How Weather Affects the Noise You Hear from Highways (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

80 Conclusions How does the weather affect the sound we hear from highways? As this research study has shown, the weather can have a substantial impact on highway sound propagation. Previous research has shown, and this research has confirmed, that sound from highways is affected in two primary ways by meteorology: • The absorption of noise in the atmosphere is affected primarily by temperature and humidity and its effect increases as the frequency of the sound increases. • Changes in the vertical temperature and wind gradient affect how sound refracts in the atmosphere. The effect generally increases with distance. In downwind conditions with wind increasing with height and temperature inversions, sound tends to bend toward the ground, generally increasing sound levels. In upwind conditions and temperature decreasing with height, sound is refracted upward, creating “shadow zones” where there are lower sound levels. Turbulence of the air scatters sound into the shadow zone, resulting in increased sound levels there. Barriers and buildings can create micro-meteorological conditions that affect the vertical wind gradient and turbulence. For example, on the leeward side of a highway noise barrier, increased wind shear and turbulence created by the barrier can lead to increased sound in the shadow zone. Sound Monitoring To confirm and expand upon these results, the research team collected sound and meteo- rological data from February 25 to March 8, 2017, along I-17 north of Phoenix, Arizona. Data were collected at a location with no highway noise barrier and at a location with a highway noise barrier. The Barrier location measurements were made on the last four days of the measurement campaign. At the No-Barrier location, sound level meters were placed 15, 30, 60, 120, 240, 480, and 960 meters from the highway and at the Barrier location, sound level meters were placed at 15 meters from the highway, and then behind the Barrier at 60, 90, 120, and 150 meters from the Barrier. Meteorological data were measured with two 10-meter meteorological masts at the No-Barrier location, a 3-meter mast at the Barrier location, and remote sensing devices for mea- suring temperature and wind profiles at a nearby school. No-Barrier Location At the No-Barrier location, the data clearly show a meteorological effect. The effect was great- est at the 480-meter position. For example, the difference in sound level between the morning and afternoon peak traffic hours was about 18 dB, despite similar traffic volume, vehicle mix, and speed during those two periods. The mornings were characterized by temperature inversions C H A P T E R 7 Conclusions and Suggested Research

Conclusions and Suggested Research 81 and the afternoons were characterized by unstable atmospheric conditions with temperatures decreasing with height. The data show that the sound speed profile is highly correlated with these effects. The sound speed profile combines the effects of the wind profile and temperature profiles. A negative sound speed gradient (decreasing sound speed with height) results in upward refraction and a posi- tive sound speed gradient results in downward refraction. When plotted against the difference in sound level relative to the 15-meter position, the data form an “S” curve, showing decreas- ing sound levels with negative sound speed gradients and increasing sound levels with positive gradients. However, there are several uncertainties inherent in the data that affect the results: • Atmospheric complexity. The measured sound speed gradient is for a single location, but the effect at any one location near a highway combines the impact of multiple atmospheric states (i.e., multiple combinations of wind direction, wind speed, and temperature). The analysis of the gradient only took into account what was occurring at two measurement heights, but the real gradient is nonlinear and more complicated than this. • Distributed source. The calculated effective sound speed only represents a single propagation direction (the direction of the line of microphones), while the other directions from other parts of the road are ignored. • Air turbulence. Turbulence due to wind shear and buoyancy will tend to increase sound levels in the shadow region. • Background noise. Though an extensive program was implemented, not all background noise could be filtered out of the data. • Spatial meteorological changes. There were two meteorological towers. While representative, there may be small meteorological variations along the propagation path. • Other measurement errors. There may be other small errors due to instrumentation and the temporal changes over time. For example, this compares an arithmetic average sound speed difference with an equivalent continuous average sound level, which may lead to the over- weighting of higher sound levels during any 5-minute period. • Source distance. The source is scattered over approximately 12 lanes of traffic with different lane distribution throughout the day. Despite these considerations, the effective sound speed