Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram
Office of Planning, Environment, & Realty (HEP)
HEP Events Guidance Publications Glossary Awards Contacts

Measurement of Highway-Related Noise

5. Vehicle Noise Emission Level Measurements for Highway Noise Prediction Models

This section describes recommended procedures for the measurement of vehicle noise emission levels. Among other purposes, emission levels are required to input user-defined vehicles in the FHWA Traffic Noise model (FHWA TNM®).(3) The TNM is used to predict sound levels in the vicinity of highways and to design highway noise barriers. The procedures described below are consistent with the methodology used during the development of the Reference Energy Mean Emission Level (REMEL) Data Base for the FHWA TNM.(4,36)

5.1 Site Selection

5.1.1 Site Characteristics

To minimize site specific effects associated with vehicle-noise emission level measurements, it is recommended that between five and ten unique sites be selected. These sites should possess the following geometric characteristics:

Click on image for full description

Figure 8. Site geometry.

The above characteristics and parameters are presented for vehicle noise emission level measurements in general; Section 5.6.1 presents specific requirements and measurement parameters associated with inputting user-defined vehicles in the TNM.

5.1.2 Microphone Location

The microphone system should be placed 15 m (50 ft) om the center of the near travel lane, with the microphone diaphragm positioned for grazing incidence, 1.5 m (5 ft) above the plane of the pavement (See Figure 8). Additionally, systems may be optimally positioned at other offset distances, e.g., 7.5 and 30 m (25 and 100 ft), for the purpose of characterizing measurement-site drop-off rate.

5.1.3 Vehicle Types

Roadway vehicles are typically grouped into five acoustically significant types, i.e., vehicles within each type exhibit statistically similar acoustical characteristics. These vehicle types are consistent with the FHWA TNM, and are defined as follows:

One of the primary purposes for performing REMEL measurements is for the purpose of characterizing user-defined vehicle types (See Section 5.6.1). Such types may include motor homes or electric cars.

5.2 Noise Descriptors

The maximum, A-weighted sound-pressure level with fast exponential time-averaging (LAFmx) should be used for the development of vehicle noise emission level relationships. Additionally, spectral data, although not required, may be useful during analysis. Specifically, since TNM computations are performed in one-third octave-bands, it may be helpful to verify consistency with the spectral data currently in the model.(4)

5.3 Instrumentation (See Section 3)

5.4 Sampling Period

The sampling period for each vehicle pass-by will vary, but should be chosen to encompass a time period such that a minimum rise and fall in the noise-level time-history trace of 6 dB is achieved, with 10 dB being preferred (See Section 5.4.1). Rise and fall are defined, respectively, as the difference between LAFmx and the minimum measured level associated with either the start or end of a given pass-by (whichever difference is smaller). This criterion ensures acoustic quality of the pass-by event, and may be determined by (1) observing the display of the sound level meter; or (2) examining the time-history chart produced by a Graphic Level Recorder (GLR). A GLR is the preferred instrument for establishing event quality.

5.4.1 Event Quality

The event quality for each pass-by should be determined during data measurement and prior to data analysis. Event quality is characterized by three type designations (Type 2, 1, or 0).

Events with a rise and fall of the optimum 10 dB or greater are designated as Type 2, the highest quality event. Events with a rise and fall of between 6 and 10 dB are designated as Type 1. Events with a rise and fall of between 3 and 6 dB are designated as Type 0, and in most cases should not be used. Events with less than a 3-dB rise and fall should be discarded.

In special situations, events in which the ambient is less than 10 dB below the LAFmx and events designated as Type 0 may be used in the analysis. More specifically, it may be necessary to relax the 10-dB ambient requirement, discussed in Section 5.1.1, to 6-dB. This situation may occur, for example, during the measurement of low-speed automobiles or during the measurement of hard-to-find vehicle types, e.g., buses. The LAFmx for these events may be corrected for ambient via energy-subtraction before data analysis as follows:

Ladj = 10 x log10 (10{0.1Lc} - 10{0.1La})           (dB)


For example:


Ladj = 10 x log10(10(0.1 x 55.0)-10(0.1 x 47.0)) = 54.3 B

Furthermore, it may be necessary to use events designated as Type 0. These events may be corrected only if the 10 dB-ambient requirement is maintained, and as such, the rise and fall of these events can be attributed entirely to nearby vehicles. This correction is to be performed by subtracting from the measured LAFmx, the sound energy due to “contaminating” vehicle(s) as follows:

Ladj = 10 x log10(10{0.1Lc}- 10{0.1La})           (dB)


This method is only viable if a time-history trace is available. In such instances, the sound due entirely to a contaminating vehicle can be estimated through linear extrapolation (See Figure 9).

