We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Evaluating indices for use with Sentinel-2 to detect ghost fishing nets
Summary
Scientists tested whether satellites can spot "ghost nets", lost fishing gear that drifts in the ocean, breaking down into the microplastics that end up in seafood and drinking water. They floated a test raft of old fishing nets off the Scottish coast and used satellite images to try to detect it, finding that a specific method (comparing red and near-infrared light) worked best, though picking plastic out from the natural background was still tricky. This matters because better satellite tracking could help clean up ocean plastic before it breaks down into the tiny particles now showing up in our food and bodies.
This study evaluated the suitability of the multispectral Sentinel-2 satellite data to detect abandoned or lost fishing gear at sea, off the coastline of Scotland. These lost “ghost nets” are often produced using polyamide and/or polypropylene-based plastics, and it is estimated that these comprise 46% of the “Great Pacific Garbage Patch”. Current methods of plastic detection and clean-up include passive detection and speculative indirect modelling methods relying on ground observations from coastlines and ocean current surface transport modelling. Satellite remote sensing can provide a more direct method to detect ghost nets, with multispectral platforms differentiating plastic materials apart from natural environments. In this research, a 10m² raft was constructed out of disused polyamide-nylon fishing nets and polypropylene piping collected locally, and floated on the surface in a near-shore environment of North-East Scotland. This was done three times between the 6th and 13th June 2021, and Sentinel-2 acquisitions of the raft were captured in a controlled environment. The image acquisitions were pre-processed using Sen2Cor and ACOLITE to correct for atmospheric interference, as well as Continuum Removal. Multiple plastic detection indices were evaluated, including the Floating Debris Index, Plastic Index, and novel Band Ratio/Depth/Peak Indices. Clear identification of plastic netting between background samples was difficult, however spectral comparison between representative samples showed differences that could be used as identifiers for plastic netting detection. The novel Band Ratio Index using Red and Near-Infrared Spectral Bands provided better results of detection on the Scottish coastal environment against representative samples. This zip file contains the originally sourced spectral analysis data, the Sentinel-2 aqcuisition images (produced and provided by ESA), as well as the spatial data, spectral processing in ENVI and ACOLITE softwares, and additional notes on the workflow to produce it.