difference used here is a useful quan- tity with a considerable predictive power for the sound level. Barrier Location The meteorological effect at the Barrier location was not as significant at the No-Barrier loca- tion. This indicates that the sound speed profile has a considerably smaller effect on the sound level behind the barrier, at least out to 150 meters, the furthest microphone position. The research team concluded that sound levels at positions well inside the acoustic shadow region behind a tall barrier, in this case with an effective height of 7.6 meters, are less affected by meteorological effects than sound levels at unscreened positions. Modeling The team modeled the sound levels from the highway using several approaches: Harmonoise. This model, developed in the EU, takes into account the refraction due to meteorology. The approach of Harmonoise, to curve the ground to represent effects of curved sound rays, is a possible approach for modifying TNM to account for meteorological effects. The research team modeled the project area around I-17 in Harmonoise using six wind classes and five stability classes. The modeling showed meteorological effects down to -25 dB and up to 14 dB. Classified into the same meteorological classes, the measurement at I-17 had a lower range, on

82 How Weather Affects the Noise You Hear from Highways the order of -6 to +8 dB, although the measurements conducted did not cover all of the combi- nations of Harmonoise wind and stability classes. Variations in vehicle mix and ground type did not have a significant impact on the Harmonoise modeled results. Nord2000. NCHRP Report 791 lists meteorological effects for several combinations of dis- tance, vector wind speeds, temperature gradients, vehicle mixes, and ground conditions. This model had a similar range of meteorological effects as Harmonoise, from -21 to + 15 dB, although with a different distribution as a function of distance. There is good agreement between the mea- surements made at I-17 and the model within 120 meters for the temperature gradient table except under strong inversions. The measurements results indicate much smaller changes com- pared to neutral conditions than the levels modeled in the NCHRP Report 791 report. Distances beyond 120 meters show the highest differences, especially for strong upwind, moderate upwind, and strong inversion conditions. However, the measurements did not cover all of the combina- tions of NCHRP Report 791 wind and temperature gradient classes. The NCHRP Report 791 Nord2000 modeling tables indicate that the barrier is less effective during inversion conditions, which is contrary to the measured adjustments within the dis- tances measured. The microphone positions, out to 150 meters from the barrier, were in the acoustical shadow zone. Under strong inversions or downwind conditions, the sound rays bend downward, but not enough to affect the sound levels within this distance. In addition, during the measurements, there may have been less turbulence and wind shear than was assumed in the models. Both increase the level of noise in the shadow zone. The NCHRP Report 791 tables treat wind and temperature effects separately. Thus, combinat- ing these effects cannot be estimated with these tables. This limitation is addressed in the lookup tables presented in Chapter 3. FHWA TNM. TNM assumes a nonrefracting atmosphere. For comparison, researchers com- pared the TNM model to the results found under completely neutral conditions. This was difficult because just two of the 5-minute measurements periods were defined as neutral. Initially, the results showed that TNM overpredicted the actual sound levels. The research team concluded this was due to the rubberized asphalt pavement used on the I-17. After adjusting for this effect, the TNM model predictions improved, with increasing over-prediction as distance increased from the highway. The best agreement occurred at the Barrier location. PE modeling. PE models are accurate numerical models for atmospheric sound propaga- tion. The research team compared the results of the Harmonoise model with the PE model. The Harmonoise model gave results that were in reasonable agreement with the PE model. While there were differences between the two for a nonturbulent atmosphere, these differences were largely eliminated in a turbulent atmosphere. It is not clear why the models show a larger meteorological effect than the measurements do. Part of the explanation may be that strong downward refracting conditions were not included in the measurements, so the effect may have been greater had these been included. Another explana- tion could be that the turbulence strength was higher during the measurements than assumed for the calculations. Also, at the larger distances decreases in measured noise could be smaller than predicted since we are approaching background levels. This is an area that needs further research. Conceptual Models and Tools Models One objective in this research was to provide guidance on how to quantify meteorological effects on roadway noise propagation. Four approaches that could be used to quantify meteo- rological effects were found.

