Quantifying the operator effect in LSPIV image-based velocity and discharge measurements

The operator effect is a prominent error source in image-based velocimetry methods. Video sampling, ortho-rectification parameters, motion analysis parameters and filters can strongly impact velocity and discharge measurements. This has been reported in the literature (e.g. Detert, 2021) and highlighted by the Video Globe Challenge 2020, a video gauging intercomparison (Le Coz et al., 2021). The parameter choices made by the operator must be assisted to contain errors and to make image analysis methods accessible to non-specialists.

An investigation of the operator effect (or parameter effect) in various situations is proposed. The analysis focuses on the LSPIV measurements carried out during the Video Globe Challenge 2020. This contest involved around 15 participants with varying levels of experience, challenged over 8 videos. All the LSPIV measurements were replayed based on the data submitted by the participants. The objective was to identify the most sensitive parameter(s) for each video, based on an extensive analysis of the replayed velocity and discharge results.

The data retrieved were: video sampling rate, number of frames, ortho-rectification resolution, IA and SA sizes, correlation based and vector based filters, surface velocity coefficient (a.k.a. alpha) and transect interpolation parameters. To ensure valuable comparisons, grid points and video sequencing were fixed the same for all the participants. Replaying LSPIV measurements allowed to play with the parameters methodically and to quantify their impact on the measured discharge deviation from the reference.

Several lessons were learned from these analyses thanks to the variety of conditions offered by the 8 videos. A tendency to under-estimate the discharge in case of inappropriate parameters was observed. The influence of the video sampling rate has been noticed in many cases. It turns out to have more impact than the motion analysis parameters. The dataset was used to evaluate the benefit of automated parameters setting tools, e.g. ensemble correlation, automated time-interval, automated video sequencing.

 

Detert, M. (2021). How to avoid and correct biased riverine surface image velocimetry. Water Resources Research, 57, e2020WR027833. https://doi.org/10.1029/2020WR027833

 

Le Coz, J., Hauet, A., and Despax, A. (2021). The Video Globe Challenge 2020, a video streamgauging race during the Covid-19 lockdown, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2116, https://doi.org/10.5194/egusphere-egu21-2116, 2021