We have identified a problem with Method 1 as set out in the TRL document

CONVERTING THE UK TRAFFIC NOISE INDEX LA10,18h TO EU NOISE INDICES FOR NOISE MAPPING EPG 1/2/37 (2002).

Method 1 shows that you convert the L10 of each segment to Leq and then sum the Leqs to get the period Leq. There are two conversion formulae - one for motorways and one for non-motorways. For L10s of around 70 dB, the Leq is about 3 dB below the L10, but as the L10 drops, the difference reduces. For non-motorways, the formula has a constant term of 24.5 dB, which means that no matter how low the L10 gets, the Leq is never less than 24.5 dB.

When developing the formula, TRL did not consider that the effect of segmenting the roads would be to create a lot of low L10 values, which get converted into an Leq which is much too high.

We would like to make it clear that NoiseMap implements the method correctly as described.

As an immediate fix, we recommend that people **set all roads to motorways **if they are using method 1. This will result in a much smaller over-prediction. In the longer term, we will revise our implementation of method 1 so that it accumulates the noise from all segments into sub-totals for motorway and non-motorway, and then converts these sub-totals to Leq. This is how the revised (2005) method 3 works. When TRL/DEFRA issed a revision to method 3, it is unfortunate that they did not notice that the same problem would affect method 1. We have not looked at method 2 - probably that has a similar issue, but we think that it is rarely ever used.