Subaerial Gravel Size Measurement Using Topographic Data Derived From a UAV-SfM Approach

Woodget, Amy and Austrums, Robbie (2017) Subaerial Gravel Size Measurement Using Topographic Data Derived From a UAV-SfM Approach. Earth Surface Processes and Landforms, 42 (9). pp. 1434-1443. ISSN 0197-9337 Online: 1096-9837

Preview	Text Woodget and Austrums 2016 Submitted.pdf - Submitted Version Download (1MB) | Preview
Preview	Text Subaerial Gravel Size Measurement Using Topographic Data Derived From a UAV-SfM Approach.pdf - Accepted Version Download (196kB) | Preview

Abstract

Fluvial grain size plays a fundamental role in determining the condition and availability of aquatic habitats. Remote sensing provides rapid and objective methods of quantifying fluvial grain size, and typically provide coarse grain size outputs (c. 1m) at the catchment scale (up to 80km channel lengths) or fine resolution outputs (c. 1mm) at the patch scale (c. 1m2). Recently, drone based approaches have started to fill the gap between these scales, providing hyperspatial resolution data (<10cm) over reaches up to a few hundred metres in length. This ‘mesoscale’ is of importance to habitat assessments and is aligned with the ideals of the ‘Riverscape’ concept. Most drone based grain size measurement approaches use textural variables computed from drone orthoimagery. To date however, no published works provide quantitative evidence of the success of this approach, despite significant differences in platform stability and the image quality obtained by manned aircraft versus drones. With interest in drone surveys growing rapidly, such error quantification is essential for making reliable, evidence-based recommendations about the suitability of drones for routine management of fluvial environments. Here we provide an initial assessment of the accuracy and precision of grain size estimates produced using two different drone-based methods; (1) the image textural variable ‘negative entropy’, and; (2) the roughness of point clouds derived from drone imagery processed using structure from motion photogrammetry. Data is collected from a small gravel-bed river in Cumbria, UK. Results from jack knife analyses show that the point cloud roughness method gives more accurate and precise measures of grain size at this site, as indicated by the mean (0.0002m) and standard deviation (0.0184m) of residual errors. However, both methods struggle to provide grain size measures with sub-centimetre precision. We suggest that blur within the drone imagery prevents better precision, resulting from an inadequate camera gimbal.

Actions (login required)

View Item