USE OF ANALOG, DIGITAL AND LASER SCANNER EQUIPMENT IN THE ESTIMATION OF DENDROMETRIC VARIABLES
Name: MILLENA DUARTE MACHADO
Publication date: 29/06/2023
Examining board:
Name |
Role |
|---|---|
| GILSON FERNANDES DA SILVA | Advisor |
Summary: Forest inventories are crucial procedures in obtaining qualitative and quantitative information about the composition, structure and condition of native or planted forests, and in the management and decision -making process in forestry companies. The most measured variables in a forest inventory are the diameter at 1.30 m from the ground level (D) and the total height (H) of the trees, as they are easier to obtain and, in the case of planted eucalyptus forests for wood production, are important for the study of growth, in the quantification of volume, biomass and carbon, in addition to serving as indicators of the productive quality of a place. The study was divided in to two chapters, the first of which aimed to evaluate the accuracy of a Portable Laser Scanner (PLS) combined with a Simultaneous Mapping and Locating (SLAM) system to identify the position of Eucalyptus grandis trees and estimate the diameter at 1.30 m from ground level (D) and the total height (H), and the accuracy of these estimates at different point cloud densities. The second chapter aims to evaluate the accuracy of the Suunto and Haga hypsometers and the PLS-SLAM for estimating the total height in relation to the reference value, obtained by the total station. The study of the two chapters was carried out in a plot containing 71 individuals of Eucalyptus grandis located in Jerônimo Monteiro, Espírito Santo. In chapter one, all 71 individuals in the pl ot had their D and H measured with the aid of a compass to estimate the diameter and total station for height, and a PLS-SLAM sweep was also carried out in the area. Subsequently, the point cloud obtained with the PLS was processed in order to extract the D and H estimates from the analyzed trees, in five different point densities, 36,000, 1,000, 500, 100 and 10 returns.m-2, in addition, for the variable H, processing was carried out manually and automatically. In chapter two, the total height estimates made with the Suunto hypsometer, Haga hypsometer and the PLS-SLAM through automatic and manual processing were compared with the reference measurements estimated with the total station. In both chapters, the accuracy of these estimates was evaluated using the root mean error (RMSE) and the relative and absolute bias, the correlation between
the data was evaluated using Pearson 's correlation coefficient (r) and, the Graybill's F statistics (1976) to test the similarity of the variables. As main results, in chapter one, it was observed that point clouds of up to 100 returns.m-2 produced results with satisfactory accuracy for D and H, and that the degradation of the point cloud tended to reduce the accuracy of the estimates of the D and H. And in chapter two, it was found that the PLS-SLAM through manual processing was significantly more accurate in estimating H than automatic processing and the Suunto and Haga hypsometers.
Keywords: Forest inventory; Mensurement; Estimates; LiDAR.
