Browsing by Author "Yrttimaa, Tuomas"

Decaying dead wood is a key factor for forest biodiversity. In boreal forests, many threatened and specialised species are dependent on dead wood. Therefore, information on quantity and quality of dead wood is needed. Conventionally, the inventory of dead wood is based on measurements and observations done in the field with traditional forest measurement tools, which, however, could be replaced by terrestrial laser scanning (TLS). TLS provides a dense point cloud on its surroundings with a millimetre-level of detail enabling versatile measurements at the levels from an individual tree to an entire sample plot. Previous studies have proven TLS to efficiently provide information for mapping standing trees, but the feasibility of TLS for dead wood inventory has not yet examined. The objective of this study was to develop an automatic method for mapping downed dead wood using TLS. TLS data were collected from 20 sample plots (32 m x 32 m in size) using the multi-scan approach with five scanning positions on each plot. All downed dead tree trunks with a diameter exceeding 5 cm at the middle of the trunk were measured in the field and considered as the field reference. Cylinder fitting and surface model segmentation were utilised when the downed dead wood trunks were automatically detected from the point clouds. The trunks were also detected visually to reveal all the potential of the use of a dense point cloud in mapping downed dead wood from a sample plot. Dimensions, volume, geometry-related quality attributes and position in the sample plot were automatically determined for each trunk detected from the point cloud. Based on trunk attributes, a map representing the spatial distribution of downed dead wood, as well as estimates for attributes describing the quantity and quality of downed dead wood at the plot level, were constructed. Finally, a diameter distribution for downed dead wood in the study area was comprised. This study revealed that TLS is a valid method for mapping downed dead wood from sample plots. By utilising the TLS point clouds, 68 % of the downed dead wood volume was detected automatically, while the total volume of downed dead wood was estimated with an RMSE of 15,0 m3/ha. The mapping accuracy could be improved with the visual interpretation of the point cloud, in which case 83 % of the dead wood volume was detected, and the estimate for the total volume of downed dead wood was determined with an accuracy of 6,4 m3/ha. On average, the length of the detected tree trunk was underestimated while the diameter was overestimated since the trunk was not able to be detected entirely from the point cloud. According to the results, the reliability of TLS based dead wood mapping increases alongside the dimensions of the dead wood trunks. The density of plot vegetation, however, causes shading and reduces the trunk detection accuracy. Therefore, when collecting the data, extra attention must be paid to the quality of the point cloud.