Remote Sensing of Coastal Marine Plastic Debris

The increasing presence of plastic in the world’s oceans and along coastlines is an environmental issue that has captured the attention of both the public and the scientific community. Since the late 20th century, the amount of plastic in oceans, seas, rivers, coastal waters, and on beaches has been rising steadily. Marine plastic debris (MPD) are associated with a range of negative impacts including habitat degradation, harm to organisms, reduced productivity in blue economies, and effects on human wellbeing. The global distribution and high mobility of MPD present significant challenges in monitoring and quantifying the total amount of plastics in marine and coastal environments. In situ data are scarce and estimates of MPD in oceans and on coastlines heavily rely on interpolation between field measurements. Remote sensing has long been used to track environmental change, including monitoring variables within the marine environment. Until the last decade however, the lack of appropriate spatial and spectral resolution rendered satellite remote sensing of MPD unfeasible. With increasing access to sensors with improved resolutions, there has been a rapid increase in scientific output using various imagery for tracking plastic debris across marine environments. In particular, there has been marked interest in the use of multispectral and hyperspectral sensors with greater numbers of spectral bands. This increased spectral resolution is useful for detecting plastic debris using differing spectral characteristics compared to other co-occurring elements. While these methods cannot usually identify individual plastic items, they can detect sub-pixel aggregations, provided enough material is present to produce sufficient spectral difference between plastic-contaminated and non-contaminated pixels. Floating plastics in coastal waters and open oceans have been the primary focus of this developing field. However, while detecting floating MPD in open waters using spectral remote sensing has gained traction, similar research on detecting MPD washed up along shorelines is less developed. While large volumes of MPD are found in open waters, removing plastics from beaches is considered a much easier and cheaper option, and many governments and NGOs devote resources to this endeavor. As such, enhancing our understanding of the spectral characteristics of beached MPD and developing new detection methods can improve not only our comprehension of global plastic debris stocks, but also the efficiency of remediation efforts. This thesis presents new insights into the feasibility of spectral remote sensing for MPD on beaches and introduces a new spectral detection algorithm developed after examining the limitations of these detection methodologies. The first step was to determine whether plastic debris deposited on beaches from surrounding oceans could be separated their using spectral characteristics. in situ MPD on the beaches of the Cocos (Keeling) Islands were recorded, and their spectral characteristics were analysed to determine if they were distinct from non plastic-contaminated regions. The results showed that plastics on beaches were spectrally distinct in the shortwave infrared spectrum, and could be separated from non plastic-contaminated regions if at least 3% of the area was covered by plastic. Following this initial work, the limitations of spectral remote sensing of beached MPD were investigated through determining the minimum detectable limits for different kinds of plastics commonly found along shorelines. Spectral feature analysis was used to better understand differences in spectral characteristics between different plastics, and also between weathered and non- weathered plastic. The results confirmed that sub-pixel aggregations of MPD could be separated from sand, though the minimum detection limit varied by polymer and whether the plastics had been exposed to environmental degradation. Minimum plastic surface covers between 2–8% were required, depending on the type of plastic. With the limitations of MPD spectral detection established, investigation of several candidate satellites with appropriate spectral resolution was undertaken. This was done through the creation of a library of mixed plastic-sand spectra. The suitability of these satellites were assessed in parallel with investigations of potential new spectral indices for shoreline plastic detection. Sentinel-2, while previously demonstrated as suitable for detecting floating plastics, was found to not be optimal for detecting shoreline plastics. However, Worldview-3 and EnMAP were found to have sufficient bands to resolve key plastic spectral features which do not overlap with spectral features from co-occurring inputs, such as sand and seaweed. Furthermore, several indices were nominated with strong relationships between sub-pixel plastic surface cover across several different sands. The most promising index from this study was then employed on a larger spatial extent. The novel Coastal Marine Plastic Index (CMPI) was developed for use with SWIR Worldview-3 imagery and was used in a case study over a beach in Australia. The results demonstrated the utility of the Worldview-3 SWIR sensor for shoreline detection of plastic debris, with the CMPI successfully separating pixels with as little as 10% plastic surface cover from the surrounding environment. The CMPI was able to separate plastic contaminated pixels from both nearby sand and water to a greater degree than other spectral indices that it was bench-marked against. The global distribution of marine plastic debris, coupled with the difficulty in collecting data in remote locations, serve to motivate ongoing efforts to develop global methods of mapping plastics. While significant progress has been made in recent years improving the state of plastic detection with satellites over open waters, there is still a paucity of methods for shoreline plastics. Ultimately, the goal is to achieve accurate, global mapping of MPD accumulated along shorelines, and thus provide both baseline monitoring of the ongoing marine plastic debris crisis and improved outcomes for MPD removal from the marine environment. The findings in this thesis underscore the utility of satellite remote sensing for detecting marine plastic debris along shorelines, contributing towards this goal. Through this work, an improved method for detecting plastic debris on shorelines was produced. Despite beaches being complex systems with competing spectral inputs influenced by many factors, the unique spectral characteristics of plastics can be utilised to detect sub-pixel quantities of contamination.