Short Communication: Challenges and Applications of Structure-from-Motion Photogrammetry in a Physical Model of a Braided River

Abstract

Abstract. Extending the applications of Structure-from-Motion (SfM) photogrammetry in river flumes, we present the main challenges and methods used to collect a large dataset (> 1000 digital elevation models) of high-quality topographic data using close-range SfM photogrammetry with a resulting vertical precision of ∼ 1mm. Automatic target-detection, batch processing, and considerations for image quality were fundamental to successful implementation of SfM on such a large dataset, which was used primarily for capturing details of gravel-bed braided river morphodynamics and sedimentology. While the applications 5 of close-range SfM photogrammetry are numerous, we include sample results from DEM differencing, which was used to quantify morphology change and provide estimates of water depth in braided rivers, as well as image analysis for mapping bed surface texture. These methods and results contribute to the growing field of SfM applications in geomorphology and close-range experimental settings in general.

Experiments on restoring alluvial cover in straight and meandering rivers using gravel augmentation

“…Photo surveys using high‐resolution overhead cameras were completed after the water was drained from the flume each time the discharge was stopped. Two digital SLR cameras were mounted on a trolley 2.9 m above the channel and in a near‐vertical, fully convergent configuration (Leduc, Peirce, & Ashmore, 2019). Pixel resolution on the ground was approximately 0.7 mm.…”

“…Pixel resolution on the ground was approximately 0.7 mm. Images were acquired by running the trolley along the flume with 70–80% overlap longitudinally (Leduc et al, 2019). Ground control targets were independently surveyed with a total station and 3‐D intersection was used to calculate their coordinates at a sub‐millimetre precision (Leduc et al, 2019).…”

“…Images were acquired by running the trolley along the flume with 70–80% overlap longitudinally (Leduc et al, 2019). Ground control targets were independently surveyed with a total station and 3‐D intersection was used to calculate their coordinates at a sub‐millimetre precision (Leduc et al, 2019). Photogrammetric processing was completed in Agisoft PhotoScan (e.g., see Leduc et al, 2019; Morgan, Brogan, & Nelson, 2016).…”

“…Ground control targets were independently surveyed with a total station and 3‐D intersection was used to calculate their coordinates at a sub‐millimetre precision (Leduc et al, 2019). Photogrammetric processing was completed in Agisoft PhotoScan (e.g., see Leduc et al, 2019; Morgan, Brogan, & Nelson, 2016). The automated ground control target recognition feature in the software yielded ortho‐images with nominal resolution and precision of the order of 1 mm.…”

Morphodynamic research challenges for braided river environments: Lessons from the iconic case of New Zealand

“…Baynes et al, 2018a) or structure‐from‐motion photogrammetry (e.g. Leduc et al, 2019; Spreitzer et al, 2020) has allowed the detailed quantification of channel form (e.g. Garcia Lugo et al, 2015), spatially distributed erosion and deposition rates, and lateral channel dynamics (e.g.…”