Forest delineation based on LiDAR data and vertical accuracy of the terrain model in forest and non-forest area

Abstract

This paper deals with the use of airborne laser scanning data (ALS) in the process of the automatic delineation of forest and the generation of digital terrain models (DTM) in forested and non-forested areas. The study area where the procedures presented were examined is part of the University Forest Enterprise, Technical University in Zvolen (48°37' N, 19°04' E). A partial modification of existing solutions that iteratively takes into account the criteria of minimum area, height, width and crown coverage is presented within the forest delineation. At the same time this approach also evaluates the mutual distance of identified crowns and the presence of buildings. Compared with manually identified forest boundaries, the accuracy of the automated procedure in the study area reached the value of 93%. In the DTM generation, various alternative methods of interpolation and conversion were used, while ALS data from the summer and winter seasons were also available. The results showed that laser scanning in the area of interest provided systematically overestimated data for the DTM generation. The largest deviations of the DTM were found in terrains based on ALS data from the summer season, with a significant slope, regardless of the complexity of the afforestation structure (except for the youngest forest). In older stands and unforested areas, both with a moderate slope, the DTM accuracy achieved was in the range ±6-17 cm.