Detecting moose (Alces alces) browsing damage in young boreal forests from airborne laser scanning data

• Published in: Canadian journal of forest research Vol. 46; no. 1; pp. 10 - 19
• Main Authors: Melin, M, J. Matala, L. Mehtätalo, A. Suvanto, P. Packalen
• Format: Journal Article
• Language: English
• Published: Ottawa NRC Research Press 01.01.2016; Canadian Science Publishing NRC Research Press
• Subjects: Alces alces; ALS; balayage laser aéroporté; Boreal forests; branches; browsing; Clearcutting; dommage forestier; Environmental aspects; forest damage; Forest dynamics; forest structure; forestry; Forests; Habitats; Herbivores; humans; Lasers; lidar; Moose; Observations; Optical radar; orignal; professionals; regression analysis; structure forestière; Taigas; Finland; ANE, Finland
• ISSN: 1208-6037; 0045-5067; 1208-6037
• DOI: 10.1139/cjfr-2015-0326

Large herbivores can have large impacts on their habitats through extensive browsing. Similarly, human actions can have large impacts both on habitats and on the animals utilizing the habitats. In Finland, the increase in clear-cut areas has been highly positive for moose in particular, because these areas provide an easy and abundant source of winter food. For the forest owners, moose browsing causes growth and quality losses or even the destruction of whole stand. We aimed to identify moose browsing damage from airborne laser scanning (ALS) data and to predict damaged areas. The data was used to detect the difference in forest structure caused by moose browsing (lost branches and twigs) in relation to reference areas without moose browsing. The damaged areas were located, measured, and confirmed by forestry professionals, and ALS data was collected after the damage. In the end, the structural differences that browsing caused proved to be clear enough to be detected with metrics calculated from ALS data. Many variables were significantly different between the damage and no-damage areas. With logistic regression, we were able to differentiate the areas with significant, large-scale damage from no-damage areas with a 76% accuracy. However, the model was too keen to predict false-positive cases (classifying no-damage areas as damaged). It was shown that ALS data can be used in detecting moose browsing damage in a case where the damage is extremely severe (like in here). Yet, to make the results more accurate, better field data about the damaged areas would be needed.