Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

Office of Planning, Environment, & Realty (HEP)
PlanningEnvironmentReal Estate

HEP Events Guidance Publications Awards Contacts

Validation of FHWA's Traffic Noise Model (TNM): Phase 1

Final Report August 2002

Continue on to Contents

FHWA-EP-02-031

DOT-VNTSC-FHWA-02-01

Prepared for
U.S. Department of Transportation
Federal Highway Administration
Office of Environment and Planning
Washington, DC 20590

Prepared by
U.S. Department of Transportation
Research and Special Programs Administration
John A. Volpe National Transportation Systems Center
Acoustics Facility, DTS-34
Cambridge, MA 02142-1093

Notice

This document is disseminated under the sponsorship of the Department of Transportation in the interest of information exchange. The United States Government assumes no liability for its contents or use thereof. This report does not constitute a standard, specification, or regulation.

The United States Government does not endorse products or manufacturers. Trade or manufacturers' names appear herein solely because they are considered essential to the object of this document.

Abstract

The Volpe Center Acoustics Facility, in support of the Federal Highway Administration (FHWA) and the California Department of Transportation (Caltrans), has been conducting a study to quantify and assess the accuracy of FHWA's Traffic Noise Model (TNM) and make recommendations on its use. The TNM Validation Study involves highway noise data collection and TNM modeling for the purpose of data comparison. Phase 1 of the study has been completed. For this phase, over 100 hours of traffic noise data were collected at 17 highway sites around the country. The 17 sites included: open areas next to the highway with acoustically soft ground (e.g., lawn); open areas with acoustically hard ground (e.g., pavement or water); and areas next to the highway with an open area behind a single noise barrier. In comparing the measured sound levels to the TNM-predicted sound levels, several variables were examined, including distance from the roadway, wind conditions, and percentage of heavy trucks. This report discusses theresults of Phase 1 of the TNM Validation Study.

Acknowledgments

The authors of this report wish to express their great appreciation to all involved in making the TNM Validation Study: Phase 1 a success. The list of contributors is long and appears in Appendix A - many thanks to all. We would like to specifically thank Bob Armstrong and Steve Ronning of the Federal Highway Administration, Office of Natural Environment, for essential support and guidance in making this a successful study. We also wish to thank the California Department of Transportation, Division of Environmental Analysis, in particular, Keith Jones, Rudy Hendriks, Bruce Rymer, Jim Andrews, and Bala Nanjundaiah, for providing their support and field measurement participation.

Executive Summary

Introduction

The Volpe Center Acoustics Facility (VCAF), in support of the Federal Highway Administration (FHWA) and the California Department of Transportation (Caltrans), has been conducting a study to quantify and assess the accuracy of FHWA's Traffic Noise Model® (TNM) and make recommendations on its use. The TNM Validation Study involves highway noise data collection and TNM modeling for the purpose of data comparison. The number of sites required to do a comprehensive study reflects the incorporation of numerous TNM features, either isolated or grouped with other TNM features. This large task is more manageable divided into multiple phases; in this manner, interim results are available to TNM users.

Phase 1 of the study has been completed. For this phase, over 100 hours of traffic noise data were collected at 17 highway sites around the country. The 17 sites included: open areas next to the highway with acoustically soft ground [e.g., field grass (effective flow resistivity (F) . 150 cgs Rayls) or lawn (F . 300 cgs Rayls)]; open areas with acoustically hard ground [e.g., pavement or water (F . 20,000 cgs Rayls)]; and areas next to the highway with an open area behind a single noise barrier. In comparing the measured sound levels to the TNM-predicted sound levels, several variables were examined, including distance from the roadway, wind conditions, and percentage of heavy trucks. A brief review of the study, including the results, is presented in this section of the report. For more details, please refer to the remainder of the report.

Field Measurements

Phase 1 measurement sites had characteristics of those most commonly modeled by TNM users and were relatively simplistic so as to isolate individual features of TNM. All 17 sites for Phase 1 were either open areas (i.e., free from interfering objects, reflective or absorptive, in the sound propagation path) or featured a noise barrier (wall or berm). Most sites were flat, the exceptions having ground undulations or substantial changes in elevation.

Instrumentation was deployed at each measurement site for capturing acoustical, meteorological, traffic, and site survey data. A-weighted equivalent sound levels in 5-second periods were captured using microphones, spectrum analyzers, sound levels meters, and digital audio tape recorders. One-second time intervals of temperature, relative humidity, wind speed and direction, and ambient atmospheric pressure were captured using automated meteorological stations. Highway traffic was continuously recorded using video cameras. A site survey was completed using a differential global positioning system. Other supporting instrumentation was also deployed. At each measurement site, approximately 6 hours of data were collected.

