UT’s remote sensing and drone experts are collaborating with the Australian Institute for Marine Science (AIMS) to experiment little drones, specialized hyperspectral cameras and machine learning can study the Great Barrier Reef easily and efficiently, more than the manned aircraft.
The leader of the QUT project, Associate Professor Felipe Gonzalez pointed out that the team studied three reefs from 60m in the air, and the coral bleaching was recorded by the AIMS divers from below the water. Prof. Gonzalez said that taking data from the air and comparing it to that gotten from under the sea, will nurture the system on how to study and classify bleaching levels.
The captured images give a resolution of around 9 centimeters which is good enough to spot separate coral bleaching. The unmanned aircraft can also cover a larger area in a day and can be used quickly. While it’s true that reefs have been studied with hyperspectral images and satellite, drones will be better with their resistance to cloud covering and lower deployment cost.
The new aerial system features miniaturized hyperspectral cameras which grew costly and could be handled by manned aircraft. One fascinating thing about drones is that the mounted hyperspectral cameras cover about 270 bands in the visible and almost infrared regions of the spectrum. Likewise, the drone-mounted camera provides way many details than the ordinary eye can see, and at a very high resolution.
Apparently, scientists can work better with more details of any bleaching event, and drones with hyperspectral cameras are apt for quick and detailed response during and after bleaching cases. The Great Barrier Reef houses about 3,000 reefs spread over 2,300 kilometers, causing an expensive and hassle-full survey with the traditional methods.
Professor Gonzalez and his team are working on artificial intelligence that classifies the various spectral patterns for things within the coverage. Every object displays a unique signature via the hyperspectral camera, like a biometric. The signature for coral is different from that of sand, and so on. Most importantly, a distinct coral colony will appear as a separate signature with changing bleaching levels, making it easy to study those individual changes with time.
Professor Gonzalez added that the more fingerprints available, the more accurate and efficient the system functions.
Gonzalez and his group from QUT conducted a drone study on Ningaloo Reef in Western Australia recently, and their innovations spread across marine robots, agricultural robots and using UAVs to find and study gas leaks and pests.