Estimating three-dimensional tree shape of 40-year-old evergreen conifer using terrestrial laser scanner and a voxel-based analysis

Abstract

Estimating three-dimensional shape of 40-year-old evergreen using terrestrial laser scanner and a voxel-based Three-dimensional point cloud data can be extracted using a terrestrial laser scanner, and three-dimensional forest information is applicable to various forestry studies. It is required that the estimation of a three-dimensional structure is accurate enough. Here, a voxel-based data processing seems to be an effective method because it divides space using the virtual cube with reduced data. In this study, we attempt to estimate a three-dimensional structure in evergreen coniferous species in a mature stand. Measurements were conducted using a major species of the boreal forest, Abies sachalinensis , in a 40-year-old plantation site. Using point cloud data obtained by terrestrial laser scanner, the vertical trend of trunk diameter was firstly measured, and then the stem volume and branch-leaf biomass at every 2 m layer were estimated using a voxel-based method with eight kinds of candidate size options. They were compared to the actual data acquired after artificial felling of trees. Trunk diameter was successfully estimated from ground to tree height of more than 60%, showing an RMSE of 2.30 cm. It is also demonstrated that error tends to increase as the measurement position gets higher. Stem volume represented high correlation with the manually measured values, except for the case that uses a maximum size of voxel size（0.2 m）.Especially, the voxel size of 0.08 m