The initial step in the development of a remote sensing mission for marine plastic detection is to evaluate the skill of current optical remote sensing techniques and to test different thresholds for detection. To do so, experimental, in-situ and modelling studies need to be performed. As part of the ESA-OPTIMAL project (Optical Methods for Marine Litter Detection) this work presents results from a field campaign incorporating in-situ hyperspectral radiometry measurements, airborne hyperspectral data and Sentinel-2B MSI concurrent imagery. The aim of this campaign was to evaluate the detection of large marine plastics on the shore and provide guidance for further system development to ESA.

Field measurements were carried out in May and June 2018 on the shore of Whitsand Bay (UK). Five different plastic targets were deployed with sizes of at least 7 m x 7 m and their in-situ reflectance were collected with an Analytical Spectral Device (ASD) FieldSpec Pro hyperspectral spectroradiometer on loan from the National Enviroment Research Council (NERC) Field Spectroscopy Facility in UK. Aerosol content was recorded by a simultaneously operated Microtops sunphotometer. The in-situ data were used in combination with a standard radiative transfer code (6S) to compute remote sensing reflectance at the satellite height. Matching satellite imagery from Sentinel-2B were compared with the simulated spectra. Preliminary results show that although spectral signatures unique to plastic can be recorded by hyperspectral in-situ instrumentation, no current bands in Sentinel-2 MSI can be used to uniquely identify plastic presence on the dry beach. Subsequently, a sensitivity analysis on bandwidth with the potential for detection of plastic using additional bands was investigated.

In addition to the satellite matchup, a hyperspectral imager (Specim Aisa Fenix) operated by the NERC Airborne Research Facility, collected data from their aircraft at five different altitudes between 750 m and 2200 m. The airborne data were atmospherically corrected using the ATCOR4 transfer model. Results from this study allow the investigation of the pixel size resolution to be carried out, providing evidence of the required optimal pixel size to detect unique spectral features of plastic.

Area D - Board 409

Marine litter is a global concern, affecting all the oceans of the world. Every year, millions and millions of tons of litter end up in the ocean worldwide, causing environmental, economic, health and aesthetic problems. The IFADO (Innovation in the Framework of the Atlantic Deep Ocean) project aims to create new marine services to support the European Marine Strategy Framework Directive implementation with the North Atlantic Ocean as a study case. As part of this project, it is of decisive importance to identify convergence zones, pathways and main sources of marine litter in the North Atlantic Ocean. This information is essential in the perspective of preventive and cleaning actions.
The North Atlantic gyre is notably composed by the Gulf Stream on the west boundary, the North Atlantic Current, the Canary current on the south-east European coast and the North Equatorial Current. In this system, the subtropical gyre is known to be a great convergence zone. The Lagrangian particle modelling tool Opendrift, developed at the Norwegian Meteorological Institute, has been used to evidence convergence areas. The tool was fed with global ocean current maps derived from satellite observations (GlobCurrent products) or computed by an ocean circulation model (CMEMS analyses), and different scenarios of marine litter release were modelled in the North Atlantic Ocean. Indicators were developed to highlight the preferential residence zones and pathways for plastic particles. The goal was to understand the fate of plastic particles released from land (80% of marine litter that end up in the ocean are coming from land: beaches littering, river discharge...) and the trajectories of particles floating far offshore (shipping and fishing litter).
This presentation is dedicated to the capabilities of ocean current products derived from satellite observations to simulate plastic particles trajectories at a basin scale in order to validate pathways and convergence zones of floating marine litter. Comparison to results obtained with modelled ocean current shows a good consistency of both products. A time residence indicator has been built to highlight the convergence zones. Correlations between release and stranding of particles are explored in some regions of interest to help define a monitoring strategy.

Area D - Board 410

As part of the EU Copernicus programme, the European Space Agency (ESA) operates the Sentinel-2 missions that yields optical imagery of high spatial resolution (up to 10 m) and frequent revisit (5 days or less).The Sentinel-2 A and B Multispectral Instruments (MSI) share the same orbit, separated by 180 degrees. The Sentinel-2 mission has been designed to cover large swaths (290 km on ground) with high geometrical and spectral performances in 13 spectral bands from VNIR to SWIR.

