Dense image matching of terrestrial imagery for deriving high-resolution topographic properties of vegetation locations in alpine terrain

• Published in: International journal of applied earth observation and geoinformation Vol. 66; pp. 146 - 158
• Main Authors: Niederheiser, R., Rutzinger, M., Bremer, M., Wichmann, V.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.04.2018
• Subjects: Biogeomorphology; Dense matching; Multi-view-stereo; Photogrammetry; Point clouds; Structure-from-Motion; Topography; Vegetation; Topography; Biogeomorphology; Structure-from-Motion; Point clouds; Vegetation; Dense matching; Multi-view-stereo; Photogrammetry
• ISSN: 1569-8432; 1872-826X
• DOI: 10.1016/j.jag.2017.11.011

•249 photogrammetrically modelled alpine vegetation locations.•Rapid capturing and automated processing.•Multi-scale 3D topographic analysis and solar irradiation potential.•Alpine vegetation location characterisation. The investigation of changes in spatial patterns of vegetation and identification of potential micro-refugia requires detailed topographic and terrain information. However, mapping alpine topography at very detailed scales is challenging due to limited accessibility of sites. Close-range sensing by photogrammetric dense matching approaches based on terrestrial images captured with hand-held cameras offers a light-weight and low-cost solution to retrieve high-resolution measurements even in steep terrain and at locations, which are difficult to access. We propose a novel approach for rapid capturing of terrestrial images and a highly automated processing chain for retrieving detailed dense point clouds for topographic modelling. For this study, we modelled 249 plot locations. For the analysis of vegetation distribution and location properties, topographic parameters, such as slope, aspect, and potential solar irradiation were derived by applying a multi-scale approach utilizing voxel grids and spherical neighbourhoods. The result is a micro-topography archive of 249 alpine locations that includes topographic parameters at multiple scales ready for biogeomorphological analysis. Compared with regional elevation models at larger scales and traditional 2D gridding approaches to create elevation models, we employ analyses in a fully 3D environment that yield much more detailed insights into interrelations between topographic parameters, such as potential solar irradiation, surface area, aspect and roughness.