The types of sites and the locations of the acoustical and meteorological instrumentation are seen in the following table.

Table ES.1. Phase 1 Measurement Sites by Type.
Site Type Number of Sites Ranges of Microphone Distances

d=dist from rodway
bb=dist behind barrier

Open area acoustically soft ground
4
d=50 to 800 ft (~15 to ~245 m)
Open area acoustically hard ground
4
d=50 to 1273 ft (~15 to ~390 m)
Noise barrier
9
bb=50 to 300 ft (~15 to ~90 m)

TNM Modeling

Each measurement site was modeled using TNM. The input objects were taken directly from the site survey map and maps drawn during site scoping and measurements; these include all roadways, receivers, noise barriers, terrain lines, and ground zones. Once a TNM base case was xxicompleted for a particular site, a new run was created for each 5-minute data block. This amounted to as many as 70 TNM runs for each measurement site. For each 5-minute period, the corresponding traffic data (scaled from 5 minutes to 1 hour), temperature, and relative humidity were entered. All runs were then calculated, resulting in an hourly, A-weighted, equivalent sound level for each data block.

Data Analysis

After initial processing, the measured and TNM-predicted sound levels were imported into spreadsheets for analysis. For both sets of data, the 5-minute data blocks were logarithmically combined into 15-minute data blocks for final analysis and presentation.

The data sets were also processed in two ways: (1) the TNM-predicted sound levels were calibrated to the measured sound levels using a reference microphone so as to make a direct comparison of measured sound propagation and TNM-predicted sound propagation; and (2) the TNM-predicted sound levels were not calibrated to the measured sound levels so as to add another level of comparison, comparing measurements and predictions with possibly slightly different sound source characteristics. The calibration for the first way of processing was accomplished by applying a calibration value (the difference between a site's measured sound levels at the reference microphone and the TNM-predicted sound levels at the same position) to the predicted sound levels at all other positions. This calibration process eliminates biases due to possible site-specific emission levels.

Since TNM currently calculates sound levels for a windless environment, the data were further processed in two other distinct ways according to the wind speed. The two processing methods were: 1) no data blocks were discarded due to wind conditions (this data set is referred to as the all-wind data); and 2) any data blocks that at any time achieved a "very windy" condition [winds exceeded ~11 mph (5 m/s)] were removed (this data set is referred to as the strong-windremoved data). The process was assumed to eliminate data subjected to severe refraction and/or xxiipossible turbulence; it was also assumed that the removal of data characterized as "very windy" eliminated any data that may have been contaminated by wind noise at the microphone.

For final presentation the data were compared in several ways. First, direct comparisons of TNM-predicted sound levels and measured sound levels were made, then the differences as a function of the following variables were calculated: distance from the roadway, height above the ground, wind speed, wind direction, and percentage of heavy trucks. Additional analysis was performed using alternate TNM runs in order to make recommendations on the use of TNM.

Results

Overall, for the calibrated data, TNM is performing very well. The following graphic shows a direct comparison between TNM-predicted and measured sound levels for the strong-windremoved data. The TNM-predicted sound levels were calibrated to measured sound levels using a reference microphone. The data are plotted with the horizontal axis being the measured sound levels and the vertical axis being the TNM-predicted sound levels. Each 15-minute data block (15-min Leq) is represented as an orange X, where the number of data points is stated in the lower right corner of the figure. A dashed blue line represents the linear fit and solid green lines show the 95 percent confidence band. A solid black diagonal line symbolizes perfect agreement between TNM-predicted data and measured data. Data points that fall above (to the left of) this line indicate over-prediction and points that fall below (to the right of) this line indicate underprediction. It should be noted that the uncalibrated results (not shown in this graphic) indicate some over-prediction, but the bias is essentially eliminated after calibrating the TNM-predicted data using a reference microphone.

Direct Comparison of TNM and Measured Data; All Sites (calibrated);

Figure ES.1. Direct Comparison of TNM and Measured Data; All Sites (calibrated); Strong Wind Data Removed.

(Note: Data for 16 of the 17 measurement sites are shown in this plot; no data points for Site 04CT remained after eliminating the strong wind data.)

