Modelling submerged fluvial substrates with structure‐from‐motion photogrammetry

Document Type

Article

Publication Date

8-30-2019

Department

Department of Biological Sciences

Abstract

Natural sediment regimes of fluvial systems are variable and important to the biological and physical structures of rivers, yet watershed degradation has led to increased fine sediments entering and aggrading in rivers. As a result, quantifying substrate composition is important for targeting and monitoring restoration. Conventional methods for assessing substrate composition (e.g., pebble counts) can be time‐consuming and biased. We examined the use of the photogrammetric technique, structure‐from‐motion (SfM), as an alternative method by measuring streambed roughness. We expanded its application to submerged substrates in an artificial streambed to assess if roughness could predict pebble count substrate size percentiles across a range of manipulated levels of fine sediment aggradation. We then assessed the use of SfM in a free‐flowing river streambed. Results from the artificial streambed with coarse substrates (≤31% added fine sediment) revealed that repeated SfM models of the same streambed had a high degree of similarity (mean difference = 1 mm) and a strong relationship between SfM‐derived roughness and pebble counts (r2 > .95). This relationship was weaker (r2 < .66) and violated regression variance assumptions when substrates had up to 47% (55.7 kg) fines added, possibly due to SfM characterizing details not captured by pebble counts. In the natural streambed, there was a strong relationship between percentiles from the SfM model roughness and pebble count diameter (r2 = .96). SfM appears to be an efficient and appropriate alternative to direct substrate measurements across a broad range of streambed substrate compositions and thus a useful tool to model streambed morphology.

Publisher's Statement

© 2019 John Wiley & Sons, Ltd. Publisher’s version of record: https://doi.org/10.1002/rra.3532

Publication Title

River Research and Applications

Link to Full Text

Share

COinS