Find Floating Debris on Water from Satellite

No, and no tool using free satellite imagery honestly can. What the index measures is material floating on water that reflects more near-infrared light than open water does. Plastic behaves that way. So do seaweed, foam, driftwood, wave caps and sun glint. The published study that separated plastic from those materials used a classifier trained on plastic targets deliberately placed at a known location, and even then worked on patches tens of metres across. Without that ground truth, a candidate is somewhere worth looking, not a finding.

The Floating Debris Index from Biermann and colleagues, 2020, which builds on the Floating Algae Index from Hu, 2009. It compares the near-infrared reading against the baseline it would have if nothing were floating there, using red edge and short-wave infrared to set that baseline. The Plastic Index of Themistocleous and colleagues, 2020, is offered alongside for comparison.

A Sentinel-2 pixel is ten metres, so a hundred square metres of surface. Debris has to be gathered into a patch that fills a good share of one before the signal rises above the water around it. Scattered litter, and anything below the surface, is invisible. This is a screening tool for large accumulations, not a litter survey.

The index assumes the background is water, which absorbs near-infrared strongly. Over land that assumption fails and the number means nothing, so everything outside water is removed before any candidate is counted.

Not loose litter, no. At ten metres a dump site is a handful of pixels of mixed material, and no simple index separates plastic from soil, metal and rubble. Plastic sheeting over fields and greenhouses is a different case and genuinely is detectable, because it covers whole fields with one material.

Because there is no universal value. The index shifts with sun angle, water clarity and sediment, so the tool sets a starting point from the water in your own area and lets you move it. If a patch appears and disappears with a small nudge, that is telling you the signal is weak.

Report it as a place worth checking, with the date and the index named, and say the material is unidentified. Presenting a candidate as detected plastic would be wrong, and the first person to visit the site and find seaweed would be right to say so.