Aboveground biomass estimation in linear forest objects: 2D- vs. 3D-data

Abstract

Wood-chips of linear forest objects (hedge banks and roadside plantings) are used as sustainable energy supply in wood-chip heating systems. However, wood yield of linear forest objects is very heterogeneous and hard to estimate in advance. The aim of the present study was to compare the dry mass estimation potentials of two different non-destructive data: (i) Canopy area (derived from aerial images) and mean age at stump level (2D), (ii) volume of vegetation cover based on structure from motion (SfM) via unmanned aerial vehicle (3D). These two types of data were separately used to predict reference dry mass (ground truth) in eleven objects (5 hedge banks and 6 roadside plantings) in Schleswig-Holstein, Germany. The predicting potentials were compared afterwards. The reference dry mass was ascertained by weighing after harvesting and drying samples to constant weight. The model predicting reference dry mass using canopy area and mean age at stump level achieved a relative root mean square error (RMSE) of 52% (42% at larger combined plot sizes). The model predicting reference dry mass using SfM volume achieved a relative RMSE of 30% (16% at larger combined plot sizes). This result indicates that biomass is better described by volume of vegetation cover than by canopy area and age.

Keywords

• sfm
• aerial images
• hedge banks
• roadside plantings
• uav
• dry biomass