Page tree
Skip to end of metadata
Go to start of metadata

Tips for non-contact stream gaging using captured video

Introduction

The United States Geological Survey (USGS) and Environment and Climate Change Canada (ECCC) are exploring the possibility of using large-scale particle image velocimetry (LSPIV) to obtain measurements of surface velocity in streams and rivers. It is envisioned that LSPIV could be a valuable tool for measuring discharge when traditional measurement techniques are not possible, for verification of theoretical measurements, or as a “backup” to direct measurements of discharge. For example, this method might be especially suited to streams that experience very rapid changes in stage (and discharge), such as those that experience flash flooding. LSPIV may also be used to measure surface velocities for model calibration or other hydraulic studies.

The following guidelines have been developed based on the recommendations and experience of staff at the USGS, the ECCC and researchers at the Center for Water Research and Technology (CETA) of the National University of Cordoba, Argentina. These guidelines are expected to evolve as we learn from our experiences.

An example field note has been provided and should be completed for measurements to be submitted to the Surface Velocity Work Group:

Figure 1

Images illustrating the application of Large Scale Particle Image Velocimetry. a) Mean displacement fields of the water surface of the Picso River, Pisco, Peru. b) Mean displacement fields of the flow over the dam spillway of Arroyo Corto River, Córdoba, Argentina

Quick Start Guidelines

  • Most cameras can be used to capture video of sufficient quality for LSPIV processing. However, avoid wide angle lens, and other sources of image distortion. Look for resolutions at least 640 x 480 pixels.
  • Record video for at least 1 min to enable the best chance for quality results. In processing, the video may be subsampled to a shorter period, but the longer duration videos afford a better chance of providing a steady video with good seeding/surface tracking potential.
  • The camera platform (tripod, person, etc.) should be as stable as possible. If a tripod or other fixed mount is not an option, try to stabilize the image by bracing against a fixed object, or shooting from your knee. There are several internet examples for how to stabilize video shot without a tripod.
  • If taking an oblique-angle video (i.e., from the ground/bank), avoid perspective angles lower than 15°.  NOTE: this applies to oblique perspective video. If you capture video perpendicular to (directly above) the flow, and can see the entire region of interest and fixed locations, that, too, is acceptable.
  • Ensure the frame of view of the video includes the entire width of the measurement cross section, and has the highest angle (as close to 90°) as possible (For example, looking down on the water is preferable to looking across the water).
  • Ensure that the frame of view includes fixed locations (e.g. banks, trees, structures) on both sides of the channel within the image. Try to get fixed objects that are close to the water surface elevation.
  • Include a minimum of four (4) fixed and permanent reference points, where the distances between those points can be measured, either at the time of recording or at a later date (stakes or other markers may be temporarily installed if existing reference points are not available or suitable).
  • Film a zone where surface flow disturbance patterns are more uniform with time. Some surface disturbance is fine, but if that disturbance is erratic or propagates along the stream due to some process other than advection by the bulk flow, it will contaminate the LSPIV results and lessen their veracity. For example, wind waves going upstream will effectively reduce downstream surface velocities. Moreover it has to be assumed that the roughness is moving in the same plane as the water surface. (e.g. strong depression of the water surface should be avoided).
  • The best river sections for capturing video have stable bottoms not subject to erosion. It is understood that this is not always known or possible.
  • Avoid reflections, shadows, and sparkling patterns on filmed surface.
  • Consider the effects of pier wake and other flow disturbances on velocities when taking video from a bridge. It may be preferable to shoot video looking upstream of a bridge, or from the banks depending on how the bridge affects the flow.

Detailed Guidelines

  1. Any camera sensor can be used that has video resolution larger than 640 x 480 Pixels (includes most smart phones). If higher definition settings are available, use them. Video should be captured at a rate of 15 frames per second (fps) or faster. For high velocity flows (a dam spillway for example), even higher video fps settings/capabilities are preferred and potentially necessary. Cameras capable of rates upward of 60 fps are becoming commonplace consumer products. Videos captured should be as stable as possible, either by supporting the camera on/against a fixed object, such as the railing of a bridge or ideally by using a tripod. Panning and zooming with the camera should avoided.

    Figure 2

    Example set up where the digital camera is supported on a railing of a lock gate.
  2. Lenses that distort the image should not be used, such as a wide angle lens. When using a zoom lens, be careful not to use the wide-angle setting. In the event that such a lens is used, it is important to record and save the lens type information with the data, so that corrections for distortion can be made during post-processing.

