Hyperspectral remote sensing as an environmental plastic pollution detection approach to determine occurrence of microplastics in diverse environments

Abstract

Microplastic (MP) pollution poses serious ecological and human health risks, necessitating advanced detection methods to support targeted remediation efforts. Hyperspectral remote sensing was hypothesized to offer a promising solution, utilizing Near-Infrared (NIR) and Short-wave Infrared (SWIR) spectroscopy for identifying and differentiating materials, including plastics, in the environment. In this study, a total of 228 unique substrate-plastic-concentration combinations containing polyethylene (PE), polyethylene terephthalate (PET), polylactic acid (PLA), polypropylene (PP), polyvinyl chloride (PVC), or styrene-butadiene rubber (SBR) at varying concentrations (0 %, 0.15 %, 0.5 %, 1.5 %, 5 %, 15 %, 50 %, 100 %) were mixed with different substrates (soils, concrete, vegetation, and water). The mixtures were analyzed with NIR spectroscopy, and 8240 raw spectra were preprocessed to remove instrumental and path distortions. Results showed that detection sensitivity varied by substrate, with the polypropylene (PP) index being identified as the most sensitive to the presence of all plastics in the present study. Principal Component Analysis further revealed the association of increasing plastic concentration with key wavelengths, which were used to develop band equations for detecting each plastic via hyperspectral image analysis. These band equations were validated with hyperspectral imagery from AVIRIS-NextGen to map plastic pollution at a landfill site in Houston, Texas, USA, a plastic sink that would reasonably contain plastic pollution. This investigation demonstrates the potential of hyperspectral imaging for mapping terrestrial plastic and microplastic pollution, offering a scalable tool for MP monitoring to support remediation strategies.

Keywords: Environmental plastic detection; Hyperspectral remote sensing; Near-infrared spectroscopy; Plastic mapping.