Publications‎ > ‎international‎ > ‎

Estimation of effective plant area index for South Korean forests using LiDAR system

Estimation of effective plant area index for South Korean forests using LiDAR system
KWAK Doo-Ahn1, LEE Woo-Kyun1*, KAFATOS Menas2, SON Yowhan1,
CHO Hyun-Kook3 & LEE Seung-Ho3
Abstract: Light Detection and Ranging (LiDAR) systems can be used to estimate both vertical and horizontal forest structure. Woody components, the leaves of trees and the understory can be described with high precision, using geo-registered 3D-points. Based on this concept, the Effective Plant Area Indices (PAIe) for areas of Korean Pine (Pinus koraiensis), Japanese Larch (Larix leptolepis) and Oak (Quercus spp.) were estimated by calculating the ratio of intercepted and incident LIDAR laser rays for the canopies of the three forest types. Initially, the canopy gap fraction (GLiDAR) was generated by extracting the LiDAR data reflected from the canopy surface, or inner canopy area, using k-means statistics. The LiDAR-derived PAIe was then estimated by using GLIDAR with the Beer-Lambert law. A comparison of the LiDAR-derived and field-derived PAIe revealed the coefficients of determination for Korean Pine, Japanese Larch and Oak to be 0.82, 0.64 and 0.59, respectively. These differences between field-based and LIDAR-based PAIe for the different forest types were attributed to the amount of leaves and branches in the forest stands. The absence of leaves, in the case of both Larch and Oak, meant that the LiDAR pulses were only reflected from branches. The probability that the LiDAR pulses are reflected from bare branches is low as compared to the reflection from branches with a high leaf density. This is because the size of the branch is smaller than the resolution across and along the 1 meter LIDAR laser track. Therefore, a better predictive accuracy would be expected for the model if the study would be repeated in late spring when the shoots and leaves of the deciduous trees begin to appear.
KeyWords: leaf area index, plant area index, LiDAR, k-means clustering, gap fraction, beer-lambert law