SEMI - AUTOMATIC ASSESSMENT OF NORWAY SPRUCE (PICEA ABIES) IN MODERN DIGITAL AERIAL PHOTOGRAPHS

摘要

In present times, forest conversion due to climate change confronts forest owners and forest administrations with great challenges if vast areas of forested surfaces are concerned by the effects of temperature rise and drought. In Bavaria, especially Norway Spruce (Picea abies) will lose terrain in favor of other tree species by decreasing vitality and the proliferation of bark beetle damage. The Bavarian Forest Service is confronted with the problem, that there is only little information available about the approximately 1.6 Mio ha of nongovernmental forest property in Bavaria to fulfill its advisory tasks, because of data privacy convention. For this reason, the Bavarian State Institute of Forestry conducted a research project focusing on possible semi-automatic approaches of anonymous information extraction from digital aerial photographs. This method mainly comprises the segmentation and object-based classification of 4-channel 16 bit aerial photographs in combination with a preceding image homogenization process. The project showed a classification accuracy of 80percent (young stands) up to more than 90percent (old stands). A reliable classification (up to 80percent accuracy) has even been achieved for Norway Spruce stands younger than 30 years, that cannot be interpreted by visual stereoscopic interpretation. The new methodology enables the Bavarian State Institute of Forestry to gain an identification of Norway Spruce for vast non-governmental forest areas. In a subsequent investigation, the method has to prove its cost efficiency and is to be transferred on larger areas and to other tree species of interest in the debate of environmental change.