Non-destructive individual tree aboveground biomass estimation in tropical rainforest using terrestrial laser scanner

Abstract

Recent methods for detailed and accurate biomass and carbon stock estimation are driven by advances in remote sensing technology. However, this method heavily relies on the availability of species and area dependent allometric equations, which has been long based on the destructive method. This study introduces a non-destructive laser-based approach for individual tree aboveground biomass estimation by developing a semiautomatic approach of individual tree measurement using the collected point cloud. Biomass of individual trees was derived from tree parameters estimated using terrestrial laser scanner (TLS) data and assessed with field collected data. This study also improvised available allometric models for aboveground biomass estimation based on tree species and individual tree properties obtained from TLS. Point cloud for this study were generated using TLS (Riegl-VZ400) representing 118 random trees from 39 plots established in Royal Belum forest reserve in the state of Perak, Malaysia. Individual tree census was carried out to collect detailed primary tree attributes such as diameter at breast height and tree height. The scanning process using TLS was done to acquire point cloud in multiple positions to ensure good visibility of individual tree. Detailed tree measurement was carried out on the point cloud generated from TLS and the results were compared with the ground collected data. The volume of tree trunk is estimated based on cylinder model fitting on point cloud. The biomass of tree trunk is calculated by multiplying the volume with the species dependent wood density values. The biomass of branches and leaves were estimated based on the same concept and the point cloud were fitted with convex-hull approach. The estimated biomass from TLS was compared with the biomass estimated using existing allometric equations. Measurements of individual tree attributes from the point cloud produced diameter at breast height estimates with of 0.06 cm root mean square error with overestimation of 0.03cm. The root mean square error value for tree height and crown base height estimates is 7.10m and 4.31m with underestimation of 3.07m and 1.05m respectively. In general, the estimated biomass of tree trunk shows strong correlation with biomass value obtained from the allometric equation with r value of 0.97. The estimated branch and leaves biomass show poor relationship with biomass estimated using existing allometric equations with r value of -0.12 and 0.24 respectively. The findings on speciesspecific non-destructive laser-based approach suggests similar correlation pattern observed for biomass of stem, branches, leaves and total aboveground biomass of all tree species with mean of r value of 0.92, -0.12, 0.24 and 0.91 respectively. The proposed methodology and results obtained in this study allow generation of species-specific allometric equations in which suitable with LiDAR-derived variables for individual trees biomass estimation which is a promising alternative approach to the destructive method.