A highway noise prediction model which considers 20 s continuous equivalent sound level measure as
the basic vehicular noise has been developed and is called the Leq(20 s) model. The Leq(20 s) model is
believed to provide accurate predictions by measuring the sound level of individual vehicle without
assuming the vehicle noise source as a point source. To verify the rationality of Leq(20 s) model, a
mathematical derivation was performed based on the Federal Highway Administration (FHWA) traffic
noise prediction model in this study. The derivation process indicated two defects in the Leq(20 s) model.
One was that given the lack of a revision item for finite length road, the Leq(20 s) model cannot be applied
to some special roads, such as in predicting the traffic noise level of a receiving point which is located at
the tunnel portal of a highway. The other was that the Leq(20 s) model had some theoretical deviations
from the model of mathematical derivation from the FHWA model. When the speed of a vehicle ranged
from 20 km/h to 120 km/h, the deviations ranged from 0.80 dB(A) to 0.13 dB(A). The deviations could be
reduced if the constant item of 22.55 dB(A) in the Leq(20 s) model was revised to 22 dB(A). To compare
the accuracy of the Leq(20 s) measure with the measure of the mean energy emission level, which was
used as the basic vehicular noise in the FHWA model, a series of simultaneous measurements of a 5 s
continuous equivalent sound level [Leq(5 s)] and the maximum emission levels of individual vehicles
was conducted. The measurements of Leq(5 s) were compared with the calculations based on the hypothesis
of monopole and dipole sources from the maximum emission levels of individual vehicles measured
on roadsides. Result indicated that the mean value of measured Leq(5 s) was close to that calculated from
the maximum emission level with the hypothesis of monopole source for light vehicles. However, for
medium and heavy vehicles, the mean value of measured Leq(5 s) was between the two values calculated
from the maximum emission level with the hypothesis of monopole and dipole sources. A mean error of
1.2–1.6 dB(A) was obtained for Leq(5 s) when the mean energy emission level was employed in the FHWA
model for medium and heavy vehicles. The accuracy of traffic noise prediction can be improved by
considering Leq(20 s) as the basic vehicular noise instead of the mean energy emission level.