Hyperspectral imaging for multiscale assessments of marine litter and plastic pollution

Plastic and chemical pollution has become the fifth largest global planetary threat, a threat growing at such a rate that it exceeds current monitoring capabilities. Over the past decade, there have been substantial technological advancements in space and airborne hyperspectral imaging (HSI) with potential widespread applications for pollution detection and monitoring.

NGI has continued to scale research related to environmental and pollution assessments but has not explored large-scale monitoring of pollution and land-use change through satellite and drone-based data beyond a few small efforts.

The assignment

We propose to investigate the application of hyperspectral imagery for detecting marine litter and plastics in the coastal zone near Oslo and other regions of Norway coordinated with a beach clean-up initiative.

Laboratory-based hyperspectral measurements of collected marine litter items and pure plastic polymers have already been acquired. They are currently available as training datasets for developing algorithms to be applied to larger-scale Drone and Satellite imagery. 

We will test and develop data acquisition protocols and processing pipelines (including Machine Learning and Deep Learning algorithms) and identify essential spectral wavelength regions for the detection and mapping of land-based and marine pollution to investigate possible applications, necessary equipment specifications, and approaches to pursue in future EU and NFR proposals.