Most plastic input into the ocean is through riverine transport, hence monitoring of plastic litter (PL) in rivers is essential to quantify inputs for global reduction of plastics into the marine environment. Detecting floating conglomerations of waste with satellite remote sensing would be an effective first-order identifier since plastic makes up to 40% of the floating waste in Indonesian rivers. Indonesia is also the 2nd largest contributor of plastic marine litter (PML). We propose a novel monitoring approach that uses a variety of satellite datasets to identify conglomerations of floating litter in targeted rivers with a width of 60 - 100 m. Land-water boundaries can be discerned on Sentinel-1 SAR and Sentinel-2 MSI imagery, and our tests with the latter also show that floating waste correlates with higher reflectance values in the visible and NIR range. To make our classifier generic and avoid false positives caused by shallows or high suspended particulate matter, this is complemented by texture analysis on high-resolution TerrraSAR-X, which enables us to identify changes in surface roughness correlated with floating litter. When available, hyperspectral EnMAP data will be used to focus on absorption features from the PL floating in the river or deposited on the banks and beaches. We teamed up with River Recycle, which has developed a technology that locally removes riverine floating waste and cleans, sorts and analyses different fractions for a year, thus recording seasonal variations. Cameras with IR filters will be installed upstream, enabling us to validate and improve our algorithms for the detection and estimation of PL fluxes. Finally, our regional hydrological and particle transport models incorporate plastic behavior to add validation and prediction to observed variations in PL fluxes. This comprehensive approach will lead to a new remote-sensing based method for PL identification and quantification, suitable for guiding regional cleaning missions.
