Seagrass meadows are crucial marine habitats. They store large amounts of carbon dioxide and provide a nursery for many marine species, supporting biodiversity. Unfortunately, these habitats are declining worldwide. To protect and restore them, nonprofit organizations and governments need reliable monitoring methods to determine what works and what doesn’t.
Period: August 2024
Location: Flensborg Fjord, Denmark
Key takeaways
In this technology demonstration, we have:
Collected 4K images across designated areas, spanning approximately 1500 m², in approximately 3 hours, showing that our method can scale efficiently.
Provided location information with 2 m accuracy, which can help monitor growth or decline of individual seagrass patches over longer time periods.
Achieved significant reduction of the time needed for processing compared to traditional methods. Because of our high quality images and geo-referencing, processing of the LobsterMap was completely automatic.
Challenge
Current methods for monitoring seagrass—like using divers, drop/drift cams, or towed video—are not practical for large areas.
Divers have to take frequent breaks, because it is demanding work. Drop cams provide sparse snapshots of the seabed, making it impossible to determine the degree of patchiness of the seagrass beds and therefore skewing any coverage estimate. Towed video is arguably the most suitable for large areas, but often fails to capture smaller details such as seagrass health, epiphyte cover, or other notable biological observations.
Furthermore, with any of these approaches, laborious processes are required to geo-reference the data. Finally, with these methods, the images and observations have to be processed manually. This is because the illumination, distance and angle to the seabed, etc, are irregular and thus difficult to automatically analyze.
In summary, these approaches have three pain points: they take too much time, often lack precise location data needed for accurate comparisons, and produce inconsistent results that are hard to process.
An example of the 'drift cam' method: the user suspends a camera (in this case a GoPro) into the water and lets the current or wind move the vessel. It is a simple method that gives a clear indication of the presence or absence of seagrass. But the poor quality does not allow for much further analysis, and there's no way of knowing exactly where the data was recorded without an external GPS recorder. This video was taken by the Lobster team while the Scout 2 was surveying to compare the results.
Solution
The team addressed these challenges by conducting a survey with our advanced underwater drone, the Scout 2. Our drone collected detailed and accurate data across a large area with minimal human effort. The data collected was processed into a LobsterMap for easy access by end-users.
Logistics: the team arrived on site with our company van. It took about half an hour to unpack the equipment. Using waders, the team (two people) carefully put the Scout 2 in the water.
Preparation: the day before, one team member had pre-programmed the Scout’s missions, telling it exactly where to go and what survey parameters to used. A survey altitude of 1.0m, and 0.4 m/s cruise speed were used. These are fairly typical settings for a high resolution survey.
Survey process: using the Lobster Remote, the Scout is activated and the missions are started. The operator has to do nothing except press a button. Once the Scout finishes a mission, the operator starts the next mission, until everything is completed. The water depth varied between 1.8 m and 8 m, and visibility was between 3 and 4 m.
Live feed: During this demonstration, the team had an antenna buoy attached to the Scout so they could have a live feed of the images it produced, and check the data quality. The team also wanted to check whether the Scout would get entangled in the seagrass, as it was flying so close to the beds. This was not a problem.
Survey data: in just over 3 hours, the Scout completed seven missions. Four 100m transects were completed perpendicular to shore, to find the shallowest and deepest point of seagrass habitat. Three 20 x 20 m squares were surveyed at 100% coverage to investigate specific areas: one with dense beds, one with patchy beds, and one with barely any seagrass at all. This resulted in a total of 400 GB of images.
Processing: after the survey, the Scout was connected to Lobster’s server to offload the data and process overnight. The LobsterMaps were then available for inspection.
Here you can see some highlights of the LobsterMaps that were created:
You can scroll through the seabed to discover for yourself what lies beneath the surface.
A closeup of the dense seagrass bed shows luscious green blades.
Zooming in even further, all the tiny details become clear.
The images are geo-referenced with an accuracy of 2 m. This allows for surveys to be conducted at the same location across different seasons or years, enabling direct comparisons of the habitat over time. The ease of conducting surveys with this method also lowers the cost and time required to collect high-quality data, making it more accessible for regular monitoring.
Fieldwork is a big part of what we do at Lobster Robotics. We’ve seen too many companies build a product, ship it out, and then find out months or years later that some parts don’t work as expected—or at all. We believe in getting hands-on with our technology to keep improving it, making sure it’s practical and user-friendly, especially when it comes to complex tech like underwater robotics.
Impact
Our method for monitoring seagrass meadows is designed to protect these important ecosystems. It's cost-effective, accurate, and easy to scale, making it a great tool for long-term research and conservation. Plus, our tech can really make a difference in understanding how human activities, like offshore windpark construction, impact marine habitats. For example, when export cables from these projects cut through areas like seagrass beds, our detailed LobsterMaps can help developers reroute cables and assess any damage, so they can take the right action.
Behind the scenes
