Job Background

This job was carried out in April 2018 for Central Plains Water Ltd with the scope being to conduct a bathymetric survey of the intake canal to assist with the building of a hydraulic model.

The aerial survey was carried out using a DJI Phantom 3 UAV and depth calibration data was collected by staff from NIWA using a remote controlled Jet Boat equipped with a depth sounder a (JetTec R/C boat with Tritech PA500 Echosounder).

 
Equipment used for obtaining depth calibration data

Equipment used for obtaining depth calibration data

Flyover video showcasing the final DEM overlaid onto the aerial imagery.

Ground control points were placed on the edges of the canal to allow an accurate DEM of the above water ground surface to also be generated

A o.1m resolution orthophoto was generated and the Red, Green and Blue (RGB) values were extracted from the image at each depth point.

Machine learning techniques were then applied in order to find a relationship between RGB and depth.

Results

A seamless DEM of both the above and below water ground levels was created by combining the optical bathymetry results with the above ground survey from the drone using photogrammetric techniques.

Overall results showed an excellent match with the calibration data giving a mean absolute error of 0.16 m and a R2 Goodness of Fit of 0.9.

A flyover video showing a visualisation of the final DEM can be viewed above as well as a comparison between the surveyed bed level and the final DEM is presented below.

Location of surveyed profile location used for validation

Location of surveyed profile location used for validation

Comparison of surveyed bed level profile with results of optical bathymetry technique

Comparison of surveyed bed level profile with results of optical bathymetry technique

Results comparison with surveyed cross sections from 2016

Results comparison with surveyed cross sections from 2016

As a further independent validation, comparison was made with physical survey which was carried out in 2016. Minor changes in bed level are to be expected from 2016, however a reasonable fit with the data was observed.

We would love to have more opportunities to explore and develop these techniques further. If you are interested in learning more about these techniques or trialing them at your site, please get in contact with Matthew Gardner.