Image-based Lagrangian Particle Tracking in Bed-load Experiments

The manuscript presents a protocol for the conduction of bed-load sediment transport experiments where the moving particles are tracked by image analysis. The experimental facility, the procedures for run realization and data processing, and finally some proof-of-concept results are presented here.

Transcript

The overall goal of this procedure is to measure the trajectories of individual particles transported by a flow as bed-load over a large area of observation to avoid experimental bias related to the maximum length of measurable tracks. This method can help answer key questions in bed-load transport mechanics as it provides detailed information about the particle's motion and rest events. The main advantage of this technique is that the transported particles can be tracked for long distances, enabling long exams to be observed.

We first had the idea of this method while analyzing data from previous experiments, when we realized that particles can be transported over longer distances with respect to the observation windows. First, prepare a set of steel plates covered in a layer of the sediment particles of interest. Coat the sediment surface in a water-resistant black paint.

Next, place PVC supports on the bottom of the flume and lay the plates on the supports. Adjust the plate placement as needed to ensure that the sediment bed is continuous. Then cover the channel with transparent acrylic lids.

Set the flume to the desired slope. Then turn on the flume pump and fill the channel with water. Set the flow rate using the regulation valve.

Set the elevation of the pressure head to slightly above the channel lids using the tail water regulator. Ensure that the flow is covered without exerting significant force on the lids. Check the flow rate and pressure head elevation every 15 minutes until the change between measurements is small, indicating a stable flow.

Then, mount an ultrasonic velocity profiler in a holder with a predetermined inclination. Apply ultrasound gel onto the probe tip. Place the probe above the flume lid with the probe tail towards the channel inlet.

Proper placement of the probing gel is crucial to obtain good velocity profiles for characterization of the background hydrodynamic conditions. Connect the probe to the acquisition module and configure the instrument for instantaneous velocity profiles. Acquire the desired number of velocity profiles.

Then, reverse the position of the probe and acquire another set of velocity profiles. Acquire velocity profiles in this way for each measuring location on the flume. Then, calculate the average velocity value for each location and determine the stream-wise and vertical velocity components.

Adjust the values the values accordingly for the medium through which the acoustic beam traveled. Identify a range of elevations at which the stream-wise velocity component profile shows a linear trend in a semi-logarithmic plot. Fit the curve to a logarithmic equation and estimate the shear velocity.

To begin the experiment, set the frame rate and resolution of two action cameras to the desired parameters. Mount the cameras on the lateral walls of the lids facing the channel bottom close enough together that the focus the areas overlap. Ensure that the flume is marked with visual reference points for known distances.

Record a short video with each camera. Based on the recordings, adjust the camera positions and orientations to ensure that the channel is in frame and the focus areas overlap. Verify that the flow through the channel is stable.

Then, begin slowly feeding the particles of interest by hand into the flume inlet at a rate of one particle every two to three seconds. We feed small amount of sediments because having few particles in the field of view is important for straightforward sediment tracking. At small discharges, however, some particle may be stuck in the focus area.

Start recording with both video cameras. Turn off the room lights to create a marker for later camera synchronization. Maintain a constant light level throughout the experiment.

Continue feeding particles into the flume for the desired experiment duration. Then, stop the cameras and remove trapped particles from the sediment bed. Repeat the experiment under other hydrodynamic conditions as needed.

To begin processing the images extracted from the videos, first apply a radial transformation to the the pixel coordinates so that the flume sides appear as straight lines. Determine the conversion the conversion factor from pixels to distance based on the bed elevation and the reference markers for known distances. Then, create a new image sequence in the fluid flow image analysis software.

Fill in the time interval between frames and the pixel to distance conversion factor for the image sequence. Select the files of interest and run the process. Generate intensity maps of a random selection of particle images from the sequence.

Identify an appropriate threshold value for the intensity of the particles of interest. Then, create a filter pipeline for the sequence. Set the filter to background removal.

Create a new particle identification algorithm using a single threshold. Fill in the particle intensity and diameter thresholds. Add the particle identification processes to the filter pipeline and then execute the processes.

Once filtering has finished, open the image view of the newly created particle record. Scroll through the frames and note the particle displacements between images. Then, create a new PTV analysis pipeline.

Create a new analysis and select distance in the costings tab. Fill in the search window position and dimensions. Add the new process to the pipeline and execute the process.

Use track reconnection to fix any interruptions in the individual particle records. Repeat this process for the second camera recording. Then, in a specialized image processing module, select the track file for both cameras and click find track properties.

Compare images from the upstream and downstream cameras to determine the coordinate shift between cameras. Fill in the coordinate shift for the downstream camera and click make reference system uniform. Fill in the bounds of the area of overlap between the images.

Remove all trajectories smaller than the length of the overlapping area. Merge the particle track databases and fill in the overlap tolerances. Then, join the tracks.

Once the process is completed, save the results and then analyze the track data to investigate the bed-load particle kinematics. The measured flow velocity profile was asymmetric, which was attributed to a difference in roughness between the sediment bed and the channel lid. The upstream and downstream cameras show 37 and 34 tracks, respectively, over 100 seconds.

After joining the data from both cameras, 59 tracks were identified in total. The longest track spanned the full observation window for a total length of approximately 1.6 meters. The tracks were analyzed to identify when the particles were in motion or at rest on the sediment bed.

The large area of observation allowed long hops to be identified, with hops of up to 600 millimeters being observed under these hydrodynamic conditions. Shorter, faster hops were found to occur most frequently. After watching this video, you should have a good understanding on how to perform sediment transport experiments of particle tracking over large areas.

While you are attempting this procedure, remember to ensure that the lighting of the experiment is good even though the images can be collected in many ways, starting from high-quality images, simplifies a lot of the following work. After it's development, this technique paved the way for researchers in the field of sediment transporter to explore a variety of indicators for process interpretation and modeling.

Read Article

Journal (Environment)

Journal (Environment)

Image-based Lagrangian Particle Tracking in Bed-load Experiments

Chapters

Transcript

Related Videos

Get cutting-edge science videos from JoVE sent straight to your inbox every month.

Your free access has ended.