Conclusions and Suggested Research 83 • Research models, such as PE models, are useful in that they provide highly accurate results when the downrange and vertical sound speed profile can be quantified. However, they are not appropriate for regulatory use, as they are difficult to implement, require extensive meteoro- logical data, and require significant processing power. However, research models are helpful in validating engineering approaches, such as done in this study and others [117] and developing algorithms for engineering models. • Statistical models can be used to estimate the meteorological effect by correlating the differ- ence between the expected and observed sound levels with various meteorological parameters. In this study, various models were developed using such parameters as distance, wind and temperature gradients, several representations of atmospheric stability, AERMET parameters, and calculated sound speed gradients. The best fit was found using the measured sound speed gradient at two heights, with distance interactions. The model was able to explain 91% of the variation in sound level at the No-Barrier location. However, the statistical model was less appropriate for the Barrier location. Given the complex relationship between the barrier, micro-meteorology, and the size and shape of the shadow zone with barrier position and height, the research team concluded that a statistical approach is not suitable for estimat- ing meteorological effects in the presence of barriers or other obstructions unless additional variables can be identified and quantified. However, given the fast computational times of statistical models, they could be used as first-order approximations at No-Barrier locations. • Lookup tables are useful when the effect can be described by two or three variables. Many variables for lookup tables were considered for meteorological effects, including temperature and vector wind gradients, such as used in NCHRP Report 791, various stability classifications such as day/night with cloud cover, Pasquill-Gifford stability, and inverse Obukhov length. The research team moved forward with a lookup table using measured values of vector wind speed at 10 meters and temperature profile. This is like the approach used by NCHRP Report 791, but it uses different cut points for the temperature profile and considers the combined effect of temperature gradient and wind speed. Note that vector wind speed is used and not the vertical gradient of vector wind speed, since the former is easier to measure and roughly correlated to the gradient near the ground (since the theoretical wind speed at the ground is zero). This table had a range of meteorological effect of -9 to +7 dB. Lookup tables were also developed in the cases where only vector wind speed or temperature gradient is known and a table that shows the meteorological effect if the effective sound speed gradient near the ground is known. Since these tables were based on measured rather than modeled data, results for strong downwind conditions are missing (since this condition did not occur during the monitoring). • Engineering models such as Harmonoise and Nord2000 were considered as the basis for engineering tools to estimate the meteorological effect. However, both models showed more extreme effects than were measured at I-17 at both the Barrier and No-Barrier locations. However, the approaches used for these models, such as curving the ground to account for the curved sound rays, could be used in a future upgrade to TNM to account for meteorological refraction. • Updating TNM to include meteorological effects will involve additional research that is not included in the scope of this research project. However, several approaches to this were con- sidered in the study, including integration of the methods in this list. All approaches, except incorporation of research models, were found to be feasible. If TNM is updated using one or more of these methods, validating it using the data collected in this and other studies, and with research models where possible is suggested. AERMET Tool Early on in this study, the researchers identified a disconnect between what data are needed to estimate the meteorological effect and what data are available. In particular, the vector wind and temperature profiles needed to estimate the effective sound speed profile cannot be found easily

84 How Weather Affects the Noise You Hear from Highways in NWS websites or other weather data depositories. However, all states use the AERMOD air quality dispersion model and its required met data files from AERMET. Therefore, the researchers developed a tool to calculate effective sound speed profiles from AERMET data files. This AERMET Sound Speed Profiler tool is a spreadsheet that uses variables from an AERMET file for any location and then calculates an effective sound speed profile for every hour of the year. The output is a series of charts showing the effective vector wind gradient, temperature gradient, and resulting effective sound speed gradient averaged for each hour of the day. The user can specify the wind direction and height above ground for the linear gradient. This tool is available on the TRB website. The results from this tool can then be used to determine how frequent various meteorologi- cal conditions exist that lead to higher sound levels than predicated by TNM, for example. This can be helpful for outreach efforts, explaining these effects in response to complaints of highway noise during periods of downward refraction, and for developing inputs to the statistical models and lookup tables developed here. Implications for Noise Impact Assessments In the United States, modeling highway noise for regulatory purposes is currently done under acoustically neutral