Using Structure from Motion photogrammetry to assess large wood (LW) accumulations in the field

Abstract

River systems in forested catchments are strongly influenced by wood and sediment loading from both natural (e.g. mass wasting) and anthropogenic (forestry) sources. In order to effectively manage impacts on property, infrastructure and the broader ecosystem, as well as to better assess the interactions between large wood (LW), channel morphology and flow hydraulics, more efficient and accurate surveying techniques are required. We present an evaluation of Structure from Motion (SfM) photogrammetry and point-cloud-based techniques for capturing the geometry and volume of LW in the field. Wood accumulations can be notoriously complex 3D objects, with multi-scalar features, and thus difficult to quantify with photogrammetric methods. We introduce a novel workflow to generate volumetric information of organic material accumulations, using SfM and a suite of meshing algorithms. The volumes of two accumulations are assessed individually, using several meshing techniques, and are compared with more conventional volume estimates. Results show volumetric variations of <19% between the interpolated volumes from point clouds and those of meshes, and a slight overestimation of volumes using simple geometric primitives. We discuss the nature of noise and errors that arise within processed results from SfM software (Pix4D) and offer some criteria for generating high quality point cloud models of highly irregular and complex organic material piles. More detailed 3D mesh representations of LW accumulations will be effective for improving computer-aided modelling of jam-induced floods, backwater effects and impacts on channel morphology. Measuring the size and geometric distributions of wood elements, assessing deposit volume and porosity, or estimating the proportion of organic fine material in the deposit, are important quantities to be assessed for wood accumulations in rivers. The photogrammetric methodology presented herewith will be useful for forestry and freshwater managers to estimate LW quantities, but also to develop accurate 3D models of prototype LW accumulations, independent of camera model and environmental condition, for a range of applications and LW accumulation assessment.