Direct and Indirect River Plastic Detection from Space

Plastic pollution is a threat to both human health and ecosys-tems. Quantification efforts often rely on observations in the field, which can be costly and labour-intensive. Satellite remote sensing offers an opportunity for upscaling plastic detection in time and space, as it enables covering large areas and using historical images. To date, several approaches have been successful for the detection of marine plastics, but detecting riverine plastics with satellite imagery are lacking. In this paper, we explore the use of satellite imagery for direct and indirect detection of floating plastics in rivers, ultimately calculating river surface plastic densities. The direct detection was done using Worldview-3 imagery and enables to detect large objects (~ 1.2 m of size). The indirect approach uses plastic entrapped in water hyacinths as a proxy for plastic density at the river surface. Sentinel-1 imagery allowed to estimate water hyacinth coverage, which we combined with field data to estimate entrapment ratio and plastic density within hyacinths. Items as small as 2.7 cm could be detected with field data. The indirect approach detects a thousand more plastic items (2.1 × 10
⁴
items km
^-2
) than the direct approach (5.8 × 10
¹
items km
^-2
), a likely result of the larger range of item size detectable with this method.