Terrestrial structure-from-motion: spatial error analysis of roughness and morphology

Schwendel, Arved C.; Schwendel, Arved; Milan, David J.

Abstract

Structure-from-Motion (SfM) photogrammetry is rapidly becoming a key tool for morphological characterisation and change detection of the earth surface. This paper demonstrates the use of Terrestrial Structure-from-Motion (TSfM) photogrammetry to acquire morphology and roughness data at the reach-scale in an upland gravel-bed river. We quantify 1) spatially-distributed error in TSfM derived Digital Elevation Models (DEMs) and 2) identify differences in roughness populations acquired from TSfM photogrammetry versus TLS. We identify an association between local topographic variation and error in the TSfM DEM. On flatter surfaces (e.g. bar and terrace surfaces), the difference between the TSfM and TLS DEMs are generally less than ±0.1 m. However, in areas of high topographic variability (>0.4 m) such as berm or terrace edges, differences between the TSfM and TLS DEMs can be up to ±1 m. Our results suggest that grain roughness estimates from the TSfM point cloud generate values twice those derived from the TLS point cloud on coarse berm areas, and up to four-fold those derived from the TLS point cloud over finer gravel bar surfaces. This finding has implications when using SfM data to derive roughness metrics for hydrodynamic modelling. Despite the use of standard filtering procedures, noise pertains in the SfM DEM and the time required for its reduction might partially outweigh the survey efficiency using SfM. Therefore, caution is needed when SfM surveys are employed for the assessment of surface roughness at a reach-scale.

Citation

Schwendel, A. C., Schwendel, A., & Milan, D. J. (2020). Terrestrial structure-from-motion: spatial error analysis of roughness and morphology. Geomorphology, 350, Article 106883. https://doi.org/10.1016/j.geomorph.2019.106883

• Journal Article Type: Article
• Acceptance Date: Sep 27, 2019
• Online Publication Date: Oct 21, 2019
• Publication Date: Feb 1, 2020
• Deposit Date: Nov 18, 2019
• Publicly Available Date: Oct 22, 2020
• Journal: Geomorphology
• Print ISSN: 0169-555X
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 350
• Article Number: 106883
• DOI: https://doi.org/10.1016/j.geomorph.2019.106883
• Keywords: Digital Elevation Model (DEM); Error; Roughness; SfM photogrammetry; Terrestrial Laser Scanning (TLS)
• Public URL: https://hull-repository.worktribe.com/output/2970753
• Publisher URL: https://www.sciencedirect.com/science/article/pii/S0169555X19303745?via%3Dihub
• Contract Date: Nov 18, 2019