Scientists are harnessing the power of artificial intelligence (AI) to quickly identify and monitor giant icebergs in satellite images. Unlike the conventional approach which requires several minutes for a human to outline a single iceberg, AI can accomplish the same task in less than 0.01 seconds, making it 10,000 times faster. This innovation is crucial to locate icebergs and monitor their shrinkage over time, as it helps quantify the amount of meltwater they release into the ocean.

The British Antarctic Survey revealed that massive ice sheets in Antarctica will continue to melt at an accelerated rate throughout the century. This will inevitably contribute to rising sea levels around the world in the coming decades. Last year, A68a, one of the largest icebergs known to scientists, thawed in the South Atlantic Ocean after drifting for five years from the Antarctic Peninsula. It is estimated to have been over 100 miles long and 30 miles wide. This melting iceberg not only released 1 trillion tons of fresh water into the ocean but also pumped nutrients into its environment, potentially altering the local ecosystem.

Identifying icebergs accurately is challenging as they, along with sea ice and clouds, all appear white in satellite images. Additionally, the shape of the Antarctic coastline can resemble icebergs. To overcome these obstacles, researchers trained a neural network for the first time to detect large icebergs in satellite images. They used data from the European Space Agency’s Sentinel-1 satellite, which is equipped with radar lenses capable of capturing the Earth’s surface regardless of cloud cover and lack of light.

The AI system achieved a 99 percent accuracy in detecting icebergs in satellite images, including correctly identifying icebergs ranging in size from 54 square kilometers to 1052 square kilometers. The research team believes that machine learning will enable scientists to monitor remote and inaccessible parts of the world in near-real time. Furthermore, the AI tool outperformed conventional automated approaches by not misconstruing individual bits of ice as a collective iceberg.

This groundbreaking research opens up possibilities for an operational application, allowing for easier observation of changes in iceberg area for several giant icebergs. By utilizing AI, scientists can gain valuable insights into the effects of iceberg shrinkage and its impact on the surrounding ecosystem. The study was published in the journal The Cryosphere.