Control calculation of the simple joint-Nordic model for rail traffic noise

The predictions of a simple rail traffic noise model were checked against measurement data acquired in the Nordic Countries. Data from 284 individual trains together with averaged data from another 427 trains exhibited considerable scatter. The comparison showed that the proposed model was accurate to within 2 dB on an average. As adjustments it was suggested that the mean prediction difference were set to zero for the maximum level L Amax and that the resulting bias in the equivalent level L Aeg estimate were suppressed by stressing the use of the angle correction. Scatter with different train types could only be reduced by more detailed information on noise emission.