Field Evaluation of Reflected Noise from a Single Noise Barrier—Phase 1 (2016)

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.

41 C H A P T E R 4 Findings and Applications FHWA Method Findings Data were collected under a fair range of meteorological classes, as shown in Table 11, with the number of equivalent five-minute period groupings by site by meteorological class. A grouping represents three or more five-minute periods with equivalent source data (based on the reference microphone level and the average speed by direction of travel) and meteorological class. Reference will be made to the meteorological class as appropriate in the discussion of the findings. Table 11. Number of equivalent groupings by location by meteorological class. Meteorological Class Number of equivalent groups I-24 Briley I-90 SR-71 MD-5 Upwind Lapse 31 Calm Lapse 13 4 Calm Neutral 4 4 4 10 Calm Inversion 31 15 Downwind Lapse 12 15 Downwind Neutral 6 7 Total 48 39 16 6 47 Based on the data collected and analyses procedures described above, the findings are outlined individually below. Finding 1: Measured broadband unweighted sound pressure levels and A-weighted sound levels are generally higher at the Barrier microphones than at the No Barrier microphones. As a first step in analyzing the data, the running Leq(5min) for each microphone pair at Barrier and No Barrier sites were graphed and level difference plots developed. These graphs and analysis give an overall picture of the measured levels, both in terms of unweighted and A-weighted sound levels. These graphs are prior to any grouping of the five-minute periods by source and meteorological class equivalence. As an example, the BarRef01 and NoBarRef02 levels at I-24 are shown in Figure 24 (unweighted). Then, Figure 25 shows the differences in both the unweighted and A-weighted levels for this microphone pair. A positive value means the Barrier level was higher than the No Barrier level. Similar graphs for all microphone pairs at all locations are in Appendix B. Table 12 presents the approximate range of differences in the Barrier and No Barrier running Leq(5min) for all of these pairs. These ranges are termed “approximate” because the data are prior to any attempt to group the five-minute periods into equivalent periods based on source and meteorological class. Since the MD-5 data were collected in two separate periods, ranges for both daytime and nighttime are shown for MD-5. Level differences are shown for both unweighted and A-weighted data. In general, the ranges in the differences were generally greater for the unweighted levels than for the A-weighted levels. The

42 reason is probably due to the sampling including 1/3 octave bands ranging from 20 Hz to 10 kHz. There tends to be more variation in levels in the very low and high bands—from sources ranging from heavy trucks on to insects—which would affect a broadband unweighted level but get filtered out of broadband A-weighted level calculation. Each location is discussed below. Figure 24. Running Leq(5min), I-24, unweighted sound pressure level, dBZ, BarRef01 and NoBarRef02. Figure 25. Differences in running Leq(5min), I-24, BarRef01 minus NoBarRef02. 83 84 85 86 87 88 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 So un d Pr es su re L ev el , d BZ Time BarRef01 NoBarRef02 -3 -2 -1 0 1 2 3 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 Di ffe re nc e in le ve l, dB Time dBA dBZ

43 Table 12. Approximate range of differences in the Barrier and No Barrier running five-minute Leq for all locations (Barrier minus No Barrier). Microphone Pair Type of Level Approximate Range of Level Differences by Location (dB) I-24 Briley I-90 SR-71 MD-5 (day) MD-5 (night) BarRef01 minus NoBarRef02 Unweighted 0.5 to 1.5 n/a 0 to 1 0 to 1.8 -1 to 1.8 -2 to 2.2 A-weighted 0.5 to 1.5 n/a 0 to 1 0 to 1 -0.5 to 0.5 -1.5 to 0.7 BarCom03 minus NoBarCom05 Unweighted 0 to 1 0 to 2 0 to 1.5 -0.9 to 1 -1 to 2 -2 to 1.5 A-weighted 0 to 1 -3 to -1 0.4 to 1.3 -0.7 to 0.5 0.5 to 2.4 0 to 1.5 BarCom04 minus NoBarCom06 Unweighted 0 to 0.5 -1 to 1 -0.7 to 1.5 0 to 4 -0.5 to 1.8 -1 to 2 A-weighted -0.5 to 0.5 -1.5 to 0.5 0.2 to 1 1 to 3.9 -0.5 to 1 -0.5 to 1 I-24 For virtually all of the running Leq(5min) periods, the BarRef01 levels, both unweighted and A-weighted, were higher than the NoBarRef02 levels by a range of 0.5 to 1.5 dB. These higher levels were expected because the BarRef01 microphone was located halfway between the barrier and I-24. For a large majority of the running Leq(5min) periods at the community lower microphones, the BarCom03 levels, both unweighted and A-weighted, were higher than the NoBarCom05 levels by a range of 0.0 to 1.0 dB, with some differences as much as 1.5 dB. For most of the running Leq(5min) periods at the upper microphones, the BarCom04 levels, both unweighted and A-weighted, were higher than the NoBarCom06 levels by a range of 0.0 to 0.5 dB, with some differences as much as 1.0 dB. For the other periods, the A-weighted levels at NoBarCom06 are 0 dB to 0.5 dB higher than the BarCom04 levels. Briley In general, the running Leq(5min) decreased over time as the traffic decreased from the evening rush hour into the later evening. At the lower microphones, the unweighted sound pressure levels at BarCom03 were typically higher than at NoBarCom05, from a few tenths of a decibel to just over 2 dB. However, the A-weighted sound levels at BarCom03 were generally 1.5 dB to 2 dB lower than the NoBarCom05 levels in the first three hours of the measurement and 2 dB to 3 dB lower in the last hour. The results were different for the upper microphones. The differences in the unweighted sound pressure levels at BarCom04 and NoBarCom06 varied from positive to negative over most of the measurement period and became generally negative (NoBarCom06 higher than BarCom04) later in the evening. The A-weighted sound levels at NoBarCom06 tend to be slightly higher than at BarCom04 in the early part of the measurement, with the difference increasing as the measurement period moved later into the evening. Insect and frog noise from trees near NoBarCom05 and NoBarCom06 became major sound contributors starting early in the evening. I-90 For the reference microphones, both the unweighted and A-weighted running Leq(5min) at BarRef01 were on the order of 0 dB to 0.5 dB above the NoBarRef02 levels for the first two hours of measurement (13:00 to 15:00). For the second half of the measurements (15:00 to 17:20), this difference increased to a range of 0.5 dB to 1.0 dB. The BarRef01 microphone was located atop the barrier and the barrier was just off the shoulder. It is speculated that the slightly higher levels at BarRef01 could be due to sound