For all data comparisons with the calibrated data, TNM-predicted sound levels are showing good agreement with the measured sound levels for these types of sites: open area, acoustically soft ground sites [out to 800 ft (~245 m) from the roadway]; open area, acoustically hard ground sites [out to 300 ft (~90 m) from the roadway]; and noise barrier sites [out to 300 ft (~90 m) behind the barrier]. The only difference of concern arises for open area, acoustically hard ground sites at far distances. The uncalibrated data (where site bias has not been removed) shows a general over-prediction in the TNM-predicted sound levels.

As for the effects of wind, it is seen that TNM's accuracy is dependent on the wind conditions for noise barrier sites. Also, there seems to be no apparent influence of the percentage of heavy trucks on the performance of TNM, suggesting that TNM implements heavy trucks correctly.

In addition to the above comparisons, results for alternate TNM runs were examined in order to make recommendations on the use of TNM. A summary of the results and recommendations appear in the following tables.

Table ES.2. Summary of Results.
Investigation Results Comments
Direct comparison of TNM-predicted and measured sound levels uncalibrated all-wind data average 2.6 dB over-prediction when calibrating to reference mic, bias is essentially eliminated
Direct comparison of TNM-predicted and measured sound levels uncalibrated Strong-wind-removed data average 2.6 dB over-prediction when calibrating to reference mic, bias is essentially eliminated
Direct comparison of TNM-predicted and measured sound levels calibrated all-wind data average 1.0 dB
difference from
perfect agreeme
good agreement at all
types of sites, except for
far distances at hard
ground sites (some overprediction,
~ 2.0 dB); TNM
propagation algorithms are
performing well
Direct comparison of TNM-predicted and measured sound levels calibrated Strong-wind-removed data average 0.5 dB
difference from
perfect agreement
good agreement at all
types of sites, except for
far distances at hard
ground sites (some overprediction,
~ 2.0 dB); TNM
propagation algorithms are
performing well
Differences
(calibrated TNMpredicted
minus
measured) in
sound levels as a
function of ...
distance from
roadway,
height above
ground
all-wind data average
differences for
most sites within
1.5 to 2.0 dB –
some exceptions
far distances [> 300 ft (~90
m)] at hard ground sites
show some overprediction;
no strong
trends for height above
ground
Differences
(calibrated TNMpredicted
minus
measured) in
sound levels as a
function of ...
distance from
roadway,
height above
ground
Strong-wind-removed data average
differences for
most sites within
1.5 to 2.0 dB –
some exceptions
far distances [> 300 ft (~90
m)] at hard ground sites
show some overprediction;
no strong
trends for height above
ground
Differences
(calibrated TNMpredicted
minus
measured) in
sound levels as a
function of ...
wind speed,
wind direction
all-wind data 2.0 dB wind
influence at barrier
sites
only conclusive wind influence seen at barrier sites
Differences
(calibrated TNMpredicted
minus
measured) in
sound levels as a
function of ...
wind speed,
wind direction
Strong-wind-removed data 1.0 dB wind
influence at barrier
sites
only conclusive wind influence seen at barrier sites
Differences
(calibrated TNMpredicted
minus
measured) in
sound levels as a
function of ...
percentage of heavy trucks
(only for strong-wind-removed
data)
percentage of heavy trucks
(only for strong-wind-removed
data)
no distinct trends no apparent influence of %
heavy trucks on TNM’s
performance


Table ES.3. Recommendations on the Use of TNM.
Topic Recommendation
Data Calibration TNM-predicted sound levels should be calibrated to sound levels measured at a
site. Refer to example state policies on model calibration [Hendriks 1998]
[Lindeman 2001].
Ground Undulations Substantial ground undulations [$ 5 ft (1.5 m)] should be modeled.
Grass Medians Grass medians [with widths $ 10 ft (~3 m)] should be modeled using grass
ground zones (rather than the median being defined by the default ground type
of grass).
Ground Zones Sites with mixed acoustically soft and hard ground should be modeled with the
default ground type being soft ground and the ground zones being hard ground.

Later Phases of the TNM Validation Study

Later phases of the TNM Validation Study will incorporate more site measurements and modeling along with further analysis of the Phase 1 data. Additional measurement sites will incorporate Phase 1-type sites that need further investigation, sites with multiple TNM objects, and sites with less common TNM objects.

Some items discussed in Phase 1 require further investigation. These include: general TNM over-predictions that are seen in the uncalibrated results; TNM's accuracy being dependent on wind conditions at noise barrier sites; the examination of more open area sites with unusual ground surfaces to better evaluate TNM's performance in such situations; and the impact of different TNM-modeling techniques (different user methodologies).

To view PDF files, you need the Acrobat® Reader®.

Continue on to Contents

Updated: 07/15/2011
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000