Optical techniques for river flow monitoring: an automatic procedure for the identification of the best video sequence to process by LS-PIV technique

Abstract

Technological advances over last decades gave an innovative impulse to the development of new streamflow measurements techniques, especially regarding remote flow monitoring methods that allow for non-intrusive measurements. The most widely used image-based techniques are the optical ones, such as Large-Scale Particle Image Velocimetry (LS-PIV) and Large-Scale Particle Tracking Velocimetry (LS-PTV). The dynamic movement of tracers floating on the water surface of a river is recorded, and the resulting videos are processed by applying a statistical cross-correlation analysis to detect the most probable frame-by-frame displacement, finally deriving the surface velocity field. To obtain river discharge, it is necessary to couple the geometry of a river cross-section with the assessed surface velocity field, adopting simplified assumptions on the vertical velocity profiles. The accuracy of these techniques depends on several factors, such as the processing models set-up, the tracer density and distribution, the video-length, and many other aspects related to environmental and hydraulic conditions. The aim of this work is the identification of an automatic methodology aimed to select an optimal video sub-sequence over the entire recorded video in order to optimize the performance of software based on LS-PIV technique. For this purpose, the results of multiple field campaigns on two Sicilian rivers (Italy), i.e., Oreto and Platani rivers, are exploited. Results are shown in terms of error in the estimation of the surface velocity profile along a specific transect, comparing velocity values assessed by the LS-PIV technique and by an Acoustic Doppler Current Profiler (ADCP).