<p>The atmospheric dispersion of pollutants emitted by inland vessels on the Rhine River was investigated using large-eddy simulation (LES) modeling. The PALM model system was extended to simulate the transport of pollutants within a complex river and urban environment, with emissions represented by dynamic volume sources corresponding to moving vessels. A total of twelve emission scenarios were evaluated, with variations in emission rate, vessel speed, and direction. Each scenario was repeated multiple times under quasi-stationary atmospheric conditions to create an ensemble of statistically independent realizations, allowing for the analysis of plume dispersion variability. Concentration time series were sampled at virtual receptor points located above the river, along the riverbank, and within adjacent urban areas, and were compared against in-situ measurements. The comparison revealed that the simulated plume characteristics at measurement sites were plausible in terms of peak amplitude and shape. A high degree of variability was observed among ensemble members representing a single ship, often exceeding the variability across different vessels with distinct emission rates. This underscores the challenges of inferring emission strengths from limited in-situ observations. Furthermore, peak amplitude and signal duration were found to depend strongly on receptor location and distance from the source: riverbank stations exhibited the highest concentration peaks and shortest durations, while inland urban stations showed lower amplitudes and longer durations due to increased flow obstruction from vegetation and buildings. Nevertheless, ensemble-averaged metrics, such as maximum values, dosages, and durations, exhibited a clear correlation with emission strength, thereby underscoring the model’s utility in assessing the relative impacts of inland shipping emissions.</p>