The alleged effects of climate change are reportedly responsible for coral reef bleaching, which is having catastrophic impacts on local marine species. Hoping to reverse this supposed trend, a new method of artificial intelligence (AI) cataloguing designed to identify the geographic regions where coral still thrives, aims to save some of the world’s densest and varies aquatic ecosystems.
Between ethical and economic concerns, there are plenty of reasons why saving our world’s coral reef systems are in our best interest. Aside from providing habitat to a quarter of marine species, reefs generate $375 billion (USD) in revenue to the world economy, along with food security to over 500 million people. Without our ocean’s reefs, researchers believe countless species and an entire ocean fishing industry would cease to exist. Considering how there’s only so much money, time, and resources we can devote to mitigating any damage already done, an international association of researchers hope that AI can pick up the remaining slack.
This solution has involved a team of researchers deploying underwater scooters with 360-degree cameras to photograph 1,487 square miles of reef off the Sulawesi Island coast in Indonesia. These images were then fed into a form of deep learning AI that was taught over the course of 400-600 images to identify types of coral and other reef invertebrates to evaluate the region’s ecological health. Utilizing AI to rapidly analyze photographs of coral has significantly improved the efficiency of researchers’ efforts, since it takes this technology just seconds to do what would normally take 10-15 minutes for regular coral reef scientists.
The AI technology learns similarly to the human brain, weighing up multiple minute decisions about what it’s visualizing, until it forms a picture and is confident about making an identification. Researchers are also using a custom, iterative clustering algorithm to identify coral reefs across the glove that seem most likely to benefit from conservation resources. The formula is based off 30 metrics known to impact coral reef ecology that are broadly divided into categories like historical activity, thermal conditions, cyclone wave damage, and coral larvae behavior.
The research was made possible by generous donations from various international governmental entities, along with private companies and foundations. Researchers hope these AI techniques will be further refined to help manage coral reefs on the more local level and several ecologically significant sites. Local versions of the global study researchers believe they would benefit from this data that isn’t uniformly available for reefs internationally. Information about ocean chemistry, the “adaptive capacity” of local reefs to withstand climate change or other stress on their systems, or particulars of local economic dependence on these coral reefs.
