A semi-autonomous camera-based system for global monitoring of riverine floating macroplastic transport

Understanding, predicting and preventing plastic transport through rivers benefits from monitoring and comprehension of the dynamics of plastic transport from an hourly to multi-annual scale. Existing methodologies for the monitoring of plastic transport are not suited for long, continuous observation periods or lack a rigorous comparability between locations. This paper proposes and tests a methodology for the monitoring and quantification of floating macroplastic transport in rivers that is suitable for multi-year periods and comparable between locations. Camera deployments on bridges that focus on longevity and standardization of data collection in combination with object detection and tracking techniques resulted in high-frequency data of floating macroplastic detections and flux from 11 locations worldwide spanning multiple years. This tracking and detection methodology can be transferred to new locations with low effort in model retraining. Comparison to visual counting measurements shows that we can measure the same dynamics and ranges of flux. We also demonstrate that this technique allows us to measure the surface flow velocity and tidal fluctuations of plastic debris in rivers as well as identify high transport rate events such as floods. This paper offers a comprehensive monitoring method for the transport of floating macroplastics which can be used to improve plastic transport models and measure the effects of policy and mitigation strategies of plastic debris in rivers.