Rail transport is the second most important way of transporting people and freights by land in the European Union. Rail noise affects around 12 million people in the European Union during the day and around 9 million at night. There are two possible ways to assess environmental noise: noise measurement in situ and prediction using mathematical models. The aim of the work is based on the performed measurements and selected noise predictions to evaluate the accuracy of the prediction models and assess their sensitivity to various aspects. Two measuring points in the Banská Bystrica Self-Governing Region, within Slovakia, were selected for measurement, which is characterized by increased mobility of the population. For prediction, the two methodologies were selected (Schall 03 and Methodical instructions for the calculation of sound pressure level from transport). The results show that the Schall 03 method is sensitive to the measurement location (the value reaches half of the significance level) and to the location-period interaction. The second prediction method is sensitive to systematic error (absolute term) and, such as Schall 03, to the location-period interaction. This method systematically overestimates the results. Results showed greater accuracy of both prediction models compared to the measured noise values than the results of the authors in other countries and conditions.
Keywords: environmental noise; prediction method; rail transport; sound pressure level; urban areas.