atmospheric conditions (no wind or temperature effects). However, this research has shown that noise impacts would generally increase under conditions of downward refraction and decrease under upward refraction. Under the current procedures for determining noise impacts, the sound level impacts would be estimated to be the same for widening projects. This is because the same adjustments would apply to both the existing and future sound levels. However, for new alignment projects, where there is no existing roadway, sound level increases could be higher and impact distances longer than under neutral conditions. Meteorological effects may also affect the worst noise hour determination. Typically, the worst noise hour occurs with the highest traffic volume under near free-flow conditions. How- ever, meteorological effects could alter these. As an example, in this study, the AM peak hour adjacent to I-17 was more than 10 dB above the PM peak hour (at 480 meters), even though both had similar levels of traffic and vehicle mix running at or near the speed limit. Conclusions on noise abatement could also change. The research showed that the highway noise barrier would be less effective under upwind and positive lapse conditions and much less likely to meet the reasonableness criteria in an example SHA noise policy. However, future research is needed to evaluate meteorological impacts in the presence of barriers and how that could affect appropriate designs under the FHWA regulations. The results indicate that the barrier designed for the worst noise hour under neutral conditions is more effective under downwind and inversion conditions. That is, insertion losses are higher, which increases the number of benefited residences. Therefore, a proposed barrier could be more feasible and reasonable when evaluated under downwind and inversion conditions. Guidance and Implementation SHAs currently identify noise impacts and evaluate noise abatement under neutral conditions under existing FHWA noise regulations. The FHWA regulation does not currently allow SHAs to consider meteorological effects in the noise impact and abatement analysis process. However, SHAs could evaluate meteorological effects to better understand how sound levels and noise bar- rier performance could change. Practitioners can also use meteorological adjustment to correct measurements made under nonneutral conditions for use in model validation.

Conclusions and Suggested Research 85 SHAs could also use look-up tables—like those developed for this research or those included in NCHRP Report 791—to better understand the effects at a study location. As discussed previously, the research team believes that adjustments based on measured data are most accurate. SHAs could use the noise impact and abatement spreadsheet developed for this research. This tool allows the user to input the most appropriate adjustments and noise impact and abatement criteria and obtain meteorological data from the NWS’s ASOS to evaluate the frequency of different conditions at a study location. SHAs can use data pro- cessed for air quality modeling through AERMET, along with the AERMET Sound Speed Profile Tool developed for this project to identify the frequency of upward and downward refractive conditions in an area. Finally, SHAs could include information on meteorological effects in noise reports and pro- vide estimates of sound levels under different conditions. SHAs could use the paragraphs con- tained in the brochure developed for this research as the basis for a section of the noise study report or could simply provide the brochure in an appendix. The discussion could also reference the interactive tool and provide a link. Any discussion should state clearly that sound levels under neutral conditions are the basis for noise impacts and noise abatement conclusions per the FHWA noise regulation. Outreach Materials An objective of this research was developing best practices and guidance to help explain to the public the ways in which meteorology affects highway noise. To accomplish this, the research conducted a literature search of existing materials that SHAs can use (see Chapter 6). These include: • FHWA’s Noise Barrier Design Handbook [107], • Caltrans’ “Highway Traffic Noise Fundamentals” [109], and • The European Environmental Agency noise mapping website [120] In addition, the research team developed two products that SHAs can customize for the own use when conducting public outreach: • A four-page folded brochure, in MS Word format, titled “Why is it so loud today? Under- standing how weather affects traffic noise levels in your community.” The brochure explains the basic relationships between sound levels and meteorological conditions so that a layperson can quickly read the brochure and understand the concepts. SHAs can customize any part of the brochure to meet their own needs, including the use of their own logos and contact information. • This interactive tool, “Why is it so loud today,” expands on the contents of the brochure in three modules – The Relationship Between Weather and Traffic Noise Levels – Hearing the Sound Level Difference – The Effect on Communities The tool contains audio files to allow the users to hear the differences in the sound due to the meteorological effect and contains useful animations. As with the brochure, SHAs can customize the slide deck, including adding and deleting slides, and inserting their own logos and contact information. The Interactive Tool is available on the TRB website. The website also includes a slide deck that describes the research project. This slide deck includes animations showing how temperature and wind profiles change throughout the course of the monitoring period, and video of the equipment at each microphone position.

