While the knowledge associated with the marine litter challenge is increasing rapidly, there are still many unknowns about what satellite technologies can readily bring to the solution. The great variety of scenarios where plastic is found in aquatic ecosystems, whether it originates in land or at sea, of different sizes and densities, distributed in beaches, rivers and lakes, floating on the surface, submerged along the water column and in the sea bed, makes it ubiquitous and represents complex mapping requirements.

There is no single technology able to map marine litter alone. Nonetheless, a series of remote sensing technologies exist today that are able to tackle the mapping of marine litter, directly and indirectly. High-resolution optical technologies can be used for the direct detection of high densities of plastic in cloud-free areas. The correlation between water quality parameters and microplastics has been found which motivates indirect assessment. Plastics show significant features in the 0.4–2.8 μm region of the spectrum, where most hyperspectral systems operate, and Fourier Infrared Spectroscopy has been suggested as a laboratory technique for the classification of plastics according to their composition, while Raman spectroscopy can potentially measure debris which is partially submerged in shallow waters. Within microwave technologies, SAR sensors are capable of detecting surfactants such as biogenic films, and target-like derelict fishing gear and bigger items, by polarimetry and interferometry; like SAR, altimetry could also detect marine litter as it might have an effect on reflectivity (sigma-0 or roughness). Radar technology for spaceborne remote sensing of the ocean has indisputable benefits since it provides an all-weather, all- day (independent of the solar light) wide-swath monitoring of the sea. The continuous advances in SAR coherent systems, providing high-resolution two-dimensional reflectivity images have demonstrated the suitability of this technology for marine monitoring and surveillance. The high degree of flexibility of the spaceborne SAR operation and its wide swath high-resolution imaging capabilities represents a unique and very powerful tool for ocean monitoring. Nevertheless, the particularities of marine debris, especially due to the wide variety of material types, dimensions and interaction with the background ocean, makes its detection exploiting SAR technology very challenging. The randomness of the supporting media, where the litter is enclosed, adds complexity to the detection and identification task.

A multi-sensor approach is proposed for the mapping of marine litter in a broad sense, according to the characterization of different scenarios, with particular emphasis on the potential contribution from radar techonlogies. In order to delineate pathways, and provide information over spatio-temporal gaps in the remote sensing methodologies, the use of ocean models that explicitly resolve the small-scale dynamics of the upper ocean, and backward computations where ground-truth data exist, are proposed as part of the multi-sensor methodology.

Marine litter is a global problem affecting all the oceans of the world. Millions of plastics end up in the seas affecting the marine ecosystem. Several initiatives have been planned from global players towards detection, monitoring and cleaning. State of the art techniques is needed for the detection and quantification of the marine plastics in the sea water. Unmanned Aerial Systems (drones) and satellites can be used in this direction. Already, the scientific community and space agencies are working towards the specifications of sensors detecting and quantifying marine litter.
Here, we present a test project of detecting artificial plastic targets on the sea surface using Unmanned Arial Systems (drones) and satellite images technology. This project designed to examine the ability of marine litter detection from the European satellites Sentinel-2 and Sentinel-1. Also, drones used to detect and quantify the volume of the litter on the sea surface with dedicated cameras. Three payloads used on drones for marine litter mapping: RGB, multi-spectral and thermal cameras. Inter-comparison between data from satellites and drones released the advantages and disadvantages of the detecting systems.
We created three plastic “targets”, 10 x 10 m wide, containing: a) 3700 plastic bottles, b) 138 plastic bags and c) 200 sqm fishing net. The targets’ frame was 20 inch PVC pipe and was covered with plastic net. The containing targets were attached to the net. Targets created from the Marine Remote Sensing Group (https://mrsg.aegean.gr), Department of Marine Studies, University of the Aegean. The experiment was devoted to the Word Environment Day, 5th June.
Image analysis and image processing algorithms used to evaluate the ability of satellites to detect marine litter on the sea surface. Targets were clearly detected in the Sentinel-2 image and spectral signatures were calculated. Furthermore, the percentage coverage of plastic per pixel was examined. Targets were not visible in the Sentinel-1 image due to sea state condition and spatial restraints. This experiment proved the usefulness of drone technology in marine sector since: i) acquisition can be made in remote areas with difficult access, ii) maps can be created at low cost and high resolution (typically a few centimetres per pixel), iii) surface models can be generated at true scale for measuring distance, thickness and volumes of marine litters and v) marine litter features can be observed from the most favourable point of view.

