Recent works on Sentinel-2 showed that the use of MSI sensor provides not enough spectral information to discriminate with sufficient certainty the presence of marine litter and to distinguish the latter from other surface features. PRISMA, with its 240 spectral bands, has the capacity to acquire images at higher spectral resolution, with a spatial resolution of 30m. The increased spectral resolution will allow to better discriminate the characteristics of a feature floating at or near the sea surface, thanks to: -Improved separation of spectral signatures from different features/materials (unmixing) -Reduction of other “noisy” elements (e.g. better atmospheric correction) Two different approaches will be considered for unmixing and detection: the 1° based on Signatures of Main Marine Litter’ Materials, the 2° on Machine Learning. In both cases calibration information will be collected by spectral libraries, laboratory measurements, crowdsourcing data and information from local authorities. Machine learning approach will benefit from the fusion of PRISMA, Sentinel-2 and VHR satellite data (Planet).
The project will design and test a service for the marine litter detection, identification, quantification and tracking by integrating remote sensing data and in situ measurements. AIS ship locations will be used to reduce false detections. The use of hydrodynamic models, to support the marine litter detection and tracking, will be considered. The demonstration case will be tested in Puglia Region and in one of the biggest world plastic island according to PRISMA opportunities. In-situ measurements will be provided by ARPA Puglia and by “Puglia Fishing For Litter”. The latter is a Puglia Region Authority’s project that aims to encourage the fishing boats to collect plastic debris during the fisheries activities for monitoring plastic waste quantities and types. During the trawl fishing between November 2018 and March 2019, 1200 kg of plastic debris have been recovered in Puglia.
