How to Avoid and Correct Biased Riverine Surface Image Velocimetry

Abstract

Because digital image data acquisition has become low cost and various image analysis tools have become available, the number of image-based measurement applications is increasing and so is the number of misleading applications. This paper shows how to avoid and correct problematic measurement setups and erroneous analyses by reexamining three recently published data sets on the application of surface image velocimetry to rivers. The results of these data sets were greatly improved by following image velocimetry best practices using intrinsic and extrinsic camera calibration, correcting acquisition frame rates, using appropriate image and velocimetry filters, and reducing bias due to near-zero velocities. This paper illustrates how to perform more accurate Particle Image Velocimetry analyses and presents results that are in much better agreement with direct field measurements. General recommendations are given to provide guidance to help end users perform image-based surface flow velocimetry measurements.