The Riverflow 2016 conference had a full session on recent research in image-based measurements and video analysis. It is exciting to watch innovation in process as these researchers learn to exploit the capabilities of emerging consumer technologies.
Never mind that the primary use of these technologies is so that people can instantly share their sense of place in the ‘real world’ within the virtual world where they really spend their lives. Without the billions of people motivated to lay claim to their physical existence with photos and videos, the technology for water monitoring using digital imaging would neither be accessible nor affordable.
We can be glad that this technology is both affordable and accessible. For us, it is the ‘real world’ that is our primary focus.
In the beginning, if someone needed to know how deep or fast a river was they would look at the river to ‘see’ if the current condition met their requirements. Eventually, we developed technologies that replaced these visual observations with mechanical – and then electronic – measurements of water elevation (and other parameters) that could be recorded continuously.
Digital imaging gives us the best advantages of both visual identification and electronic monitoring. Here is a high level summary of some of the relevant presentations.
Measurements of debris flow depth, velocity, and volume defy conventional technology. In the presentation “Image-based instrumentation development for measurement of debris flow characteristics” Abby Bassie demonstrated how a synchronized pair of GoPro cameras could be used to create 3-dimensional reconstructions and measurements of debris flow events.
Measuring flood discharge is in many cases impossible, or at least impractical, using conventional technology. In the presentation “Monitoring river flood using fixed image-based stations: Experience feedback from 3 rivers in France” Alexandre Hauet described how fixed location Large Scale Particle Image Velocimetry (LSPIV) could be added to an existing hydrometric network. The addition of LSPIV gaugings can significantly reduce the uncertainty of flood volumes relative to the alternative of extrapolating a rating curve beyond the highest stage, gauged using conventional measurements.
It’s often not possible, nor practical, to find an ideal location for a fixed camera installation. Unmanned Air Vehicles (UAV) are well suited for digital imaging for most consumer applications. In the presentation “Development of efficient image stabilization algorithm for airborne video images and its application to river flow measurements” Ichiro Fujito described how airborne (i.e. helicopter or UAV) images collected in either navigation mode (i.e. image background is changing) or hover mode can be used for both LSPIV and space-time Image velocimetry (STIV).
One significant advantage of image-based monitoring is the ability to collect velocity data over a whole river reach at a time. In the presentation “Unmanned aerial vehicle-based surface PIV experiments at Surb Creek” Martin Detert described an experiment collecting reach-scale surface velocity measurements using low-cost airborne LSPIV that would have been impossible to obtain using conventional point velocity field techniques. To enhance the targets used for image processing wood chips were introduced into the stream to use as tracers.
The instantaneous water surface velocity field obtained from UAV LSPIV surveys can be used to calibrate low-cost numerical models. In the presentation “Assessing the use of UAV to quantify flow processes in rivers” Gianluca Blois described how data can be collected with no risk to human operators even during extreme natural events.
One inherent problem of image-based monitoring is a dependence on ambient light, hence when flood peaks occur at night it is difficult to capture proper images. In the presentation “Development of a portable surface image velocimeter by using a far infra-red camera” Kwonkyu Yu described how a far infra-red camera was tested for use for surface image velocimetry. As well as being able to operate at night, additional resolution can be added by injection of ice-cubes to use as tracers. These environmentally benign tracers give very clear targets for image processing.
One apparent limitation of image-based monitoring techniques is that it is a surface observation and hence lacks information in the vertical profile. However, in a plenary talk “Exploiting surface turbulence metrics and secondary flows in straight river reaches and open channels” Todd Cowen explained how “the strong physical connection between surface integral length scales and the flow depth” could be exploited for quantitative infrared imaging of flow metrics. The thermal disequilibrium of turbulent flow provides plenty of targets to characterize turbulence. There is a linear relation between the depth of flow and the integral length scale of surface turbulence. This means that it is possible to measure changes in bathymetry as they are evolving in real time. There is no need to be dependent on cross section surveys that may not be valid during episodes of maximum bed stress during the peak of an event.
One major advantage of using a consumer technology that’s in everyone’s pocket is that now there are billions of people with the capability of doing streamflow measurements anytime, anywhere. In the presentation “RIVeR –Towards affordable, practical and user-friendly toolbox for Large Scale PIV and PTV techniques” Antoine Patalano described a free toolbox for image acquisition, image processing, results rectification, and flow discharge calculations. He also talked about a pilot project in Argentina where people could upload videos of flash floods for processing into discharge measurements. This is the ultimate in citizen science – where every citizen has state of the art technology ready to deploy at a moment’s notice!
In spite of an increase in cost of an order of magnitude (relative to mechanical current meters), hydroacoustic technologies for streamflow measurement went from a status of ‘interesting research’ to being the mainstream ‘go-to’ tool for hydrographers all over the world in the span of a decade or so. Given the relatively low cost of image-based monitoring technology, the ability to measure the unmeasurable, and the relative safety benefits, I doubt that it will take that long for image-based monitoring to become a dominant technology for water monitoring.