Fig. 1. Photograph and point cloud of common ash #2. (left) Ash #2 before felling (on the foreground). (right) Point cloud visualisation of the same tree from the same upward looking perspective as the photographer.

Fig. 2. Map of the scanning pattern. The locations of the harvested ash trees are depicted as vertical projections of their crown hulls. The scan positions of the modulations are shown on the right.

Fig. 3. Examples of common ash branch segment point clouds under a variation of scan position layouts. See Fig. 2 for a description of scan position variations. Segments have a zenith point of view. Points coloured on beam spot size; red bar shows the true-to-scale manually measured branch diameters at each location and the true-to-scale beam spot sizes. In blue the number of points per segment is shown. The manual diameters were 25, 25, 58.9, and 226 mm each segment. View larger in new window/tab.

Fig. 4. TLS point cloud and QSM reconstruction of common ash #2. Bottom: magnification of branches in the upper part of the crown (location indicated by the parallelogram – direction of view indicated by the arrow in panel c). (a) Full point cloud. (b) Manually segmented 10 cm long branch sections. (c) Full QSM of the tree (coloured by z-coordinate value). (d) Individual QSM cylinders that have been successfully matched with the manual measurements. View larger in new window/tab.

Fig. 5. QSM and destructive cumulative volume repartitioning in common ash based on branch diameter sizes. Grey/white bands indicate the branch diameter classes used in the destructive measurements. QSM volumes obtained with the Grid 10 modulation (coregistered using Multi-Station Adjustment 1 (MSA1)), a fuzz filtered and MSA2 point cloud.

Fig. 6. Relative (left) and absolute (right) error in the diameter of common ash QSM cylinders from the reference point cloud ‘Grid 10’ with respect to manually measured branch diameters. Comparison of hand-measured branch diameters and diameters from five QSM reconstruction (mean and standard deviation as grey whiskers). Light-grey/white bands indicate branch diameter classes. The fixed taper measurements are located at the thresholds of the diameter classes. For reasons of clarity, here no standard deviation whiskers are plotted. For visualisation purposes one outlier (error = 257% and hand-measured diameter = 3.3 mm) is not within the plotting frame. View larger in new window/tab.

Fig. 7. Comparison of the QSM volume against destructively measured volume (dashed line) in 5 diameter classes for 12 different point cloud modulations, the fuzz filtered point cloud and the Multi-Station Adjustment 2 coregistration in two common ash trees. QSM volume was extracted as the mean (points) and standard deviation (black whiskers) of five models with optimal input parameters. See Fig. 2 for a description of the scanning patterns.

Fig. 8. Common ash branch diameter error of QSMs with respect to manually measured branches in branch diameter classes. Positive values indicate an overestimation of QSM diameters with respect to manual measurements. Three modulations were compared: the ‘Grid 10’ reference modulation using the standard Multi-Station Adjustment 1 (MSA1) algorithm, the fuzz filtered point cloud and the MSA2 registered point cloud. For visualisation purposes one outlier (error = 257% and diameter class <2.5 mm; MSA1) is not within the plotting frame. Box line represents median value, the upper and lower hinges represent the 25th and 75th percentile and the whiskers extend to the min/max values, except if min/max values are further than 1.5 times the interquartile range (these points are outliers: black dots).

Fig. 9. Conceptual figure of the trade-off for optimal amount of scan positions. Theoretical implication of the number of scan positions on point cloud quality. This trade-off in amount of scan position balances the effects of occlusion and misalignment errors due to coregistration and wind.

Fig. 10. Sizes of laser beam footprint, QSM cylinders and nearest neighbour distances of common ash #1 in function of tree height. dNN: the mean distance to the closest neighbouring point in 10 cm z-slices. The laser beam footprint diameter size was calculated from the ranging distance to the scanner. The realised footprint was inferred from the range of each point in the point cloud and averaged in 10 cm z-slices. The theoretical footprint was simulated from a scanner position at ground level. ‘QSM cylinder’ are the average and 5% percentile of QSM cylinder diameters in 20 cm z-slices.