Microplastics, defined as particles < 5 mm, form a considerable portion of the plastic pollution in both inland waters and the open ocean. This pollution has increasingly received public attention following discoveries that these particles can be ingested by planktonic animals. Ingested particles can not only induce direct negative survival effects, but also be potential introduction vectors into the base of the food chain of accumulated persistent organic pollutants (POPs) or carcinogenic plastic additives. Microplastic research has mainly concentrated on open seas, while riverine plumes and coastal areas remain largely unexplored despite their hypothesized importance as both source and collection area of microplastic pollution. This work models coastal accumulation along the Adriatic coastline of microplastic particles (1-5 mm) emitted by a major river in northern Italy, the Po River, over 1.5 years. We posit that river-induced microplastic accumulation on adjacent coasts can be predicted using (1) hydrodynamic-based and (2) remote sensing-based modelling. Model accumulation maps were validated against in situ sampling at nine beaches (analyzed particle size range: 1-5 mm), with sediment microplastic concentrations up to 78 particles/kg (dry weight). Hydrodynamic modelling revealed that discharged particle amount is only semi-coupled to beaching rates, which are strongly river mouth dependent and occur within the first ten days after discharge. Particles which do not beach within this initial time period (over 80% of all modelled particles emitted by the Po River) were transported away from the Po Delta coastline and remained offshore. Sentinel-2 and Landsat-8 based remote sensing modelling presented snapshots of surface river plume form at a finer spatial resolution over a larger area than computationally feasible with existing ocean current models. This model was found to better capture river mouth relative strength, and indicated existence of two additional river mouths that were not considered influential prior to the modelling exercises. Accumulation patterns between the two models were compared and found to be in large part consistent with one another. Validation against in situ beach sampling is discussed. Models for identifying marine debris sources and forecasting accumulation areas have been put forward as a method to reduce the cost and optimize remediation activity efforts. Arguments are presented for development of an assimilated modelling system to combine the strengths of the hydrodynamic and remote sensing model methodologies. This would lay groundwork for future development of an operational monitoring system to assess microplastic pollution being emitted by a major river and its distribution along adjacent coastlines as well as into the open ocean.

On August 31, 2016 the Typhoon Lionrock landed on Primorsky Krai, Russia causing heavy rain fall around the border between the Russia and North Korea. The rain caused flooding along the Tumen River, the border between North Korea and Russia. Several days after the flooding, debris consisted with logs and household garbage were found along the East coast of Korea. The debris, of course, had negative impact on the tourism and coastal eco system. In addition to the debris landed on the East coast of Korea, there must be others drifting toward other areas or floating in the ocean. Logs floating on coastal water could be hazardous to marine traffic, and tracing the fate of the debris from the river is an important task. A Lagrangian particle model utilizing ocean circulation model output is commonly used to trace marine litter. The debris due to the typhoon is episodic and more focused approached is required. In addition, to cope with potential danger posed by large flood debris near-real time tracking is indispensable. For real time monitoring satellites are very useful. The size of less than a few meters at the most and it is too small to be identified using a satellite whose resolution is usually a few hundred meters or grater. Flooded water, however, could contain a large amount of sediment and could have distinct ocean color. By analyzing data from the Geostationary Ocean Color Imager (GOCI) that provides ocean color images of about 500 m resolution, it was possible to trace the debris from the flooding. The trajectories of the debris from Lagrangian particle tracking modelling and the patterns of anomalous ocean color are comparable along the East Coast Korea. The transport of the debris is due to the North Korea Cold Current that flows southward along the coast of Korea. The Lagrangian particle tracking model suggests that there could be debris in offshore area, but this offshore distribution cannot be confirmed from the GOCI images. This is because in offshore area the concentration of debris is low. Typhoons hit Southeast Asia every summer causing large scale flooding, and typhoon induced marine debris must occur every year. The study we present here is for the Lionrock, but this could apply to other cases.

