Calculation Accuracy

It is important to note that even if all input data is given as correctly as possible, there will always be a certain amount of uncertainty in a noise calculation. Noise values that are close to borderline between the need for noise reductions measures or not (e.g. barriers), should always be examined closely in relation to the inaccuracy of noise calculations in general.

Inaccuracy in calculation results may be caused by:

• Uncertainty in input data
• Uncertainty in calculation methods
• Uncertainty in implementing of the calculation methods

The first kind of uncertainty consists of how well we manage to give correct input data to the programs. Examples of uncertainty factors may be:

• The map is not updated.
• Terrain data is organized in an incorrect way in the terrain model.
• Incorrect traffic data.
• Equidistance and scale of the map give a certain uncertainty in the terrain data.
• In a terrain profile we assume that the terrain is the straight line between two contour lines. This is not always the case.
• There is always an uncertainty in the calculation of terrain levels in the terrain model.

Figure showing uncertainty by inaccuracy in the basis for the map

A: Inaccuracy results in this terrain
B: Correct terrain
a: Noise source
b: Sight line
c: Calculation point

It is hard to say anything in general about how large this uncertainty might be, because it depends on a number of things. A critical case is when a barrier, which as a matter of fact does not give any abatement, just “kicks in” and still gives an abatement because of inaccuracies in the basis of the map. Here errors of a few centimetres in the terrain model may amount to a difference of ± 3dBA.

Each calculation point should be considered regarding this kind of uncertainty. If we have a map with equidistance 1 m and scale 1:1000, and these are not typical critical cases, this uncertainty rarely will get larger than ± 1 dBA in Novapoint Noise.

There is also an uncertainty in how well Nordic Calculation method reflects the reality. From comparisons between Novapoint Noise calculations and measured values, it may seem as if Nordic Calculation method in general gives a little too high values. At present there is an ongoing work trying to find constant relations, here regarding an eventual revision of the Nordic Calculation method. Our tests suggest that there might be a difference of 1-4 dBA. Examples of uncertainty factors may be:

• There is uncertainty to how well the terrain profile of a section is representative of the section.
• There is some uncertainty to whether reflection planes are inserted reasonably.

The fact that a section is represented by just one terrain profile, is of course an element of uncertainty.

Figure showing that the terrain profile doesn't have to represent the whole section

A: Noise source
B: Barrier the way it is interpreted by the noise program
C: Terrain line
a: Section
b: Calculation point

Good - average accuracy The calculation principal of Novapoint Noise is based upon that the road is divided into a series of short sections and that terrain profiles are generated between the centre point of the sections and the calculation point. The shorter sections we have, the closer the terrain profiles get and the more accurate the calculation result gets. Short sections will lead to longer time of calculation. See also the chapter Uncertainty in Nordic Calculation model.

Below is shown how the length of the sections vary regarding the distance between the road and the calculation point for the two methods, and when the calculation is terminated based on the shortest distance between the calculation point and the road.

1) States length of section (m) with good accuracy. Parts of the road having distance from calculation point larger than dmin+ 800 is not included in the calculation. dmin is the shortest distance between the road and the calculation point.

2) States length of section (m) with average accuracy. Parts of the road having distance from calculation point larger than dmin+ 500 is not included in the calculation. dmin is the shortest distance between the road and the calculation point.

The third kind of uncertainty is based on how well the programs reflects the calculation methods. This uncertainty is very small in NBSTOY, roughly ± 0,5 dBA.

We have made comparisons between Novapoint Noise and NBSTOY. These have roughly given deviations of less than 1 dBA. This may indicate that Novapoint Noise reflects Nordic Calculation method to the same standard as NBSTOY.