Automatic Detection of Small Single Trees in the Forest-Tundra Ecotone Using Airborne Laser Scanning

A large proportion of Norway's land area is occupied by the forest-tundra ecotone. The vegetation of this temperature-sensitive ecosystem between mountain forest and the alpine zone is expected to be highly affected by climate change and effective monitoring techniques are required. For the detection of such small pioneer trees, airborne laser scanning (ALS) has been proposed as a useful tool employing laser height data. The objective of this study was to assess the capability of an unsupervised classification for automated monitoring programs of small individual trees using high-density ALS data. Field and ALS data were collected along a 1500 km long transect stretching from northern to southern Norway. Different laser and tree height thresholds were tested in various combinations within an unsupervised classification of tree and nontree raster cells employing different cell sizes. Suitable initial cell sizes for the exclusion of large treeless areas as well as an optimal cell size for tree cell detection were determined. High rates of successful tree cell detection involved high levels of commission error at lower laser height thresholds, however, exceeding the 20 cm laser height threshold, the rates of commission error decreased substantially with a still satisfying rate of successful tree cell detection.