Marine Litter (ML) is a major environmental issue requiring deeper insights from the scientific community for policy makers and enforcers to manage and mitigate these contaminants (GESAMP WG40, R&S93; Kershal et al, 2015). Up to date, data collection for ML applications at the oceans and coastal areas is very limited, and only a considerably reduced extension of the surface of the ocean has been actually surveyed by means of in-situ sampling (A. Cozar, personal communication, Minutes from SCOR “Flotsam” Working Group). Moreover, this sampling is not usually performed routinely at time scales representative of the ML dynamics. Recent studies have proven how important a good sampling is to properly evaluate the ML budget found at our oceans (Lebreton et al, 2018).

In the last years, the possible use of Earth Observation (EO) techniques to detect, quantify and track ML has gathered the interest of the scientific community (Workshop on Mission Concepts for Marine Debris Sensing, Honolulu, 2016; Maximenko et al, 2016, Maximenko et al, 2018). Remote sensing, and particularly Earth Observation, can bring the synoptic view, both in time and space, that is required to better understand the extension of the ML issue, as well as providing a tool for monitoring ML, henceforth supporting environmental agencies and regulatory organisms. Although, given the heterogeneity of ML -in size, composition and dynamic behaviour-, a remote sensing solution is likely to provide only partial information, depending on the marine domain and the characteristics of ML in it (Workshop in Remote Sensing for Marine Litter, ESA-ESTEC 2017). Nevertheless, the combination of in situ data and EO data from operational services -like Copernicus- can be coupled with numerical models to bring a better understanding of the ML dynamics and budgets (Lebreton et al, 2018, Brach et al, 2018, van Sebille et al, 2015), questions that are largely unexplored so far.

As part as an effort of the international community to address these scientific questions regarding ML, a white paper has been submitted and accepted for the decadal OceanObs 2019. This white paper highlights the need for an observational system for ML to includes Earth Observation as one of the technologies with best characteristics to obtain additional ML information.
In this regard, some key scientific questions are (i) the minimum observable ML plastic/group of ML size; (ii) the fraction of the total ML stock that can be detected with existing EO technologies and (iii) how would be possible to obtain optimal data with technologies fit for purpose. Following these objectives, the SSPG/UKSA-funded project “GeoInt for Marine Litter” (GML) and the ESA-funded project “Remote Sensing for Marine Litter” (RESMALI) try to embrace these challenges from two different angles. The former tackles the capability of existing EO datasets to bring information about marine litter. The later tries to define the optimal technological solution for a dedicated EO mission for ML monitoring.

In this presentation, we aim to present the work performed in both SSGP and RESMALI initiatives, which combined, try to cover the current EO capabilities for observation of ML, as well as defining the best technical options for the future. As it will be presented, Sentinel-2 data can detect and monitor ML under certain conditions, but suffers of strong limitations that cannot be easily circumvented with this instrument. The spectral information provided by S-2/MSI allows for the development of optical indexes that can discriminate between floating ML and other types of materials found at the surface of water bodies. However, Sentinel-2 was not originally targeting the observation of ML, so its characteristics are not optimal for its observation and there are considerable sources of error.

In order to tailor the technologies to this goal, a better characterization of the physical characteristics and the spectral signature of ML -in its diversity- is necessary. We present results obtained within the RESMALI project in this regard. Such information allows for the definition of the best spectral bands that could be used for optimal ML observation with remote sensing techniques. The results of these measurements permit the development of radiative transfer models that help us to understand the propagation of ML signature at the space. Nonetheless, additional technologies with potential observational capabilities for ML have been covered. The fact that existing EO data contains information about ML –as proven by GML project- encourages for the development of a dedicated EO mission for ML, which is the ultimate goal of the ESA project RESMALI.

