Determination of the geometric parameters of vegetation using airborne laser scanning data

Abstract

Vegetation is one of the most important components of the geographic environment. Creating vegetation maps and characterizing vegetation is a common task in forest management. When creating maps of the largest scales, the problem arises of determining the parameters of individual trees—their heights, crown diameters, coordinates of the position of trunks in space (the so-called tree-by-tree survey). Existing ground-based technologies for solving this problem are expensive and inefficient. Satellite imagery do not provide an acceptable result. Aerial photography is also not very suitable for solving this problem (for a number of reasons). The most promising is the use of airborne laser scanning using not too dense clouds of points of laser reflections (4–8 points per 1 m2), which makes it possible to provide such data with significant (up to several million hectares of forest per year per system) forest areas. Performing surveys with a higher density (including using unmanned aerial vehicles) seems to be irrational due to the impossibility of direct measurements of tree trunks and low productivity. The study was conducted on a test area in the North Caucasus region (Republic of Adygea) according to survey data from 2022 (autumn). This work is devoted to the development and evaluation of the results of using the methodology for determining the parameters of individual trees. To do this, it is proposed to use typical functions of relief analysis and hydrological modeling in combination with data filtering using statistical functions within a sliding window. The analysis is carried out using a regular-cell model of relative tree heights. The final result of the work is a point vector GIS layer, where the planned position of the point corresponds to the position of the tree trunk, and its height and crown diameter are recorded in the attribute table. The final results after a complete visual check on a regular cell data model and a selective manual check using orthomosaics and initial points of laser reflections can be assessed as acceptable, but in need of further improvement of the methodology.