In this paper, we examine the capabilities of the Sentinel-2 MSI for Marine Litter (ML) detection and monitoring in coastal regions, where its impact possess a significant threat to marine ecosystems. This investigation has utilised the high temporal and spatial resolution of Sentinel-2 MSI products to identify rafts of ML, enabling the isolation of a unique radiometric signature. These initial efforts to detect Marine Litter have already successfully identified multiple targets around the world. These data are vital in providing a global service, capable of leading efficient clean-up operations to tackle this global crisis.

The work done on ML compliments other ARGANS projects – such as Sen2Coral, that demonstrated the potential of the Sentinel-2 MSI mission to monitor coral reef health - and the Satellite Derived Bathymetry (SDB) project, that demonstrates the capability of utilising Sentinel-2 MSI data to deliver global IHO-compliant nautical chart data. Together, they form a suite of tools that will inform local and national response to issues that impact maritime activities and pressure coastal zones.

Importantly, the superior Sentinel-2 performance is built on the quality of their measurements. This paper demonstrates how ARGANS, thorough the DIMITRI (Database for Imaging Multi-spectral Instruments and Tools for Radiometric Inter-comparison ) toolbox maintain the high quality of MSI images and the consistency of radiometric measurements of each MSI-A and MSI-B. DIMITRI consistently validates each sensor against Earth invariant structures, resulting in excellent stability and consistency of measurements from the Sentinel-2 mission. The DIMITRI toolbox confirms that the Level-1C products delivered to users comply with the rigorous demands (a bias better than 5% and an absolute horizontal precision better than 12 metres) set out in the Sentinel-2 Mission Requirements Document (MRD). Accordingly, DIMITRI provides a crucial service, maintaining the validity of science carried out using these data.

Area D - Board 411

Plastics in marine litter are becoming a nuisance in the environment and as a result there has been a dire need to synoptically detect, track, characterise and quantify them in the ocean. We present findings from scientific evidence-based studies and datasets-of-opportunity demonstrating the potential application of optical sensors on airborne and satellite platforms in remote sensing of floating plastics from the open ocean to estuarine systems. We also evaluate the performance of these optical sensors in terms of the ‘detect, track, characterise and quantify’ requirements which also depends on the size distribution of the target aquatic plastics floating or slightly submerged. Optical sensor characteristics were also investigated in terms of temporal, spectral, geospatial coverage and resolution. The measured spectral properties of the target floating plastics were compared to open-access spectral libraries to best predict polymer types present. Additional work was aimed at determining ways to estimate the abundance of plastic debris in target areas. Implications of successful remote detection, tracking and quantification of plastic debris will be towards operational validation of field measurements over large areas and at repeated time intervals in different aquatic environments.

Area D - Board 412

Once debris and plastic pollution reach the Ocean, surface currents, tides and winds play a primary role in the transport and dispersal of floating marine litters. Multi-year numerical simulations based on lagrangian drift models can reproduce the behavior of such debris into the surface ocean transport patterns. They can be used to estimate accumulation areas or hotspots.
This paper presents several studies of hotspots determination performed with a suite of drift models developed and operated by CLS since 2009 for several drift application (search & rescue operations, particles (oil spills, algae), icebergs, lost maritime containers), called MOBIDRIFT.
The MOBIDRIFT model uses environmental data to calculate the impact on the drifted object. Based on a pure 2D lagrangian advection scheme, MOBIDRIFT considers the object shape, its concentration or its immersion and it allows the parameterization of the object drift response to each of the environmental parameters (tides, winds and currents).
The MOBRIDRIFT model can be run with several types of surface currents. These studies use the products developed by the EU Copernicus Marine Environment Monitoring Service (CMEMS). Since June 2018 the CMEMS is particularly delivering a new ocean surface current product that addresses numerous marine applications. This product based on observation-based approach is complementary to ocean current estimates provided by modelling/assimilation systems. It consists in surface and 15m depth total current estimates based on the merging of satellite and in situ observations. The operational product delivered through the CMEMS is an heritage from different R&D initiatives as the GlobCurrent project from the European Space Agency (ESA) and CNES mean dynamic topography studies. They are available in Near-Real-Time (NRT) and as Multi-Year Products (MYP) for the past 25 years.
In addition to long-term drift computation, the determination of marine litter and plastic pollution hotspots is also scrutinized through the computation of Lagrangian coherent structures (LCSs) which help in identifying trajectories in a dynamical system.

