Monitoring of marine litter through remote sensing and hydrodynamic Lagrangian models

Resumen

Marine pollution is a growing global issue. Following the proof of the significant impacts of marine debris on the ecosystem worldwide, in the last decades, the scientific community has focused the efforts on monitoring and detection to assess more effective management strategies. The first efforts were focused on the quantification and characterisation of pollutants, mostly performing in-situ campaigns both in the beach and sea environments. Up to now, the characterisation of the debris fractions has shown that the most commonly portion is represented by plastic items. These polymers are characterised by long residence times and the predisposition to be degraded in smaller items, causing even greater impact on several marine fauna species and, consequently, on humans. Recently, due to the difficulties to perform intensive in-situ sampling campaigns especially over large areas, the growing of new observation technologies has driven the scientific community to develop remote sensing techniques for marine litter detection. Several applications carried out using unmanned aerial vehicles and satellite data highlighted that there is a lack of knowledge about the spectral response of marine litter and plastic items. Additionally, their detection could be also more difficult by considering their temporal evolution due to degradation processes in the marine environment. The issue concerns marine pollution is even worse considering that once at the sea marine litter debris could be drifted by the sea currents to areas which can be very far from the source points. Thus, there is high uncertainty about the areas which can be more impacted by these kinds of pollution both at the sea (accumulation areas) and at the coast (beaching). In this framework, the existing monitoring actions (i.e. in-situ or remote detection) could be not enough to study and thus deal with the marine litter pollution problem. In this context, this thesis has been designed to test, and then to provide new insights into the marine litter issue through an integrated approach which include the in-situ monitoring campaigns, the application of remote sensing techniques following a bottom-up approach, the setup of numerical Lagrangian models to determine areas were these kinds of pollutions could be severe (accumulation and beaching areas). A monitoring campaign conducted in two urban beaches, with different management, allowed to quantify and characterise the pollution status of those areas. Results show the need to manage the beaches, organising frequently beach cleaning activities. Some of the samples collected during the monitoring, have been spectrally characterised via a Laboratory experiment, and the wavelengths/bands useful for the detection as well as the best sensors to use, were identified. Some bands in the visible, red-edge, and in some infrared bands, appear effective for the beach litter detection. WorldView-3 resulted the sensor more useful for this purpose among the nowadays operating satellite platforms. Furthermore, the interconnection between sea and beach environments was studied applying Lagrangian hydrodynamic models able to identify accumulation and beaching area. Outcomes of the models have confirmed as the marine litter is a transboundary issue that only a joint action of the countries can mitigate. Finally, the potentiality of hyperspectral satellite data was tested analysing PRISMA images evaluating the capability to detect an artificial floating target deployed at the Aegean Sea. Results showed that satellite hyperspectral images represent a useful tool for marine litter detection, although their spatial resolution limits their applicability