Revealing River Plastic Transport under different flow conditions: Long-term camera monitoring and Citizen Science in the Sarno River (Southern Italy)

Rivers are key conduits of plastic debris from land to sea, with transport often amplified during rainfall-driven high-flow events While large rivers can sometimes be monitored from space, narrow rivers and small basins—such as the Sarno River—require high-resolution, in situ approaches capable of resolving rapid, event-scale dynamics. This study consists of two main parts: 1) long-term, fixed-site monitoring, and 2) dense citizen-science observations to characterize river plastic transport under contrasting flow conditions. A low-cost RGB camera system was installed at a representative Sarno cross-section and operated continuously for one year (November 2023–October 2024), acquiring time-lapse imagery at 15 s intervals. Hydrometeorological forcing was reconstructed using ERA5-Land precipitation (hourly to monthly products) over the upstream and downstream portions of the basin to identify and contextualize low- and high-flow periods. Plastic items were detected and counted from the imagery using a YOLO-based model trained with a self-/weakly supervised strategy, implemented in two configurations: a single-class detector (plastic vs. background) and a multi-class detector (13 classes) to better differentiate plastic categories and support source/process interpretation. Analyses were performed on dates with complete camera data, spanning both low-flow conditions and rainfall-driven events.To complement the fixed-site record and capture network-scale variability during anticipated high-rainfall events, we deployed the RiverWatch app to collect geotagged images of plastic presence across multiple locations along the Sarno river network. By coupling continuous, cross-section-scale detection with event-focused, spatially distributed citizen observations, this work demonstrates a scalable pathway to quantify plastic transport dynamics in small rivers and to support monitoring strategies that are inclusive, low-cost, and transferable.Keywords: riverine plastic, image-based monitoring, YOLO, long-term monitoring, citizen science, Sarno River