An XY graph with Time on the X-axis and measured sound level on the Y-axis is shown. Click on image for full description

Figure 9. Correction for contaminating vehicles.

5.4.2 Minimum Separation-Distance

To ensure negligible contamination from vehicles other than the subject vehicle, a minimum separation-distance between vehicles should be used during the process of event selection in the field. A previous study(35) has shown that a minimum of 120 m (400 ft) between similar vehicles is required to insure that the contamination from nearby vehicles is less than 0.5 dB. In the case of sequential pass-bys of unlike vehicles, such as an automobile followed by a heavy truck, a minimum of 300 m (985 ft) is required (See Appendix C for further details).

5.4.3 Recommended Number of Samples

While, the number of samples is somewhat arbitrary and often a function of budgetary constraints, a larger number of samples will result in higher precision and a greater degree of statistical confidence in the final emission levels. Table 6 provides, as a function of speed, the recommended minimum number of samples. These numbers should be considered an absolute minimum for characterizing automobiles, medium trucks, and heavy trucks. However, for more obscure vehicle types, such as buses, motorcycles, or motor homes, it may not be practical to obtain such a significant number of samples. As a point of relative comparison, 2825 autos, 765 medium trucks, 2986 heavy trucks, 355 buses, and 39 motorcycles were sampled for the development of the TNM.

Table 6. Recommended minimum number of samples.
Speed Minimum Number of Samples
0-10 10
11-20 10
21-30 20
31-40 30
41-50 100
51-60 200
61-70 100

5.5 Measurement Procedures

  1. The instrumentation should be deployed as shown in Figure 10.
  2. Prior to initial data collection, at hourly intervals thereafter, and at the end of the measurement day, the entire acoustic instrumentation system should be calibrated. Meteorological conditions (wind speed and direction, temperature, humidity, and cloud cover) should be documented prior to data collection, at a minimum of 15-minute intervals, and whenever substantial changes in conditions are noted.
  3. The electronic noise floor of the acoustic instrumentation system should be established daily by substituting the measurement microphone with a dummy microphone (See Section 3.1.5). The frequency response characteristics of the system (if applicable) should also be determined on a daily basis by measuring and storing 30 seconds of pink noise from a random-noise generator (See Section 3.1.6).
  4. If applicable, calibration of the Doppler radar should be periodically checked in the field for accuracy and functionality, using a calibrated tuning fork, and the unit’s “internal circuit test” capability, if available.
  5. Ambient levels should be measured and/or recorded by sampling the sound level at each receiver with the sound source quieted or removed from the site. A minimum of 10 seconds should be sampled. Note: If the study sound source cannot be quieted or removed, an upper limit to the ambient level using a statistical descriptor, such as L90, may be used. Such upper limit ambient levels should be reported as “assumed.” Note: Most sound level meters have the built-in capability to determine this descriptor.
  6. A minimum of two operators are necessary for logging all field data: a vehicle observer and an acoustic observer. For each pass-by event the following data should be logged: site number, event number, vehicle class, vehicle speed, maximum A-weighted sound level (LAFmx), spectral data (if desired), meteorological conditions, and any observed anomalies or extraneous sounds.
  7. A potential pass-by event is identified when the vehicle observer confirms that the minimum separation-distance criterion is met. Note: Orange highway cones may be positioned 120 m (394 ft) upstream from the observers’ station to aid in identifying potentially acceptable events.
  8. After the vehicle passes the observers’ station, the acoustic observer should begin data capture.
  9. After the vehicle passes the microphones and before subsequent vehicles approach, the acoustic observer should end data capture. Note: If the subject vehicle’s speed varied by more than ±3 km/h (2 mi/h) and/or acoustic contamination was observed, the pass-by event should be omitted from later data analysis.

(Note: Appendix B provides example field-data log sheets.)

two-lane roadway with observer and instrumentation. Click on image for full description

Figure 10. Vehicle emissions measurement plan view.