As result of these activities by the team, a Mission Concept for the future Earth Observation of Marine Litter is also presented, including the approaches investigated in these studies, by considering both mission requirements from the community and our current knowledge about marine litter.



Brach, Deixonne, Bernard, Durand, Desjean, Perez, Van Sebille and Ter Halle (2018). Anticyclonic eddies increase accumulation of microplastic in the North Atlantic subtropical gyre. Marine Pollution Bulletin, 126, 191-196.

Kershaw, P. J., & Rochman, C. M. (2015). Sources, fate and effects of microplastics in the marine environment: part 2 of a global assessment. Reports and studies-IMO/FAO/Unesco-IOC/WMO/IAEA/UN/UNEP Joint Group of Experts on the Scientific Aspects of Marine Environmental Protection (GESAMP) eng no. 93.

Lebreton, L., Slat, B., Ferrari, F., Sainte-Rose, B., Aitken, J., Marthouse, R., ... & Noble, K. (2018). Evidence that the Great Pacific Garbage Patch is rapidly accumulating plastic. Scientific reports, 8, 4666.

Maximenko, N., Arvesen, J., Asner, G., Carlton, J., Castrence, M., Centurioni, L., ... & Crowley, M. (2017). Remote sensing of marine debris to study dynamics, balances and trends. In Preprint at https://ecocast. arc. nasa. gov/las/Reports% 20and% 20Papers/Marine-Debris-Workshop-2017.Pdf

Integrated Marine Debris Observing System (IMDOS), OceanObs 2019 (in preparation). Maximenko et al, 2018.

Van Sebille, Wilcox, Lebreton, Maximenko, Hardesty, Van Franeker, Eriksen, Siegel, Galgani, and Law (2015). A Global Inventory of Small Floating Plastic Debris. Environmental Research Letters, 10.

This work summarises the findings from the ESA funded OPTIMAL project. The main aim was to produce recommendations for a scientific and technical roadmap for the development of a mission for remote sensing detection of marine litter. Because the marine litter is dominated by marine plastics, the project focused on developing the roadmap for marine plastic only. The study comprised: 1) definition of the observational needs from the scientific community; 2) field campaign, modelling and laboratory studies to evaluate the potential for detection of current techniques and propose new approaches; 3) elaboration of synthesis and recommendations for the development of a marine plastic detection system.
The definition of observational needs was done through a consultation with the marine plastic community through a workshop. One of the main conclusions was the identification of two important observational scenarios for marine plastics detection: large size plastics (diameter > 5 cm) accumulated on shore, and microplastics (diameter < 5 mm) in the upper layer of the ocean. Following these recommendations, experimental and modelling work was designed and carried out to evaluate current optical remote sensing capability relevant to those scenarios. For the evaluation of large plastics on the shore, field measurements were carried out in May and June 2018, by placing different plastic targets on a beach and measuring their in-situ hyperspectral reflectance, in combination with satellite and airborne observations. Results from this study are presented in a separate submission.
Investigating the potential for detecting microplastics, a laboratory experiment was fulfilled to measure the optical properties of polystyrene beads as model particles for the natural microplastic populations. A range of bead sizes (0.2, 2.0 and 20 μm diameter) was analysed using a laboratory setup to characterise their abundance-specific inherent optical properties (or cross-sectional IOP: absorption, scattering, backscattering cross-sections) in the visible part of the spectra. As expected, results showed significant variations among different beads sizes for the backscattering cross section (σbbp). Standard forward modelling was carried out using as input measured σbbp , and other default optical properties from marine phytoplankton to produce remote sensing reflectance (Rrs) spectra at different levels of microplastic concentrations obtained from the literature.
Using in-situ microplastic data, collected during the Atlantic Meridional Transect in 2016, we produced Rrs estimates, to represent more realistic concentrations of microplastics. We measured relatively high microplastic concentrations ~10 particles m-3 (63 μmthese studies will be summarised in the talk with respect to recommendations for future development of remote sensing system for marine plastic detection.