Advancing Floating Macroplastic Detection from Space Using Experimental Hyperspectral Imagery

Airborne and spaceborne remote sensing (RS) collecting hyperspectral imagery provides unprecedented opportunities for the detection monitoring of floating riverine marine plastic debris. However, a major challenge in application RS techniques is lack fundamental understanding spectral signatures water-borne Recent work has emphasised case open-access reflectance reference libraries commonly used polymer items. In this paper, we present analyse high-resolution image database unique mix 40 virgin macroplastic items vegetation. Our double camera setup covered visible to shortwave infrared (VIS-SWIR) range from 400 1700 nm darkroom experiment with controlled illumination. The cameras scanned samples water captured images 336 bands. Using resulting spectra 1.89 million pixels linear discriminant analyses (LDA), determined importance each band discriminating between mixed debris, vegetation plastics. absorption peaks plastics (1215 nm, 1410 nm) (710 1450 are associated high LDA weights. We then compared Sentinel-2 Worldview-3 satellite bands these outcomes identified 12 overlap important wavelengths discrimination classes. Lastly, Normalised Vegetation Difference Index (NDVI) Floating Debris (FDI) were calculated determine why they work, how could potentially be improved. These findings enhance existing efforts pollution, as well form baseline design future multispectral systems.