5.6 Data Analysis

  1. Adjust LAFmx for calibration drift (See Section 3.1.4).
  2. Merge LAFmx data and corresponding vehicle information, including speed data, into a single file for subsequent analysis, and development of REMEL regression equations. A spreadsheet-compatible file is recommended. Note: It is extremely important not to exclude samples which appear to be outliers (e.g., samples measured for extremely loud vehicles) in the data set. Due to the nature of the field measurement procedures, specifically the use of the minimum separation-distance criteria, the data collected are truly representative of a random sample.

5.6.1 Development of REMEL Regression Equations

The FHWA’s Traffic Noise Model (FHWA TNM®) used for noise prediction and barrier analysis and design allows the user to input user-defined vehicles. However, it is anticipated that the capability to input user-defined vehicles in the FHWA TNM will not be used for entering state-specific emission levels. Based on work performed by the Volpe Center,(40) there is no indication of a need or justification for developing state-specific REMELs at this time. Until the design of highway vehicles change incrementally, or regulatory requirements warrant lower noise emission levels, development of state-specific REMELs is unnecessary.

However, the user-defined-vehicle capability in the FHWA TNM is intended for describing vehicles which differ significantly from automobiles, medium trucks, heavy trucks, buses, or motorcycles (e.g., motor homes or electric cars). Unique vehicles should be measured under the following reference conditions: constant-flow roadway traffic; level grade; and dense-graded asphaltic concrete or Portland-cement concrete.

The first step in defining a user-defined vehicle is to develop the level-mean emission level equation. To develop the equation, the measured LAFmx data should be regressed as a function of vehicle speed for each vehicle type. This can be done with any commercially available statistical analysis program. The functional form of the regression equation is as follows:

L(s) = C + [A x log10s + B] =

10 x log10[10C/10 + 10(A x logs+B)/10] =

10 x log10[10C/10 + sA/1010B/10]                    (dB)

For example:


L(65 km/h) = 10 x log10(10(50.128/10)+65(41.741/10) x 10(1.149/10)) = 76.8 dB

In the above equation, L(s) is expressed in terms of the logarithm to the base 10 of the coefficient, C, (the engine/ exhaust coefficient, which is independent of vehicle speed); and, A x log10(s) + B (the tire/pavement-term, which increases with increasing speed, s). The graphical form on a logarithmic plot of L(s) is illustrated in Figure 11 below.

An XY graph with Vehicle Speed on the X-axis and Sound Level in decibels on the Y-axis is shown. A horizontal line labeled C, the Engine/Exhaust Noise portion, remains constant as speed increases. The regression equation line (A*log10(s) + B), the Tire/Pavement Noise portion, increases as speed increases.

Figure 11. Graphical form of the FHWA TNM regression equation.

The level-mean emission level equation is then adjusted upward by a fixed value, which is a function of the relationship between the level-mean regression and the individual LAFmx values, to develop the energy-mean emission level equation. In previous REMEL studies, the adjustment from level-mean to energy-mean was computed using 0.115 2, where * is the standard error of the regression. However, due to the potentially non-Gaussian distribution of the level-mean data about its level-mean regression (the 0.115*2 adjustment assumes a Gaussian distribution), the following equation is used to compute the level-mean to energy-mean adjustment factor:

*E = 10 x log10[(1/n)*REi] - (1/n)*RLi           (dB)

For example:


*E = 10 x log10[(1/n) *RE] - (1/n) *RL = 0.649762

In the above, the RLi values represent the level residuals, which are equivalent to the value of each data point, i, at its corresponding speed, s, minus the value of regression at s; REi values represent the energy residuals, which are equivalent to 10(RLi/10); and n represents the total number of data samples.

This *E adjustment is then added to both the engine/exhaust term and the tire/pavement term of the L(s) equation, i.e., the C and B coefficients, as follows:

LE(s) = 10 x log10[10(C+*E)/10 + sA/1010(B+ *E)/10]           (dB)

From the above energy-mean emission-level regression equation, four input parameters are required to specify a user-defined vehicle type in the FHWA TNM: (1) a minimum level (the C coefficient plus *E); (2) a reference level (the emission level at 80 km/h or 50 mi/h); (3) the slope (the A coefficient); and (4) a like vehicle type. A like vehicle type is the FHWA TNM vehicle type to which the user-defined type is most similar. In determining a like vehicle type, the factors to be considered are listed in order of importance as follows: estimated subsource heights; estimated acceleration characteristics; and estimated, one-third octave-band frequency spectrum.(3,4)

Updated: 6/27/2017
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000