Tim van Emmerik , Louise Schreyers, Yvette Mellink, Ty Sok, Mauricio Arias

Plastic pollution in rivers is of increased global concern. Rivers act both as pathways for land-based plastic waste into the ocean, and as plastic reservoirs for long-term retention. Reliable observations are key to designing, optimizing and evaluating strategies to prevent and reduce plastic pollution. Several measurement methods have been developed to quantify macroplastic ($>$0.5 cm) storage and transport in rivers, including visual counting from bridges, net sampling, and images-based techniques. Method harmonization is crucial to make sure data collected using different techniques remains consistent. In turn, this would allow for comparative analysis of plastic pollution within and between rivers. In this paper, we present a harmonization approach to estimate floating plastic item and mass transport from data collected using different methods. The approach allows estimating the same values based on different measurement methods and data collection protocols. We applied our approach to the Mekong-Tonlé Sap-Bassac river system around the city of Phnom Penh, Cambodia. We estimated the floating plastic item and mass transport in the wet and dry season by combining data from net sampling and visual counting. During the wet season, plastic transport in the Mekong increased with a factor of up to 170 (item transport) and 294 (mass transport) compared to the dry season. The river plastic mass balance around Phnom Penh changed considerably, which was mainly due to the flow reversal of Tonlé Sap river between the wet and dry season. Downstream of Phnom Penh, the total plastic transport was consistently higher than upstream, although less in the wet season (1.5-1.7 times) compared to the dry season (3.8-5.9 times), emphasizing the city's role as entry point of plastic pollution into the Mekong. The largest sources of uncertainty are assumed to be caused by key differences between methods, including the size ranges, extrapolation from observation point to full river width, and the contribution of submerged plastic to the total transport. Future work should focus on including data from other methods than net sampling and visual counting, and reducing the uncertainties related to combining data from different methods. Our results show that river plastic transport dynamics are highly variable over time and space, especially around confluences, bifurcations and urban areas. With our paper we aim to contribute to further harmonization of river plastic monitoring.