Unpiloted Aerial Vehicle (UAV) image velocimetry for validation of two-dimensional hydraulic model simulations

Masafu, C., Williams, R. , Shi, X. , Yuan, Q. and Trigg, M. (2022) Unpiloted Aerial Vehicle (UAV) image velocimetry for validation of two-dimensional hydraulic model simulations. Journal of Hydrology, 612(Part C), 128217. (doi: 10.1016/j.jhydrol.2022.128217)

[img] Text
275298.pdf - Published Version
Available under License Creative Commons Attribution.

7MB

Abstract

Non-intrusive image-based techniques for measuring surface river velocities have rapidly evolved as a cost-effective and safe means for quantifying flow patterns. Large-scale particle image velocimetry (LSPIV) can provide instantaneous surface velocities over a large spatial footprint rapidly and with little pre-calibration as compared to traditional techniques. Assessment of the spatial distribution of flow velocities in hydraulic models has been comparatively harder to achieve than assessment of depth due to logistical challenges but would be aided using large observational datasets that represent the variability and distribution of flow hydraulics. Additionally, the efficacy of image velocimetry in assessing the accuracy of outputs from 2D hydraulic models has not been addressed. Here, we demonstrate how LSPIV can be used to calibrate and validate 2D model predictions in a gravel bed river reach. LSPIV velocities are depth-averaged using standard velocity coefficients (α) and then using the Probability Concept (PC) - a probabilistic formulation of velocity distributions that accounts for non-standard velocity profiles, typical in field settings. UAV surveys were used to acquire video for LSPIV and imagery for Structure from Motion (SfM) topographic modelling. We use spatially dense acoustic doppler current profiler (aDcp) velocity data for benchmark assessment of the velocity outputs of HEC-RAS 2D model simulations. 2D model prediction error, based on seeded LSPIV velocities, was within range (4.2%) of the aDcp parametrised model, with improvements in modelled versus predicted velocity correlations (up to 7.7%) when using PC to depth average LSPIV velocities. Validation bias reduced significantly (11%) with tighter error distributions when compared to the aDcp based model. Although additional hydraulic measurements are required to parametrise the Probability Concept algorithm, the performance of 2D hydraulic models calibrated/validated with LSPIV velocities is on par with traditional techniques, demonstrating the potential of this non-intrusive, low-cost approach.

Item Type:Articles
Additional Information:CM was funded by the Lord Kelvin Adam Smith (LKAS) PhD scholarship at the University of Glasgow. GNSS equipment was provided by Natural Environment Research Council’s Geophysical Equipment Facility (NERC GEF) loan 1118.
Status:Published
Refereed:Yes
Glasgow Author(s) Enlighten ID:Williams, Professor Richard and Masafu, Mr Christopher and Shi, Dr John Xiaogang
Authors: Masafu, C., Williams, R., Shi, X., Yuan, Q., and Trigg, M.
College/School:College of Science and Engineering > School of Geographical and Earth Sciences
College of Social Sciences > School of Social & Environmental Sustainability
Journal Name:Journal of Hydrology
Publisher:Elsevier
ISSN:0022-1694
ISSN (Online):1879-2707
Published Online:19 July 2022
Copyright Holders:Copyright © 2022 The Authors
First Published:First published in Journal of Hydrology 612(Part C): 128217
Publisher Policy:Reproduced under a Creative Commons License

University Staff: Request a correction | Enlighten Editors: Update this record