<p>Macroplastics have been found in many compartments of freshwater systems amongst which floating at the water surface. To quantify the floating macroplastic flux in rivers, the visual counting method was developed. This method is based on visual observations from bridges, and has already been applied in various river systems across the world, including the Rhine-Meuse delta. A two-year dataset of monthly field measurements on ten bridges across the Rhine, Meuse, and IJssel rivers has been collected. This dataset revealed the high variability of the floating macroplastic flux in both time and space. Except for extreme flooding events, the fluctuations are not always simply related to the discharge or season. This finding raises the questions of how to assure representative field observations. Representative field observations are important, as they are typically inter- and extrapolated in time and space. If the timing or location of the measurement is not representative of the ‘normal’ condition, then the extrapolation of that measurement will be associated with a large uncertainty range, resulting in over- or underestimations of the total floating macroplastic flux in the river. As field observations play a major role in calibrating and validating river plastic transport and emission models, it is essential to minimize the uncertainty of field-based floating plastic transport estimates. To optimize the visual counting method and explore its limits, we executed three experiments. The first experiment demonstrated that the temporal variability at bridge level is high, but can be attenuated by repeated measurements. The second experiment showed how many observation points on the bridge are sufficient to account for the spatial variability of the macroplastic flux across the river cross profile. The third experiment determined that the size limit of the visible macroplastics is 1 cm<sup>2</sup> on bridges that are up to 5 meter above water level and 4 cm<sup>2</sup> for bridges up to 15 meter above water level. The findings of these experiments endorse the effectiveness of the visual counting method and allow for a substantiated implementation of this method in floating macroplastic monitoring campaigns across river networks worldwide.</p>

Brazil has invested considerably in the reservoir construction during the past decades, mainly for irrigation and hydro-power generation. Despite their large impact on catchment hydrology, reservoir dynamics are often not included in hydrological models due to their complexity. In this study, we investigated the effect of including reservoir dynamics (realism) in hydrological models on the model performance (accuracy). Combined, realism and accuracy form the model fidelity. We used the HBV-EC and GR4J models to simulate hydrological processes and daily streamflow of 403 catchments across Brazil in two scenarios, with and without reservoirs. The model performances were assessed with the Kling Gupta Efficiency (KGE) and its components, and were compared between the models and scenarios. We found a significant increase in the HBV-EC model performance when the reservoirs were taken into account, although the overall performance was relatively poor. The average KGE increased from 0.21 without the reservoirs to 0.40 with the reservoirs. The GR4J model, on the other hand, showed better overall performance, but without the improvement when including the reservoirs; the average KGE slightly decreased from 0.57 to 0.56. In the catchments with the largest reservoir capacity, HBV-EC in the scenario with reservoirs outperformed GR4J in both scenarios. We note that better model performance can still be obtained with a smaller spatial scale or other methods of including reservoirs, which require more data and detailed studies. With this paper, we demonstrate that model performance can improve when including reservoir dynamics, but this depends on model structure and does not always increase model fidelity.