No abstract

Terrestrial laser scanning is a promising technique for automatic measurements of tree stems. The objectives of the study were (1) to develop and validate a new method for the detection, classification and measurements of tree stems and canopies using the Hough transformation and the RANSAC algorithm and (2) assess the influence of distance to the scanner on the measurement accuracy. Tree detection and stem diameter estimates were validated for 16 circular plots with 20 m radius. The three dominating tree species were Norway spruce (Picea abies L. Karst.), Scots pine (Pinus sylvestris L.) and birch (Betula spp.). The proportion of detected trees decreased as the distance to the scanner increased and followed the trend of decreasing visible area. Within 10 m from the scanner, the proportion of detected trees was 87% on average for the plots and the diameter at breast height was estimated with a relative root-mean-square-error (RMSE) of 14%. The most accurate diameter measurements were obtained for pine, which had a RMSE of 7% for all the full 20 m radius plots. The RANSAC algorithm reduced noise and made it possible to obtain reliable estimates.

Properties of individual trees can be estimated from airborne laser scanning (ALS) data provided that the scanning is dense enough and the positions of field-measured trees are available as training data. However, such detailed manual field measurements are laborious. This paper presents new methods to use terrestrial laser scanning (TLS) for automatic measurements of tree stems and to further link these ground measurements to ALS data analyzed at the single tree level. The methods have been validated in six 80 × 80 m field plots in spruce-dominated forest (lat. 58° N, long. 13° E). In a first step, individual tree stems were automatically detected from TLS data. The root mean square error (RMSE) for DBH was 38.0 mm (13.1%) and the bias was 1.6 mm (0.5%). In a second step, trees detected from the TLS data were automatically co-registered and linked with the corresponding trees detected from the ALS data. In a third step, tree level regression models were created for stem attributes derived from the TLS data using independent variables derived from trees detected from the ALS data. Leave-oneout cross-validation for one field plot at a time provided an RMSE for tree level ALS estimates trained with TLS data of 46.0 mm (15.4%) for DBH, 9.4 dm (3.7%) for tree height, and 197.4 dm 3 (34.0%) for stem volume, which was nearly as accurate as when data from manual field inventory were used for training.

No abstract

Individual tree crowns may be delineated from airborne laser scanning (ALS) data by segmentation of surface models or by 3D analysis. Segmentation of surface models benefits from using a priori knowledge about the proportions of tree crowns, which has not yet been utilized for 3D analysis to any great extent. In this study, an existing surface segmentation method was used as a basis for a new tree model 3D clustering method applied to ALS returns in 104 circular field plots with 12 m radius in pine-dominated boreal forest (64°14'N, 19°50'E). For each cluster below the tallest canopy layer, a parabolic surface was fitted to model a tree crown. The tree model clustering identified more trees than segmentation of the surface model, especially smaller trees below the tallest canopy layer. Stem attributes were estimated with k-Most Similar Neighbours (k-MSN) imputation of the clusters based on field-measured trees. The accuracy at plot level from the k-MSN imputation (stem density root mean square error or RMSE 32.7%; stem volume RMSE 28.3%) was similar to the corresponding results from the surface model (stem density RMSE 33.6%; stem volume RMSE 26.1%) with leave-one-out cross-validation for one field plot at a time. Three-dimensional analysis of ALS data should also be evaluated in multi-layered forests since it identified a larger number of small trees below the tallest canopy layer.