The majority of plastic polluting our oceans does not float on the surface, but under the waves, making it difficult to track by satellite. But a new system of search algorithms powered by supercomputers could be a game-changer. With adjustments, it could become a valuable tool in the fight against this global scourge. A study published in Nature Communications details how a machine learning program trained on 300,000 multispectral and infrared images made it possible to identify accumulations of plastic waste over 94.5 km² in the Mediterranean.
“Windows”, indicators of pollution hot spots
The researchers used a database of satellite images to identify windrows, floating filaments made of waste and plastic formed by ocean currents and wind. Although often ephemeral, these structures can extend for kilometers before dispersing into the ocean.
Although representing only a tiny part of marine litter, windrows could serve as indicators of pollution hotspots. Between July 2015 and September 2021, the team detected and tracked 14,374 windrows, an exponential increase in documented structures. Many have been verified by ships in the region.
Mapping marine litter accumulations
Manuel Arias, co-director of the project, explains: “Until now, looking for waste aggregates several meters in diameter on the surface of the ocean was like looking for needles in a haystack, because the formation of these trails require the presence of a large quantity of waste and little wind to avoid their dispersion. »
Most filaments were 1 km long, but 27 documented windrows reached 10 to 23 km. In total, pollution is estimated at 94.5 km², the equivalent of around 7,500 football fields.
Understanding the influences on waste accumulation
Analyzing patterns of marine pollution allowed the team to infer several factors influencing where and when ocean waste accumulates, such as nearby population density, geography and precipitation.
A region frequently subject to torrential rains, for example, dumps garbage into the water more often, while desert coastal areas contribute much less. These hotspots are mainly located within 15 km of a coast and frequently return to shore in the following weeks or months.
Towards better management of marine waste
Researchers believe this new system could help cleanup projects identify areas where ocean plastic pollution is concentrated and even predict where and when it might disperse. They also believe the program can be expanded to track other issues, such as ship cargo losses, oil spills and search and rescue situations.
“Marine litter reflects the failures of our civilization to manage waste,” the team writes in its study. “Satellite observations are one of the best hopes for large-scale monitoring. »
This technological advance opens up new perspectives in the fight against plastic pollution of the oceans. By providing a better understanding of where and when waste accumulates, it could help optimize cleaning and prevention efforts. Combined with other measures such as reducing plastic production and improving waste management, this innovation could be one step closer to cleaner, healthier oceans.
SRQ Backlot