Volume estimates of trees with complex architecture from terrestrial laser scanning

Abstract

Research on urban forest carbon storage requires new allometric equations that relate easily measured aspects of tree form to total tree volume. Existing methods for creating these equations are time consuming and can require destructive sampling. In this study we used a terrestrial lidar sensor; data produced by this system consists of the three-dimensional locations of points that lay on object surfaces (in this case the trunk, branches, twigs, etc.). The challenge in calculating tree volume was the aggregation of these individual points into objects whose dimensions could be reliably estimated. We converted the lidar observations into 3-dimensional voxels, identified narrow (< 20 cm) sections along the stems, and merged sections on the basis of their connectivity. The distribution of returns from small branches was analyzed on the basis of branch size and the sampling pattern of the sensor. Two hundred and fifty measurements of stem diameter were made on 14 trees of two species using both lidar and field measurements; the estimates were highly correlated (r²> > 98%) and unbiased. The processing algorithm was applied to lidar data from 179 trees of 11 species; the accuracy of results was dependant on tree volume, species and tree form.