Experimental and numerical investigation of the vertical distribution of macro plastic transport in rivers

There is a significant mismatch between the estimated amounts of ocean plastic and the expected plastic ingress by rivers (De Fockert et al., 2024; OECD, 2022). Most plastic transport monitoring data used for global modelling of plastic flux estimations is based on the number of items transported at the water surface (González-Fernández, 2023). However, the amount of the submerged plastic items transport in rivers is not accounted for or is based on approximations (Blondel & Buschman, 2022; Hurley et al., 2023). Notably, submerged plastic transport can exceed surface transport by 4-5 orders of magnitude (Vriend et al., 2023).

Numerical models often oversimplify the macro plastic transport by focusing solely on rising velocity, neglecting physical processes like particle characteristics and free surface interaction (Wickramarachchi et al., 2024). This study addresses these aspects through a detailed experimental investigation on the vertical distribution of near-neutrally buoyant macro-plastics in a controlled laboratory environment to provide validation data to improve the current parametrisation in the particle tracking model Delft3D-PART as part of the Delft3D open source software suite (Stupary et al., 2015).

Experiments were conducted in a  45m long, 1.2m high, and 1m wide flume at Deltares. Large quantities of different types of plastic items were released close to the flume bed 30m upstream of the measurement location, where the position of each item was measured and counted to obtain a vertical distribution. The plastic items used in the campaign were similar to commonly found riverine litter such as bags, foils, cups and spheres, with varying size and plastic type (PP & HDPE). Additionally, the hydrodynamic conditions were varied allowing testing of different turbulent flow conditions.

Acoustic Doppler Velocimeters (ADV) measurements were performed to characterize the flow field and turbulence. Computer vision AI algorithms were used to track the plastic particle positions within the water column, enabling the construction of plastic vertical distribution profiles.

Similar to Valero et al. (2022), the experiment confirmed distinct surfaced and suspended transport layers for near-neutrally buoyant plastics for low turbulent flows. Under low turbulent flow conditions, plastic items concentrated at the free surface, confirming dominance of buoyancy and surface tension effects over turbulent mixing. Within the suspended transport layer, the plastic particles exhibited an inverse Rouse profile. At higher turbulence, the vertical distribution of observed plastics became more uniform for plastic bags, while smaller sized plastics remained well-represented by the inverse Rouse profile. This suggests that classical inverse Rouse theories, which neglect particle size, may not adequately describe the plastic observation profiles of larger sized plastic transport in rivers.

Based on these findings, a Delft3D-FLOW hydrodynamic model was developed and validated against the ADV measurements, in which the particle tracking model Delft3D-PART was adapted to incorporate surface interaction effects based on particle dimensions. This parameterization enables more accurate simulation of riverine plastic concentrations by considering hydraulic dynamics, surface interaction, and plastic dimensions. The improved model parameterization will enhance the accuracy of predicting plastic transport and contribute to the development of effective mitigation strategies.