For plastic detection, characterisation and tracking, we need rich spectral information, ideally hyperspectral data. We would like hyperspectral data up to short-wave infrared (2500 nm) and around 0.5 m ground sample distance. Neither of these characteristics are existent in current Earth observation missions, and will still be difficult to find in future Earth observation missions.

To mitigate current challenges, the MUSS2 project investigated the following:

1. Using novel spectral enhancement methods to generate simulated hyperspectral images from the multispectral images of the Copernics Sentinel-2 MultiSpectral Instrument, by using dictionary learning and spectral response function modelling;
2. Investigating newly available PRISMA full stretch hyperspectral satellite images;
3. Validating effectiveness of the synthetic Sentinel-2 hyperspectral data for marine plastic identification via simulated mixed pixels with marine plastic.

Finally, the project team reported results from spatial enhancement with hyperspectral-multispectral fusion using real targets.