Influence of Traffic Input Data Quality on Road Noise Estimates Using the CNOSSOS-EU Method.

Accurate traffic input data are essential for reliable road noise mapping within the CNOSSOS-EU framework. However, European countries often rely on heterogeneous data sources and measurement practices, which may introduce uncertainties in noise estimates and reduce the comparability of results across regions. This study evaluates the performance of three traffic data collection methods, specifically microwave radar traffic counters, artificial intelligence-based cameras, and Google API-derived flows, in three representative test sites located in Italy and Romania. Traffic flows and vehicle category distributions obtained from each method were used as inputs for noise simulations, and predicted levels were compared with in situ noise measurements. A second analytical approach was developed to estimate short-term noise levels at a 10' resolution by combining CNOSSOS-EU power models with propagation matrices computed using commercial sound propagation software. The results show that both radar counters and cameras provide reliable inputs for day/evening/night indicators, although counters may miss flows under complex traffic conditions, and cameras may overestimate counts at high volumes. Google API-derived flows perform well only when traffic exceeds approximately 150 vehicles per hour and when the traffic model is carefully calibrated. Manual counting confirmed that all three input data collection methods exhibit non-negligible traffic loss, which contributes to a systematic underestimation of simulated noise levels when using average flow-based modeling. Differences between methods become more pronounced when analyzing short time intervals rather than aggregated indicators. Overall, this study highlights the strengths and limitations of each data source and provides guidance on their appropriate use for road noise assessment and strategic mapping.