Verifying the accuracy of the leaf area density distribution of an individual tree derived from terrestrial laser scanning while considering the penetration of beams into the crown and the influence of wind

A terrestrial laser scanner-based method for estimating the leaf area density (LAD) distribution is examined while considering the degree of penetration of the laser beams into the tree and the influence of wind, which can lead to errors in outdoor measurements. Two Japanese zelkova (Zlkova serrata) trees of 6 m in height were used. Each tree was scanned from four positions, each at a distance of 4.5 m from the tree. To evaluate the influence of wind on the estimation accuracy of the LAD distribution, laser-beam-transparent sheets were installed around one tree to block the wind. LAD distributions with a voxel size of 0.3 m * 0.3 m * 0.3 m were estimated based on a previously developed method in which the contact frequency between laser beams and leaves was calculated. Through these estimations, two different methods for using data acquired from multiple scanning positions were examined. In the first method, all scanned data and the calculated LAD were integrated (Integration). In the second, the LAD was calculated for each scanning position, and the maximum LAD was adopted for each voxel (Selection). The estimated leaf area was compared with the measured area by stratified clipping. When the wind was blocked, the difference in the estimation accuracy between the Integration and Selection methods was small, even though the number of incident laser beams on each voxel in the Selection method was smaller than that in the Integration method. The estimation errors in determining the leaf area for the upper, middle, and lower layers of the tree were 10-15 % for both methods. When the wind was not blocked and the wind speed reached 0.5 m/s, the LAD was overestimated by both methods, but the difference between the LAD estimated with and without the sheets was within 10 % in the Selection method. Conversely, the LAD estimated by the Integration method was 20 % greater than that estimated by the Selection method. These results indicate that the Selection method is suitable for estimating the LAD distribution in outdoor spaces.