An evaluation of models which predict noise levels from road traffic in terms of the LEQ (24-hour) index was undertaken in 1998 by comparing predicted levels with noise levels measured at 20 sites. These sites included motorways, highways, urban arterials and urban streets in New Zealand.
Three main LEQ models were evaluated: the Nordic, the FHWA, and the CETUR models. In addition, the CRTN model was included as a LEQ -type model with both UK and New Zealand road surface corrections, and a simplified version of CETUR.
The Nordic model was able to correctly predict noise levels within ±2 dBA at all 20 sites. CRTN was able to predict LEQ (24 hour) to the same level of accuracy as L10 (18 hour), and predicted 15 sites. The CETUR and FHWA models were less successful, with only 65% correct predictions.
The LEQ models could give strongly indicative predictions (80% successful) for traffic flows as low as 10 vehicles per hour or less. For successful prediction, accurate account needs to be taken of road surface effects on noise levels.