF-based ground filtering, noise removal of non-ground points using Statistical Outlier Removal (SOR), height normalization, individual tree segmentation using Treeiso, diameter at breast height (DBH) estimation via circle fitting, and volume calculation using both 2.5D volume and allometric equations.
The results show that the total number of points in the study area was 291,003,897. After CSF application, 170,879,807 points (58.72%) were classified as ground points and 120,124,090 points (41.28%) as non-ground points. Following SOR filtering, 105,683,270 points remained. After height normalization and removal of points below 0.3 m, 104,848,559 points were used for Treeiso segmentation. A total of 2,397 individual tree objects were extracted, which were further classified into 600 accepted trees, 1,565 multi-stem or merged candidates, and 232 removed objects. Among the accepted trees, 207 were classified as overstory trees, 182 as midstory trees, and 211 as understory vegetation. The total estimated tree volume was 174.0263 m³. These findings demonstrate that LiDARbased analysis enables precise quantification of individual tree attributes, including location, height, DBH, and volume, even in complex forest environments such as steep terrains. Compared to conventional sampling-based approaches, this method provides detailed three-dimensional structural information and volume-based estimation of damaged trees. Therefore, it is expected to contribute to the development of a more objective and reproducible framework for estimating damaged tree volume in EIA practices.
