Outdoor noise measurements compared with predictions by Nordic models

Outdoor noise measurements of road traffic at 16 sites, rail traffic at 7 sites, and of two large industrial sources are reported. The measured sound levels are compared with the levels calculated by the respective Nordic prediction models. The prediction of the equivalent level of road traffic noise was very accurate: the mean prediction error was less than 0.5 dB. The maximum level predictions were obviously worthless. The rail traffic noise comparisons contain three sets of data: 228 train passings of the Finnish measurements and additional Danish and Norwegian measurement data with 529 train passings. The rail traffic noise model was shown to be reasonably accurate. The mean prediction error for the sound exposure level was less than 1 dB and the internal bias between exposure and maximum levels in the model was 1 to 2 dB. Tentative values for the necessary train type corrections are given for some Nordic types. The results of the industrial noise comparison were not satisfactory, which reflect the various problems encountered with large sources and propagation to long distances. Advanced signal processing and propagation calculation techniques were applied for checking the results. These showed that results with an apparently good agreement may in fact be unreliable due to contamination by background.