| Summary: | 碩士 === 國立屏東科技大學 === 森林系所 === 103 === Forest canopy structure is composed by the various species. Sun light is a main factor to affect the crown structures after tree competition. However, thinning operation is an appropriate way to control canopy density, which can adjust the competition conditions in the different crown structures. Recently, Airborne Laser Scanning (ALS), also referred to as Light Detection and Ranging (LiDAR), has been established as a standard technology for high precision three dimensional forest data acquisition; it could get stand characteristics with three-dimensional information that had develop potential for the structure characteristics of forest canopy. The 65 years old, different planting density of Cryptomeria japonica experiment area was selected for this study in Chitou area. We used LiDAR image to estimate LiDAR characteristic values by constructed CHM, voxel-based LiDAR, multiple echoes, and assess the accuracy of stand characteristics with intensity values and field data; moreover, to explore the correlation between the LiDAR characteristic values and stand characteristics via different spatial scales and different canopy density. Furthermore, extracting canopy thickness with point cloud frequency distribution of voxel-based LiDAR, and discuss the vertical structure of the stand through variance in the different planting distances by one-way ANOVA analysis. The competition index was calculated with field data, and estimate competition index of LiDAR via multiple linear regression. The tree crown extraction by three different scales through multi resolution segmentation method, and used local maxima to detect individual tree location. Respectively assess the size of crown and accuracy, in order to understand the feasibility of extraction with crown profile and individual tree location of the LiDAR in three different segmentation scale. The results showed that the highest accuracy with stand characteristics was stand high which estimate by LiDAR, its average accuracy of 91.03%. LiDAR raster grid size was 20 m × 20 m for the correlation was the best, however, the higher canopy density will reduce the accuracy of the LiDAR characteristic values to estimate the stand characteristics. The significantly affect canopy thickness and the degree of competition in different planting distances. The canopy thickness, intensity of the point cloud, branch height, canopy density, and ratio of only echo of the five LiDAR indicators was obtained by competition index, that showed a high correlation (R2=0.81). Through the minimum segmentation scale was more accurate to extract tree crown in the small planting distance, but the overall situation was still showing underestimated. When segmentation scale was 5 m × 5 m for the most suitable of the detection individual tree location in small planting distance, and the segmentation scale of the 7 m × 7 m was optimal in the large planting distance. The result found stand canopy too closure in Chitou field, lead to the estimate of LiDAR caused few error. However, the LiDAR characteristic can estimate the information on the tree growth and stand structure, and profit the interpretation of the tree canopy and stand competition.
|
|---|