  3. Identify a suitable location for obtaining video of the flow. Such a location should include 4 readily visible fixed reference (control) points. The distances between the control points can be measured either at the time the video is recorded, or during a subsequent visit after the water has receded. Video may be obtained from a bridge (Figure 3) or from the shore (Figure 4), being careful to avoid wake effects from piers.

    Figure 3

    Single frame of a video recorded from a bridge where the control points were rocks
    Figure 4

     Single frame of a video recorded from shore where the control points are tree trunks
  4. Capture video over the entire width of the measurement cross section from as high an angle (close to 90°) as possible including banks from both sides in the image (Figure 5 and 6). Error in the LSPIV-processed results increases as the angle between the camera and the water surface decreases to 0° (at water surface level).

    Figure 5

    Single frame of a video recorded from a bridge where both banks are visible. Notice that the perspective of this video is neither oblique or normally oriented to the flow, but is instead at about a 45° angle to flow. This is not optimum, but valid results can still be obtained.
    Figure 6

    Single frame of a video recorded from a lock where both banks are visible. 
    Figure 7

     Single frame of a video recorded where banks are not visible and no reference points.
    Figure 8

    Snapshot of a video recorded from a lock including the reference points.
    Figure 9

    Snapshot of a video recorded from shore including the reference points (CP1 through CP4 ).
  5. Avoid reaches that are known to be susceptible to scour and fill as the streambed elevations cannot usually be measured at the time of the filming.

  6. Avoid reflections, shadows, and sparkling patterns on filmed surface, such as those shown in figure 10. Polarizing lens filters can be used to mitigate or remove surface reflections.

    Figure 10

    Single frame of a video recorded from a bridge where reflections, large shadows, and sparkling patterns are observed.
  7. Record the location coordinates of the video. Ideally this would be the latitude/longitude obtained from an accurate GPS. However, very accurate descriptions of the measurement location would also be acceptable.

  8. Record the exact date and time of the recording, using the time zone of the recorder at the site (if obtained from a streamgaging station).

Examples

San Antonio River, Córdoba, Argentina

Results of the LSPIV analysis process during a flash flood event in the San Antonio River, Córdoba, Argentina. a) Snapshot at the beginning of the event. b) Results of LSPIV processing at the beginning of the event. c) Mean flow velocity time evolution of the recorded event. d) Discharge time evolution of the recorded event and trendline.

A. Patalano, C. M. Garcia, W. Brevis, T. Bleninger, N. F. Guillén, L. Moreno, A. Rodriguez, (2015), “Recent Advances In Eulerian And Lagragian LargeScale Particle Image Velocimetry”, 36th IAHR World Congress, The Hague, Netherlands, June 2015.

LSPIV Rating

Results of the LSPIV process for many flow conditions in the San Antonio River, Córdoba, Argentina. a) Results of LSPIV processing at the beginning of an event. b) Velocity profile of the corresponding event. c) Discharge results from different conditions versus water surface elevation. ADCP results and Water National Institute (INA) rating curve are also plotted.

N. F. Guillén, A. Patalano, C. M. García, “Validación de la Técnica PIV a Gran Escala (LSPIV) para la estimación de caudales en ríos de montaña” in English: “Validation of LSPIV for flow discharge measurements in mountain rivers”, IV Simposio Sobre Métodos Experimentales En Hidráulica, La Plata, Argentina, March 2015.

Contacts

US Geological Survey

Kevin Oberg
217-328-9739
kaoberg@usgs.gov
405 N. Goodwin Ave.
Urbana, IL, 61801

Robert R. Holmes, Jr.
573-308-3581
bholmes@usgs.gov
1400 Independence Road
Mail Stop 100
Rolla, MO, 65401-2602

Frank Engel
217-328-9774
fengel@usgs.gov
405 N. Goodwin Ave.
Urbana, IL, 61801

Environment and Climate Change Canada

Elizabeth Jamieson
613-992-9337
Elizabeth.jamieson@canada.ca
Head, Hydrometric Monitoring Technology Unit
Water Survey of Canada - Relevés hydrologiques du Canada

CETA - Center for Water Research and Technology

C. Marcelo Garcia
National University of Cordoba, Argentina
cgarcia2mjc@gmail.com


Antoine Patalano
National University of Cordoba, Argentina
antoine.patalano@gmail.com

Error rendering macro 'contentbylabel' : parameters should not be empty