44 reflections off the barrier and then off the sides of the vehicles back to the microphone, especially for tractor trailer bodies. At the lower community microphones, for all of the running Leq(5min) periods, the BarCom03 unweighted sound pressure levels were on the order of 0 dB to 1.5 dB higher than the NoBarCom05 levels. For the A-weighted sound levels, the BarCom03 levels were on the order of 0.4 dB to 1.3 dB higher than the NoBarCom05 levels. At the upper and slightly farther back microphones, for most of the running Leq(5min) periods, the BarCom04 unweighted levels ranged from 0.7 dB lower than NoBarCom06 to 1.5 dB higher. The A- weighted levels ranged from 0.2 dB to 1 dB higher. At all of the microphones, as time passed the Leq(5min) dropped slowly, on the order of 1 dB to 2 dB, over the 4-hour period. In this same time period the differences in levels between the Barrier and No Barrier microphone pairs increased on the order of a half decibel. SR-71 For the reference microphones, the unweighted running Leq(5min) at BarRef01 were on the order of 0 dB to 1.8 dB higher than the NoBarRef02 levels, averaging roughly 1 dB higher. The A-weighted levels at BarRef01 were on the order of 0 dB to 1 dB higher than NoBarRef02, averaging roughly 0.5 dB. Higher levels at BarRef01 were expected because the microphone was positioned between the barrier and the road. For the microphones just off the shoulder on the opposite side from the barrier, little evidence of reflection was seen in these broadband data. The unweighted running Leq(5min) at BarCom03 ranged from 0.9 dB lower to 1 dB higher than those at NoBarCom05. The A-weighted running Leq(5min) at BarCom03 ranged from 0.7 dB lower to 0.5 dB higher than those at NoBarCom05. With these microphones so close to the far lanes of traffic, relative to the distance from BarCom03 to the barrier, little increase in level due to reflections was expected. At the distant community microphones, for virtually all of the running Leq(5min) periods, the BarCom04 levels, both unweighted and A-weighted, were higher than the NoBarCom06 levels. The unweighted levels ranged mostly from 0 dB to 4 dB higher than NoBarCom06. The A-weighted levels ranged from 1 dB to nearly 4 dB higher. For both unweighted and A-weighted cases, the average difference was 2.1 dB higher at BarCom04. MD-5 For the reference microphones, the running Leq(5min) at BarRef01 and NoBarRef02 are roughly comparable. Unweighted levels at BarRef01 ranged mostly from 2 dB below NoBarRef02 levels to 2.2 dB above them. A-weighted levels were within ±0.5 dB of each other during the afternoon session. However, due to frog noise near NoBarRef02, its evening A-weighted levels were generally higher than the BarRef01 levels. Little difference in the levels was expected because the BarRef01 microphone was positioned atop the barrier, although reflections off the vehicle bodies might increase its levels, as was discussed for the I-90 location. For the lower community microphones opposite the barrier, the daytime unweighted running Leq(5min) at BarCom03 ranged from 1.0 dB lower to 2 dB higher than those at NoBarCom05. The daytime A- weighted levels at BarCom03 generally ranged from 0.5 dB to 2.4 dB higher than those at NoBarCom05. In the evening, the unweighted levels at the two microphones were roughly within -2 dB to 1.5 dB of each other. The BarCom03 A-weighted levels ranged mostly from 0 dB to 1.5 dB higher than the NoBarCom05 levels. For most of the running Leq(5min) periods during both the afternoon and evening, the BarCom04 levels were higher than the NoBarCom06 levels. The unweighted Leq(5min) generally ranged from 0.5 dB lower to 1.5 dB higher during the day and -1 dB lower to 2 dB higher during the evening. The A-

