ANTARCTICA: Scientists from the University of Leeds have introduced an Artificial Intelligence (AI) system capable of mapping extensive Antarctic icebergs in satellite images with unparalleled speed and accuracy. The neural network-based AI completes the mapping task in a mere 0.01 seconds, a revolutionary improvement compared to the time-consuming manual methods previously employed.
The lead author of the study, Anne Braakmann-Folgmann, who conducted her research during her PhD tenure at the University of Leeds and currently serves at the Arctic University of Norway in Tromso, underlined the pivotal role of large icebergs in the Antarctic environment. “Giant icebergs impact ocean physics, chemistry, biology, and maritime operations,” Braakmann-Folgmann emphasized. Finding icebergs and keeping an eye on their extent is essential for calculating the amount of meltwater they discharge into the ocean.
The innovative approach leverages data from the Copernicus Sentinel-1 radar mission, providing images of icebergs regardless of cloud cover or lack of daylight. Traditionally, distinguishing icebergs from sea ice or the coastline in complex surroundings posed a challenge in radar images. However, the new neural network approach excels at mapping iceberg extent even under these challenging conditions. The system continually refines its predictions during the training process, adjusting parameters based on the disparity between manually derived outlines and predicted results.
The algorithm’s robustness was tested on seven icebergs, varying in size from 54 sq km to 1052 sq km, equivalent to the areas of the city of Bern in Switzerland and Hong Kong, respectively. The results showcased an impressive accuracy rate of 99%, signifying a significant leap forward in polar research capabilities.
According to Braakmann-Folgmann, who emphasized the AI system’s operational implications, being able to map iceberg extent automatically with improved speed and accuracy will make it easier for us to monitor changes in the area of several giant icebergs and pave the way for an operational application.
Mark Drinkwater from the European Space Agency (ESA) commended the research team on their innovative machine-learning approach, acknowledging that it automates what would otherwise be a manual and labor-intensive task of locating and reporting iceberg extent.
Also Read: Antarctica’s Ice Shelves Rapidly Shrinking: A Dire Consequence of Climate Change