Area D - Board 413

Our awareness of the severity and significance of marine plastic pollution is still quite new, and many basic questions about sources, sinks, pathways, and trends remain unanswered. However, it is certain that primary sources of marine plastic pollution are from land, including riverine input and storm-water run-off. The recent launch of the Sentinel satellites in 2015 (2A) and 2017 (2B) is allowing for the application of high resolution products to detect floating objects in the coastal zone – potentially including macroplastics and other debris.

Emerged algae in the intertidal zone and submerged algae in algae-culture farms are successfully detected using variations of a simple ratio normalised vegetation index (NDVI). This approach also shows promise for detection of small floating objects (sub-pixel features) accumulated at sufficient densities to highlight a 10 x 10 m pixel. We mined the AWI-LITTERBASE portal for publications on floating macroplastics, and focused on 3 areas of interest. Namely, the German Bight, coastal waters off the east coast of Scotland, and waters around the San Juan Islands of British Columbia. Using NDVI on top of atmosphere (TOA) imagery, floating objects were detected extensively in the near-coastal zone, especially near river mouths and ports. These sub-pixel features include vessels, spume (ocean foam), and in one notable case, possible high aggregations of Northern Gannets along an oceanic front. Although we cannot confirm the remaining detections flagged as plastic are indeed plastic, spectral signatures match those from 10 x 10m plastic targets deployed in the Aegean Sea by the Marine Remote Sensing Department, University of the Aegean.

Many atmospheric correction methods for marine applications rely on the assumption that water-leaving radiance is zero in the near-infrared (NIR). This assumption isn’t always robust in situations where there are valid water leaving signals in the NIR, including high turbidity, sea ice and plastics. The Polymer algorithm derives ocean colour parameters in the whole sun glint spectrum and does not necessarily require water-leaving radiance to be zero in the NIR band. By applying Polymer and adjusting this atmospheric correction to retain any signals in the NIR, we are currently developing an automated method for improved detection of floating debris in the coastal zone. With increased in situ data to validate our method, macroplastics located using Sentinel-2 could potentially be removed if detected as part of reactive clean-up programmes, or be used as a means of highlighting sources (problematic rivers).

Area D - Board 414

Remote sensing provides a promising opportunity to address the urgent need for objective detection and quantification of widespread, persistent plastic pollution in the world’s oceans. An especially high priority is the identification and development of optimal systems to accomplish a number of observation goals in variable environments, amid a selection of sensors/imagers and aerial platforms, each with individual, as well as combined advantages and weaknesses. I will discuss the suitability of a range of hyperspectral infrared (IR) imagers for the detection of plastic pollution in the marine environment and Polar regions from a small unmanned aerial vehicle (UAV), and provide comparison with other sensing methods, including visible spectrum imaging.

Plastic polymers exhibit characteristic spectral features throughout the infrared spectrum. Fourier Transform Infrared Spectroscopy (FTIR) is a conventional and commonly used lab technique that characterises polymers based on features in the longwave infrared (LWIR) region. Our research team has designed a modified optical configuration that eliminates the moving parts of a conventional FTIR imager for a compact payload on a UAV. Rather than pinpointing litter at sea, this technology provides the most promise in detecting stranded, shore-based plastic debris. Efficient, simultaneous characterisation of a collection of samples through imaging may provide additional purpose as a lower cost and compact lab-based alternative to full-size FTIR equipment.

Infrared radiation generally penetrates the water column to increasing depths with decreasing wavelength. Near infrared (NIR) and shortwave infrared (SWIR) imagers are therefore considered for detection of floating plastic debris at sea, in addition to shore-based litter. I will discuss the analysis of preliminary data collected with a hyperspectral visible-NIR imaging system along several shorelines in the Arctic region of Svalbard. The Earth’s Polar regions are particularly interesting as research areas for a number of reasons. Cold temperatures may change hardware performance and spectral measurements as observed in more temperate climates, and low solar angles influence the interaction of infrared radiation with the environment below and sensing equipment. In these remote regions, seasonally intense UV radiation promotes embrittlement and breakup of plastic debris. The “pristine” Polar environments, distant from large human populations, may be regarded as indicators or case studies for the rest of the world. Objectively and efficiently determining the extent of their contamination by remote sensing offers valuable information regarding plastic pollution in a global context.

Area D - Board 415