86 How Weather Affects the Noise You Hear from Highways Suggestions The results of this research study clearly show that the weather can affect traffic noise levels. This research study has created tools to quantify these effects and help explain these effects to the public. The research team offers these suggestions: • SHAs download and customize the brochure from the TRB website and make this available to the public via their website, or in response to complaints. • SHAs use the AERMET Sound Speed Profiler Tool developed in this study to analyze the frequency of adverse meteorological effects in their major metropolitan areas. • SHAs use the Interactive Tool PowerPoint presentation to explain how weather could affect the noise one hears from highways when doing public outreach. • For highway noise barriers proposed in areas conducive to inversions and other weather con- ditions favorable to sound propagation, consider that meteorology plays a role in barrier effectiveness and explain those effects in noise study reports and public involvement materials. • If requested, make these tools and materials available to SHAs and FHWA for training courses, such as the National Highway Institute’s “Highway Traffic Noise” course (NHI 142051). • If so desired, SHAs and FHWA may develop policies to evaluate meteorological effects for the purpose of understanding and explaining these to the public. • Conduct further research on meteorological effects. Suggested Research The research results show the potential for the development of tools, models, and policy to take into account meteorological effects in highway noise analyses. The research team suggests the following topics for further research: • Data collection. The data collected under this study was weighted toward highly stable and unstable atmospheres. Additional data is needed to help validate models under more neutral meteorological conditions. Specifically, these would include conditions with more cloud cover, lower solar elevation, and higher wind speeds. Also, instead of in-line monitoring for noise and weather as has been accomplished in the past, a monitoring array could be considered. • Lookup tables. Where these data are not available, use modeling to expand the lookup tables for various types of roads and weather conditions using modeled data. • Modeling tool. The AERMET Sound Speed Profile Tool developed under this project can be combined with the statistical model to estimate the frequency of sound pressure level differ- ences for any distance under a no-barrier scenario. This would be relatively straightforward, since the statistical models in this study were developed based on AERMET parameters. Further development could also include frequency-specific effects by 1/3 octave band. • Impacts tool. This research study developed an Impacts, Feasibility, Reasonableness Spreadsheet Tool to analyze feasibility and reasonableness of highway noise barriers under various meteorological conditions. It was used internally to conduct the impact assessment in Chapter 5. With additional research on meteorological effects in the presence of barriers, this tool should be updated and developed further for public release. • Barrier impacts. As noted in the study, creating a lookup table or statistical model in the pres- ence of highway noise barriers is more complicated. Barriers effects will depend on barrier height, the setback of the barrier from the highway, distance, and the change in wind charac- teristics created by the barrier. Additional research is needed to measure meteorological effects under more varied roadway/barrier configurations. This could be done under controlled con- ditions, given the difficulty of finding suitable research locations behind barriers. The research would have the objective to create some type of model to help planners in barrier design.

Conclusions and Suggested Research 87 • Follow-up evaluations. Evaluations of additional sites using a range of feasibility and reason- ableness assumptions to expand understanding of the effects on noise impacts and abatement conclusions. • Development of an engineering model. Given that there is a foundation of engineering mod- els that take into account refraction (i.e., Nord2000 and Harmonoise), research can begin to develop a similar approach for modifying the FHWA TNM model to account for meteoro- logical effects using physical principals, rather than statistical or lookup-table approaches (if so desired). More complex or computational intensive approaches could be considered in the future as algorithms develop and PC processing power increases. • Further exploration of differences between modeled and measured noise levels under upward refracting conditions. In this project, it was found that under upward refracting conditions the Harmonoise model yields sound levels that are considerably lower than the measured sound levels. It is of great interest to explore the possible origins of this differ- ence (atmospheric profiles, turbulence, ground). Although Harmonoise is based on some engineering approximations, calculations with the more rigorous PE model also yield in general low sound levels under upward refracting conditions, and the PE levels are of the same order as the Harmonoise levels. A possible approach could be to complement high- way noise measurements with point- and/or line-source (loudspeaker) measurements and compare the results with model results.