45 weighted levels ranged from 0.5 dB lower to 1 dB higher than NoBarCom06 during both daytime and nighttime. Finding 2: The differences in Barrier and No Barrier levels are frequency-specific and vary by location and site. There are clear examples of enhanced levels opposite the barrier compared to the corresponding No Barrier position. To see the differences by frequency band more clearly, graphs were developed that show the differences in Leq(5min) between comparable microphones for an average of all of the equivalent five- minute periods in a particular meteorological class, with their error bars. The error bars are +/- one standard deviation for each average value. Each graph shows the averages of the average level differences for the A-weighted sound level, the unweighted sound pressure level and the 1/3 octave band sound pressure levels from 20 Hz to 10 kHz. Similar plots for each group of equivalent five-minute periods are in spreadsheet files in the project record. The trends across the 1/3 octave band frequencies are generally similar in these individual groups of equivalent periods, with some differences likely related to background noise and the uniqueness of vehicle noise sources in the five-minute periods in each group. Shown in Figure 26 is a sample plot of sound pressure level spectra for the MD-5 BarCom03 and NoBarCom05 microphones. Then, shown in Figure 27 is the frequency-based average level difference graph for all of the equivalent groups in the Calm Inversion class. There are higher levels at BarCom03 in the bands from 200 Hz to 500 Hz (with the maximum at 5 dB higher at 250 Hz and 315 Hz), as well as 0.5 dB to 1 dB higher levels in the 800 Hz to 2.5 kHz bands. The 4 kHz band is 6 dB higher at NoBarCom05 than BarCom03 due to localized frog noise. The large barrier effect in the 200 Hz to 500 Hz bands in Figure 27 is evidence of a loss of some of the ground effects in these bands that is indicated by the “dip” in the No Barrier spectrum in Figure 28. More is said on this effect in Finding 8. More examples will be shown in the findings that follow, and a complete location-by-location presentation of all of the results is in Appendix B.

46 Figure 26. Sample sound pressure level spectra for BarCom03 and NoBarCom05, MD-5, Calm Inversion Group CIG-3-4, 23:15 (Leq(5min), dBZ). . 25 30 35 40 45 50 55 60 65 70 75 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k So un d Pr es su re Le ve l, dB Z 1/3 Octave Band Frequency, Hz BarCom03 NoBarCom05

47 Figure 27. Averages of the differences in Leq(5min) +/- one standard deviation (dB), all microphones, for all Calm Inversion groups, MD-5. Finding 3: Sound levels are higher and spectral content changes at a position between the barrier and the road, compared to the No Barrier site, as evidenced at I-24 and SR-71. At I-24 and SR-71, the BarRef01 microphones were set midway between the roadway and the noise barrier. As described in Section 3, the SR-71 barrier is 13 ft tall, consisting of a 7-ft concrete block wall atop a 6-ft high berm and located 50 ft from the center of the near travel lane; the BarRef01 microphone was set 25 ft from the center of the near travel lane and 10 feet above the roadway plane. At I-24, the barrier is 19 feet tall approximately 96 ft from the center of the near travel lane. The BarRef01 microphone was set 51 ft from the center of the near travel lane (33 ft from the edge of shoulder) and 45 ft in front of the barrier at a height of 10 ft above the roadway plane. At both locations, the microphones were in the paths of reflected sound. Figure 28 shows the ungrouped running five-minute A-weighted Leq data for BarRef01 and NoBarRef02 at I-24. Earlier, Figure 24 showed the unweighted Leq data for the same. Figure 29 shows the differences plots for both the A-weighted and unweighted levels for the same periods. Figure 30 then shows the differences plot for the SR-71 reference microphones’ running five-minute Leq data. At both locations, BarRef01 A-weighted and unweighted levels are higher than at NoBarRef02 by 0.5 dB to 1.5 dB. -10 -8 -6 -4 -2 0 2 4 6 8 10 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All CIG Groups

48 Figure 28. Running Leq(5min), I-24, A-weighted sound level, dBA, BarRef01 and NoBarRef02. Figure 29. Differences in running Leq(5min), I-24, BarRef01 minus NoBarRef02. 78 79 80 81 82 83 84 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 So un d Le ve l, dB A Time BarRef01 NoBarRef02 -3 -2 -1 0 1 2 3 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 Di ffe re nc e in le ve l, dB Time dBA dBZ

49 Figure 30. Differences in running Leq(5min), SR-71, BarRef01 minus NoBarRef02. Figure 31 and Figure 32 then show typical sound pressure level spectra respectively, for I-24 (an Upwind Lapse five-minute period) and SR-71 (a Downwind Neutral five-minute period). The broadband A-weighted and unweighted levels are on the left side of the graphs. Both sets of spectra show that the Barrier and No Barrier differences are frequency-specific. -3 -2 -1 0 1 2 3 9: 00 9: 10 9: 20 9: 30 9: 40 9: 50 10 :0 0 10 :1 0 10 :2 0 10 :3 0 10 :4 0 10 :5 0 11 :0 0 11 :1 0 11 :2 0 11 :3 0 11 :4 0 11 :5 0 12 :0 0 12 :1 0 12 :2 0 12 :3 0 12 :4 0 12 :5 0 13 :0 0 13 :1 0 13 :2 0 Di ffe re nc e in le ve l, dB Time dBA dBZ

50 Figure 31. Sample sound pressure level spectra for BarRef01 and NoBarRef02, I-24, Upwind Lapse group ULG-3-2, 13:26-13:31 (Leq(5min), dBZ). 40 45 50 55 60 65 70 75 80 85 90 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k So un d Pr es su re Le ve l, dB Z 1/3 Octave Band Frequency, Hz BarRef01 NoBarRef02

51 Figure 32. Sample sound pressure level spectra for BarRef01 and NoBarRef02, SR-71, Downwind Neutral group DNG-3-2, 11:38-11:43 (Leq(5min), dBZ). These spectral differences are better seen in the 1/3 octave band differences plots. For I-24, Figure 33 shows the 1/3 octave band differences of the average of all of the average differences for the Upwind Lapse meteorological class. In general, the BarRef01 levels are roughly 0.9 dB to 1.3 dB higher than the NoBarRef02 levels across the entire spectrum. At 25 Hz, the difference is 2 dB; at 200 Hz and 250 Hz, it is approximately 0.5 dB. For SR-71, Figure 34 shows the 1/3 octave band differences plot of the average of all of the average differences for the Downwind Neutral meteorological class. The BarRef01 levels are higher than the NoBarRef02 levels across virtually the entire spectrum, with exception of 8 kHz and 10 kHz. The BarRef01 levels are higher by 3 dB at 31.5 Hz, 2.5 dB at 125 Hz and 1.5 dB at 2.5 kHz. In the range from 400 Hz to 1.25 kHz, the differences are less than 0.5 dB. 40 45 50 55 60 65 70 75 80 85 90 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k So un d Pr es su re Le ve l, dB 1/3 Octave Band Frequency, Hz BarRef01 NoBarRef02

52 Figure 33. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarRef01 minus NoBarRef02, for all Upwind Lapse groups, I-24. Figure 34. Averages of the differences in Leq(5min) +/- one standard deviation (dB), all microphones, for all Downwind Neutral groups, SR-71. Finding 4: The background sound pressure level is elevated in the presence of the noise barrier at the microphone position between the barrier and the road. There is evidence that background level increased at the BarRef01 position in front of the barrier for both I-24 and SR-71. For I-24, Figure 35 presents the L90(5min) and L99(5min) for BarRef01 and NoBarRef02, in terms of overall A-weighted sound levels and unweighted sound pressure level. The upper graphs are L90 (A- weighted on the left and unweighted on the right).The lower graphs are L99 (A-weighted on the left and unweighted on the right). Then, Figure 36 presents the differences in L90(5min) and L99(5min) along with Leq(5min),computed as BarRef1 minus NoBarRef2 for the A-weighted sound levels. -4 -2 0 2 4 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All ULG Groups -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All DNG Groups

53 The results show that while the Leq(5min) averages about 1 dB higher at BarRef01 than at NoBarRef02, the L90 and L99 at BarRef01 are much higher than at NoBarRef02. This effect on these two descriptors is evidence of an increase in the background level in front of the barrier that could be attributed to the presence of reflected sound rays reaching the microphone in addition to the direct rays from the passing vehicles. The results support the idea of not only a reflection component during the moment of passage, but also approach and receding components of the reflections, as illustrated earlier in Figure 1. Thus, the sound level rises sooner and recedes later for each vehicle. This sustaining of the sound for a single vehicle overlaps with the same pattern for other vehicles, elevating the background level and reducing the time during which the background level might decrease between vehicle passages. This hypothesis is supported by the spectrogram findings to be discussed later. Figure 35. L90(5min) and L99(5min), I-24, BarRef01 and NoBarRef02 – unweighted and A-weighted sound levels.

54 Figure 36. Differences in broadband A-weighted 5-min L90, L99 and Leq, I-24, BarRef1 and NoBarRef2. Figure 37 presents the SR-71 differences in L90(5min) and L99(5min) along with Leq(5min), computed as BarRef1 minus NoBarRef2 for the A-weighted sound levels. Much like the I-24 data, the results show that while the Leq(5min) averages about 0 dB to 1 dB higher at BarRef01 than at NoBarRef02, the L90(5min) and L99(5min) at BarRef01 are much higher than at NoBarRef02: L90 by as much as 4 dB and L99 by as much as 7 dB. These differences are strong evidence of an increase in the background level in front of the barrier that could be attributed to the presence of reflected sound rays from the passing vehicles reaching the microphone in addition to the direct rays, producing a sustained sound that keeps the background level from dropping off during gaps between vehicles. -6 -4 -2 0 2 4 6 8 10 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 So un d Le ve l d iff er en ce , d B Time L90 L99 Leq

55 Figure 37. Differences in broadband A-weighted 5-min L90, L99 and Leq, SR-71, BarRef01 and NoBarRef02. The above graphs are for the broadband A-weighted sound levels and unweighted sound pressure levels only. Figure 38 for I-24, shown below, broadens the analysis to include the individual 1/3 octave bands through the use of color shading. The brown color means that the BarRef01 levels are higher than the NoBarRef02 levels and blue means that NoBarRef02 is higher. In the graph, time runs from top to bottom (increasing as one moves down each figure, with each row representing the starting minute of a running five-minute period) and the total block representing approximately four hours. The 1/3 octave bands run across from left to right, with the broadband A- weighted sound levels and unweighted sound pressure levels on the far left. Within each band’s column of data are the differences for seven Ln sound pressure level Ln values (L1, L5, L10, L33, L50, L90 and L99) and Leq in the order illustrated in Figure 39 for a single 1/3 octave band. Over 57,000 Ln sound pressure level differences are represented in the figure by color shading. In Figure 38 for I-24, vertical brown streaks are on the right sides of the data columns (representing L90(5min) and L99(5min)) in the frequency bands from 400 Hz up through 2 kHz for most of the sample period and up through 5 kHz for the first half of the sample period. These brown streaks mean that the BarRef01 background levels are higher than the NoBarRef02 background levels, evidence of a sustaining of a vehicle’s passby noise due to the creation of an image source for each vehicle as the sound reflects off the barrier. In contrast, the vertical blue streaks in the 8 kHz band are evidence of elevated background levels at the NoBarRef02, likely attributable insect noise in the vegetation behind this position. -6 -4 -2 0 2 4 6 8 10 9: 00 9: 10 9: 20 9: 30 9: 40 9: 50 10 :0 0 10 :1 0 10 :2 0 10 :3 0 10 :4 0 10 :5 0 11 :0 0 11 :1 0 11 :2 0 11 :3 0 11 :4 0 11 :5 0 12 :0 0 12 :1 0 12 :2 0 12 :3 0 12 :4 0 12 :5 0 13 :0 0 13 :1 0 13 :2 0 So un d Le ve l D iff er en ce , d B Time L90 L99 Leq

56 Figure 38. I-24 Differences in Ln(5min) by 1/3 octave frequency bands: BarRef01 and NoBarRef02. Figure 39. Order of statistical levels for a single 1/3 octave band A similar pattern is seen in Figure 40 for SR-71 between BarRef01 and NoBarRef02. Vertical brown streaks are on the right sides of the data columns (representing L90(5min) and L99(5min)) in the frequency bands from 500 Hz up through 4 kHz for nearly all of the sample period, and up through 5 kHz for the first half of the sample period. These brown streaks mean that the BarRef01 background levels are higher than the NoBarRef02 background levels. The BarRef01 levels were also higher in the 20 Hz to 31.5 Hz bands across most of the descriptors for most of the measurement period. The reason for that difference in those very low frequency bands is not apparent. Figure 40. SR-71 Differences in Ln(5min) by 1/3 octave frequency bands: BarRef01 and NoBarRef02.

57 Finding 5: Even at the reference microphone position atop the barrier, the level can be slightly higher than at the equivalent No Barrier position, as evidenced at I-90. However, little difference was seen at MD-5. The I-90 and MD-5 noise barriers are located close to the edge of shoulder of the road. The BarRef01 microphones were placed 5 feet directly above the top of the barriers. Figure 41 shows the differences in the unweighted and A-weighted running Leq(5min) between BarRef01 and NoBarRef02 at I-90. In general, both the unweighted sound pressure levels and A-weighted sound levels are higher at the Barrier microphones than at the No Barrier microphones. For the I-90 reference microphones, both the unweighted and A-weighted running Leq(5min) at BarRef01 are on the order of 0 dB to 0.5 dB above the NoBarRef02 levels for the first two hours of measurement (13:00 to 15:00). For the second half of the measurements (15:00 to 17:20), this difference increased to a range of 0.5 dB to 1.0 dB. The slightly higher levels at BarRef01 could be due to sound reflections off the barrier and then off the sides of the vehicles, especially for heavy truck trailers. There are mixed results for the MD-5 reference microphones. Figure 42 shows the running Leq(5min) at BarRef01 and NoBarRef02 to be roughly equal. Unweighted levels at BarRef01 ranged mostly from 1.5 dB below NoBarRef02 levels to 2 dB above. A-weighted levels were within ±0.5 dB of each other during the afternoon session. However, due to frog noise near NoBarRef02, its evening A-weighted levels were generally about 1 dB higher than the BarRef01 levels. Little difference in the levels was expected because the BarRef01 microphone was positioned atop the barrier, although reflections off the vehicle bodies might increase its levels, as noted above for I-90. Figure 41. Differences in running Leq(5min), I-90, BarRef01 minus NoBarRef02. -3 -2 -1 0 1 2 3 9: 00 9: 10 9: 20 9: 30 9: 40 9: 50 10 :0 0 10 :1 0 10 :2 0 10 :3 0 10 :4 0 10 :5 0 11 :0 0 11 :1 0 11 :2 0 11 :3 0 11 :4 0 11 :5 0 12 :0 0 12 :1 0 12 :2 0 12 :3 0 12 :4 0 12 :5 0 13 :0 0 13 :1 0 13 :2 0 Di ffe re nc e in le ve l, dB Time dBA dBZ

58 Figure 42. Differences in running Leq(5min), MD-5, BarRef01 minus NoBarRef02. The Calm Neutral meteorological class spectral difference plot for the I-90 reference microphone is shown as Figure 43. In general, the BarRef01 levels are 0 dB to 1 dB higher than the NoBarRef02 levels at 400 Hz and below. Above 400 Hz up through 3.15 kHz, these levels are 0.5 dB to 1 dB higher than the NoBarRef02 levels. Above 4 kHz, the No Barrier levels are higher, likely due to localized insect noise. These results are similar to the I-24 reference microphone results, where that microphone was placed between the barrier and the road. The corresponding plot for the MD-5 reference microphones in Figure 44 shows the results for the Downwind Neutral class. In general, the BarRef01 levels vary little compared to NoBarRef02 from 500 Hz through 6.3 kHz. Below 500 Hz, the BarRef01 levels were generally less than a decibel above those at NoBarRef02. The Downwind Neutral time periods were in the afternoon before the high-frequency frog noise began at NoBarRef02. -3 -2 -1 0 1 2 3 Di ffe re nc e in le ve l, dB Time dBA dBZ

59 Figure 43. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarRef01 minus NoBarRef02, for all Calm Neutral groups, I-90. Figure 44. Averages of the differences in Leq(5min) +/- one standard deviation (dB), all microphones, for all Downwind Neutral groups, MD-5. There was no evidence of the sustaining or elevating of the background level at BarRef01 for both I-90 and MD-5 based on the difference graphs for L90(5min) and L99(5min), both in terms of the broadband A- weighted levels and the 1/3 octave bands. The supporting figures and discussion are in Appendix B. Finding 6: Near the edge of the road for the lower-height microphones, the BarCom03 levels are roughly the same as the NoBarCom05 levels, being slightly higher in the very low frequency bands, as evidenced at SR-71. At the SR-71 location, the BarCom03 and NoBarCom05 microphones were located just off the shoulder of the road, 25 feet from the center of the near travel lane, at heights of 10 feet above the roadway surface. These were the closest community microphones to the road on the side opposite the barrier of all of the studied locations. Figure 45 shows the differences in the broadband unweighted and A-weighted levels for BarCom03 and NoBarCom05 for SR-71. For these broadband measures, little evidence of reflection is seen. The -4 -2 0 2 4 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All CNG Groups -4 -2 0 2 4 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All DNG Groups

60 unweighted levels at BarCom03 range from 0.9 dB lower to 1 dB higher than those at NoBarCom05. The A-weighted levels at BarCom03 range from 0.7 dB lower to 0.5 dB higher than those at NoBarCom05. With these microphones so close to the far lanes of traffic, relative to the distance from BarCom03 to the barrier, little increase in level due to reflections was expected. Figure 45. Differences in running Leq(5min), SR-71, BarCom03 minus NoBarCom05. Figure 46 then shows the averages of the differences in the BarCom03 and NoBarCom05 levels for all of the Downwind Neutral groups. The levels at BarCom03 are generally higher than or the same as those at NoBarCom05. The levels in the frequency bands from 20 Hz up through 125 Hz are 0.5 dB to 1.5 dB higher at BarCom03. From 160 Hz up through 1.6 kHz, the levels are different by only 0.5 dB or less. From 2 kHz through 5 kHz, the BarCom03 levels are a half decibel higher than NoBarCom05. The NoBarCom05 levels are less than 1.5 dB higher than BarCom03 at and above 6.3 kHz. -3 -2 -1 0 1 2 3 9: 00 9: 10 9: 20 9: 30 9: 40 9: 50 10 :0 0 10 :1 0 10 :2 0 10 :3 0 10 :4 0 10 :5 0 11 :0 0 11 :1 0 11 :2 0 11 :3 0 11 :4 0 11 :5 0 12 :0 0 12 :1 0 12 :2 0 12 :3 0 12 :4 0 12 :5 0 13 :0 0 13 :1 0 13 :2 0 Di ffe re nc e in le ve l, dB Time dBA dBZ

61 Figure 46. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarCom03 minus NoBarCom05 for all Downwind Neutral groups, SR-71. Finding 7: Near the edge of the road for the lower-height microphones, there is some evidence of an increase in the background A-weighted sound level on the order of 1 dB to 1.5 dB at BarCom03, as evidenced at SR-71. Figure 47 presents the differences in L90(5min) and L99(5min) along with Leq(5min) for the background A-weighted sound levels, computed as BarCom03 minus NoBarCom05 along SR-71. There is some evidence of the elevated background level at BarCom03 compared to NoBarCom05 even though these two microphones are very close to the edge of the shoulder for the far-lane traffic across from the barrier. While the Leq(5min) averages about ± 2 dB compared to NoBarCom05, the L90 at BarCom03 are, on average about a decibel higher than NoBarCom05 and the L99 at BarCom03 average approximately 1.5 dB higher. For both descriptors, there are also many times when the NoBarCom05 levels are higher than the BarCom03 levels. This variation in level differences is likely related to the 50-ft setback of the barrier from the center of the near lane on the opposite side of the highway and differences in traffic from one data block to the next. Regarding the barrier setback, the near-lane traffic would tend to cause the highest levels at the nearby BarCom03 and NoBarCom05 microphones, followed by the far-lane traffic and then any far-lane reflections, followed by any near-lane reflections. Regarding traffic, because there is some difference in distance between the Barrier and No Barrier sites, the exact same vehicles are not passing each mic in each 5-minute period. Also, there could be operational differences of the vehicles at the two sites, such as speed and lane changes Nonetheless, it is interesting that, on average, the trend is for the L90 and L99 to be higher at the Barrier mic. As observed when listening to the audio recordings, one senses the “presence” of the barrier at the Barrier microphone. -4 -3 -2 -1 0 1 2 3 4 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All DNG Groups

62 Figure 47. Differences in broadband A-weighted 5-min L90, L99 and Leq, SR-71, BarCom03 minus NoBarCom05. Figure 48 presents the spectral Ln differences for BarCom03 and NoBarCom05. Not a great deal of difference is seen between the descriptors for the two microphones, which is consistent with the A- weighted sound level graphs. However, there does appear to be a slight increase in the L90 and L99 in the bands in the 1 kHz to 3.15 kHz bands, as evidenced by the brown streaks on the right side of the data columns for those bands. Figure 48. SR-71 Differences in Ln(5min) by 1/3 octave frequency bands: BarCom03 minus NoBarCom05. -4 -3 -2 -1 0 1 2 3 4 5 6 9: 00 9: 10 9: 20 9: 30 9: 40 9: 50 10 :0 0 10 :1 0 10 :2 0 10 :3 0 10 :4 0 10 :5 0 11 :0 0 11 :1 0 11 :2 0 11 :3 0 11 :4 0 11 :5 0 12 :0 0 12 :1 0 12 :2 0 12 :3 0 12 :4 0 12 :5 0 13 :0 0 13 :1 0 13 :2 0 So un d Le ve l D iff er en ce , d B Time L90 L99 Leq

63 Finding 8: Farther back from the road, but still within 100 ft, the Barrier levels are higher than No Barrier levels by 0.5 to 1.5 dB and the spectrum is changed by even more in some of the frequency bands between 250 Hz and 630 Hz and in some of the bands above 800 Hz, as evidenced at I-24, I-90, and MD-5. At the I-24 location, the BarCom03 and BarCom04 microphones were set back 84 ft from the center of the near travel lane and at heights of 5 ft and 15 ft above the roadway plane, with NoBarCom05 and NoBarCom06 at corresponding positions. At the I-90 location, the BarCom03 and NoBarCom05 microphones were set back 69 ft from the center of the near travel lane and at a height of 10 ft above the roadway plane. BarCom04 and NoBarCom06 were set back 93 ft from the center of the near travel lane and at a height of 17 ft above the roadway plane. At the MD-5 location, the BarCom03 and BarCom04 microphones were set back 80 ft from the center of the near travel lane and also at heights of 5 ft and 15 ft above the roadway plane. Figure 49 shows the differences in the unweighted and A-weighted levels for the lower-height BarCom03 and NoBarCom05 microphones at I-24. For a large majority of the running five-minute Leq periods, the BarCom03 levels, both unweighted and A-weighted, are higher than the NoBarCom05 levels by a range of 0.0 to 1.0 dB, with some differences as much as 1.5 dB. For I-90 at BarCom03 and NoBarCom05, Figure 50 shows the differences in the unweighted and A- weighted levels. For all of the running five-minute Leq periods, the BarCom03 unweighted sound pressure levels are on the order of 0 dB to 1.5 dB higher than the NoBarCom05 levels. For the A- weighted sound levels, the BarCom03 levels are on the order of 0.4 dB to 1.3 dB higher than the NoBarCom05 levels. For MD-5, Figure 51 shows the differences in the unweighted and A-weighted levels for the lower- height BarCom03 and NoBarCom05 microphones opposite the barrier. The daytime unweighted levels at BarCom03 ranged from 1.0 dB lower to 2 dB higher than those at NoBarCom05. The A-weighted levels at BarCom03 range from 0.5 dB to 1.5 dB higher than those at NoBarCom05. In the evening, the unweighted levels at the two microphones were comparable. The BarCom03 A-weighted levels ranged mostly from 0 dB to 1 dB higher than the NoBarCom05 levels.

