Underwater vegetation mapping in crystal clear Ewens Ponds (SA)

This month we have been undertaking some vegetation surveys at Ewens Ponds in South Australia, in conjunction with National Parks and Wildlife South Australia. Ewens Ponds is an iconic diving location near Port Macdonnell, on the Limestone Coast. It is renowned for its crystal clear waters and a number of aquatic plant species that grow differently here. Many of these species are normally found growing in shallow water or in wet soils around the fringes of wetlands but in Ewens Ponds, they are submerged up to 5 metres below the water surface!

We undertook some surveys back in 2013, so this current study should provide some insights into how the underwater plant life has changed over the past decade. One thing we are particularly keen to do is come up with a method which is easily repeatable, so it can be done at lower cost and thus more frequently. Motivated by some techniques used on coral reefs, I have come up with a device which allows us to capture photos along a transect and link these to accurate GPS coordinates. The result is a spatially accurate photo strip, where we can map bands of different vegetation. Combining this with physical identification of the species present, we can actually map the distribution and abundance of species and, with more frequent surveys, accurately assess change over time.

What I have rigged up is a frame which fits to a body board and supports a differential GPS receiver on the skyward side and a GoPro camera on the submerged side. We can then sync GPS and camera time to link centimeter accurate GPS coordinates with an image (the camera takes a photo every second). It’s very stable and can be pushed along at a standard speed by someone in the water, or towed behind a kayak. Aside from the GPS and camera, the whole thing can be built from hardware store items or, in my case, things lying around in the shed.

Body board with GPS receiver on top, and camera on bottom being pushed along the transect (photo Claire Harding NPWSA)

The transects have been permanently marked at each end so future surveys can just run the camera from start to finish and, with little bit of computer processing, we will have repeatable photo strips and vegetation maps. An example of a stitched image is provided below.

A stitched image of a transect (note 1m quadrat for scale)

By running a transect of image capturing up and back along the transect we can also create an orthomosaic (a detailed map made by ‘stitching’ together high-res aerial images) which can be interacted with as a 3D model and zoomed in at each camera shot to give a higher resolution insight, capable of making out individual plants.

Three dimensional textured model and zoomed image from photo dataset

Of course the idea is not new and applies similar principles to those used in drone surveying, but off the shelf devices for underwater work can be expensive. Hopefully this will make it more accessible where funding is limited. In terms of what this data tells us about these ponds, that’s a work in progress and a story for another day, so watch this space!

Lachlan Farrington