Not a precise science, part 1

Computer models that predict the levels of airborne sound that will propagate within a landscape from a source, such as a high-speed train, are very sophisticated pieces of kit. But however sophisticated they may be, or may try to be, they do have their limitations. This was brought to the attention of the HS2 Select Committee recently by acoustician Graham Parry, who described sound propagation prediction as “not a precise science” (see footnote 1).

Mr Parry appears eminently qualified to make this assessment. His CV reveals that he worked on the Channel Tunnel Rail Project (HS1), and it would appear from his evidence to the Committee that he was involved in the practical tests that were held to verify the HS1 prediction model (see footnote 2).

The truth of his observation is also evident in a figure included in the Environmental Statement (ES) that I reproduce below (see footnote 3).

Source: HS2 Ltd

Source: HS2 Ltd

This shows some of the data collected from the practical tests on the TGV that Mr Parry referred to, represented as a “scatter diagram”. It was derived from measurements made at a variety of locations of the noise level resulting from pass-bys of various TGV Atlantique trains running at a variety of speeds, comparing these measured values with the predictions of the noise for those locations and pass-by speeds.

This diagram reveals that the measured data was up to 10dB(A) higher than the predicted value. As Sir Peter Bottomley pointed out (see footnote 4), the error appears approximately symmetrical, so the predictions could be up to 10dB(A) too high, but no one is likely to object if the noise levels turn out to be lower than predicted, are they?

The section of the ES that contains the above scatter diagram – section 1.3 in Annex D2 to Appendix SV-001-000 in Volume 5 – advises that a regression analysis of the TGV data “gave a standard error, for the goodness of fit between predicted and measured levels of approximately 3dB(A)”. This, it is claimed therein, “means that the difference between predicted and measured sound levels is typically within ±3dB(A)”. I guess that it means what you mean by “typically”. The author of the HS2 Ltd scatter diagram has been kind enough to draw the five and ninety-five percentile lines – these mark the boundaries within which ninety per cent of the observations lie. To my mind, the distance between the ninety-five percentile line and the line joining the points where predicted and measured levels are the same is probably a better indicator of the worst-case scenario for those who want to assess how badly HS2 noise might affect them, and the theory predicts that this distance will be 1.96 times the value of the standard error, for a large sample size. This yields about 6dB(A) as the figure we should be taking as our worst-case prediction underestimate, and tallies with the approximately 12dB(A) spacing between the two percentile lines draw on the scatter diagram.

The ES claims, in paragraph 1.3.3 in Annex D2 to Appendix SV-001-000 in Volume 5, that, “Measurements undertaken along HS1 since it came into operation have shown that the prediction method tends to over-estimate in-service noise levels”. I’m sure that this is meant to be reassuring, but it is rather too unsubstantiated to my mind. If this has been the case, where is the data to support this assertion and why is it not in the ES? Also, to what degree does the HS1 model “over-estimate”? No, I’m afraid that, with only this flimsy claim to go on, I will stick with the evidence in the scatter diagram to make my assessment of how accurate we can expect the predictions to be.

Furthermore, if we are comparing the predictions in the ES with LOAEL and SOAEL it is not only the possible 6dB(A) underestimate inherent in the prediction method that we need to take into account; we should not forget to add a further 2.5dB(A) to convert the predicted levels to façade measurements, as I explained in my blog Facing a brick wall, part 2 (posted 1 Oct 2015).

Mr Parry observed to the Select Committee that, “the noise levels referred to both within the environmental statement and here in this Committee are provided as though they were absolute certainties” and cautioned Members that this was “not the case” (see footnote 5). On the basis of the TGV tests he is surely right to express this caveat, and it is something that we should all bear in mind.

(To be continued …)


  1. Mr Parry appeared as an expert witness for Halton Parish Council and Wendover Parish Council during the afternoon session of the HS2 Select Committee that was held on Monday 7thSeptember 2015. His evidence begins at paragraph 156 in the transcript and at 15:27 hrs in the video. The comment of his quoted in my blog is recorded on paragraph 174 of the transcript.
  2. See paragraph 174 in the transcript of the afternoon session of the HS2 Select Committee that was held on Monday 7thSeptember 2015, in which Mr Parry refers to “data which I collected on the TGVA in order to develop a model for HS1”.
  3. This diagram is the right-hand part of Figure 7 in Annex D2 to Appendix SV-001-000 in Volume 5 of the ES.
  4. See paragraph 179 in the transcript of the afternoon session of the HS2 Select Committee that was held on Monday 7thSeptember 2015.
  5. See paragraph 173 in the transcript of the afternoon session of the HS2 Select Committee that was held on Monday 7thSeptember 2015.

Important Note: The record of the proceedings of the HS2 Select Committee from which the quotes reproduced in this blog have been taken is an uncorrected transcript of evidence, which is not yet an approved formal record. Neither witnesses nor Members have had the opportunity to correct the record in such instances, and it may therefore be subject to changes being made in the light of any such corrections being requested.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: