Improving MPA Management

Seminar: Imaging in Marine Science

Imaging in Marine Science

In the latest of the Bertarelli Foundation’s marine science seminar series, Prof. Heather Koldewey introduces four speakers who are all using imaging technology for marine science.

Advances in technology have allowed scientists to capture the underwater world in new and innovative ways, revealing insights in to life in the big blue and providing data to help us better protect the ocean and the life within. This seminar will explore how different imaging techniques are applied to study the ocean and help provide a picture of what the underwater world is hiding.

Speakers:

  • Joanna Harris – Manta Trust and University of Plymouth
  • Dr Dan Bayley – UCL
  • Dr Greg Asner – Arizona State University and Allen Coral Atlas
  • Dr Jyotika Virmani – Schmidt Ocean Institute

You can watch a recording of the seminar below:

Below you can find answers to questions we didn’t have time to address during the live seminar.

If overcrowding from tourists is having negative impacts on manta rays, how are you encouraging responsible citizen science photography?

Joanna: Encouraging responsible citizen science photography is extremely important. The Manta Trust has engaged in extensive research and provided key recommendations that aim to ensure that the manta ray tourism industry remains sustainable and non-detrimental to the animals’ natural behaviour (Murray et al., 2019). The outcome of this research includes a 10-step guide on how to sustainably interact with manta rays (https://swimwithmantas.org/).

What is the timescale for taking all of the photos to make these 4D maps?

Dan & Andy: We have been monitoring the reefs around the Chagos Archipelago annually using the 3D SfM mapping technique from 2015-2019. To make a ~20 x 20 m HD map takes about an hour for a diver pair to set up the site and collect pictures. The processing time depends on the computer power, number of images and the image sizes, but can take from hours to a day for a site this size. If interested in more detail on this method, we recently wrote a ‘how to’ guide: https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13476

On what scale have you developed the 4D maps? Are they limited to localised in-situ images?

Dan & Andy: Our projects have tended to be local, and the technique is usually applied over 10s to 100s of square metres, however it is possible to scale projects to larger areas. We are developing the use of an autonomous surface drone to cover larger regions of reef, and aerial drones commonly use this type of photogrammetry to survey multiple km2. It is worth noting that typically to scale up projects, the height from the ground or reef needs to increase, and so the overall resolution decreases.

With these types of coral reef mapping (4D map analysis), would this tend to specialize data gathering and analysis for researchers? How can this be made more accessible to local managers or researchers who may have limited resources and skills?

Dan & Andy: In many ways, this technique democratises small-scale spatial data gathering compared to other methods. It is possible to quantitatively map several hundred square metres of seabed using a single camera, in-situ scales, a mid-range computer, and the requisite software. Total cost buying from scratch would be <£5k (though better cameras and computers will increase resolution and decrease processing time). Mapping the same area with a multibeam sonar or lidar system would typically cost an order of magnitude more for the sensor alone, and result in poorer resolution.

The camera types you use can range from a GoPro to a high-end SLR. While you can get more detail and more control from an SLR, action cameras can produce very good results, particularly in well-lit clear water. The image capture technique just needs to be methodical to ensure sufficient coverage, and again this is detailed in the ‘how to’ guide above. To analyse the models quantitatively, there are a range of paid and open-source software types available. While the processing aspect is straight forward and can be mostly automated, the analysis needs training. However, the models can be analysed in commonly used software tools such as ArcGIS, or using open-source software, which are typically well documented. As a final point, if you wanted to just use the technique for display/educational purposes, there are programmes available which can automatically process your online, mostly at no cost.