64 Figure 49. Differences in running Leq(5min), I-24, BarCom03 minus NoBarCom05. Figure 50. Differences in running Leq(5min), I-90, BarCom03 minus NoBarCom05. -3 -2 -1 0 1 2 3 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 Di ffe re nc e in le ve l, dB Time dBA dBZ -3 -2 -1 0 1 2 3 13 :0 0 13 :1 0 13 :2 0 13 :3 0 13 :4 0 13 :5 0 14 :0 0 14 :1 0 14 :2 0 14 :3 0 14 :4 0 14 :5 0 15 :0 0 15 :1 0 15 :2 0 15 :3 0 15 :4 0 15 :5 0 16 :0 0 16 :1 0 16 :2 0 16 :3 0 16 :4 0 16 :5 0 17 :0 0 17 :1 0 17 :2 0 Di ffe re nc e in le ve l, dB Time dBA dBZ

65 Figure 51. Differences in running Leq(5min), MD-5, BarCom03 minus NoBarCom05. Next, Figure 52 presents an example of the sound pressure level spectra for BarCom03 and NoBarCom05 (lower and closer microphones to the far side of the road) at I-90 for a Calm Neutral period. The BarCom03 levels are noticeably greater in the 250 Hz to 500 Hz 1/3 octave bands. -4 -3 -2 -1 0 1 2 3 4 12 :0 0 12 :3 0 13 :0 0 13 :3 0 14 :0 0 14 :3 0 15 :0 0 15 :3 0 16 :0 0 19 :4 9 20 :1 9 20 :4 9 21 :1 9 21 :4 9 22 :1 9 22 :4 9 23 :1 9 Di ffe re nc e in le ve l, dB Time dBA dBZ

66 Figure 52. Sample sound pressure level spectra for BarCom03 and NoBarCom05, I-90, Calm Neutral class, CNG-1-1, Period 15:37 (Leq(5min), dBZ). The differences in the Barrier and No Barrier levels can be seen in the following figures. Figure 53 shows the differences in level between the BarCom03 and NoBarCom05 microphones at I-24 for an average of all of the Upwind Lapse groups. Both of these microphones were five feet above the roadway plane. In general, the BarCom03 levels are equal to or slightly greater than the NoBarCom05 levels over most of the frequency range up through 4 kHz. The increase is less than 1 dB from 31.5 Hz to 250 Hz, and on the order of 1 dB to 2 dB in the bands from 315 Hz to 1 kHz. Above 4 kHz, the levels at NoBarCom05 are higher than the levels at BarCom03. That high-frequency difference was caused by insects in the vegetation behind the NoBarCom05 microphone that were not present near the BarCom03 site. 35 40 45 50 55 60 65 70 75 80 85 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k So un d Pr es su re Le ve l, dB Z 1/3 Octave Band Frequency, Hz BarCom03 NoBarCom05

67 Figure 53. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarCom03 minus NoBarCom05, for all Upwind Lapse groups, I-24. Figure 54 shows the averages of the averages of the differences in the BarCom03 and NoBarCom05 levels for the I-90 location for all of the measured Calm Neutral groups. The BarCom03 and NoBarCom05 microphones were both 69 feet from the center of the near travel lane and 10.4 feet above the roadway surface. The levels in the frequency bands from 20 Hz up through 80 Hz were 0.5 dB to 1 dB higher at BarCom03. For 1 kHz and higher, the BarCom04 levels were approximately 1 dB to 2 dB higher than those at NoBarCom05. The most noticeable differences were in the 250 Hz to 500 Hz bands, where the levels were 2.5 dB to 5 dB (at 400 Hz) higher. Figure 54. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarCom03 minus NoBarCom05, for all Calm Neutral groups, I-90. The next location with microphones at two heights is MD-5. All of the groupings of five-minute periods that were judged equivalent for traffic parameters at the MD-5 location fell into four meteorological classes: Downwind Neutral and Downwind Lapse (daytime) and Calm Neutral and Calm Inversion (evening). The potential difference by meteorological class will be addressed later in this chapter. For now, typical sound pressure level spectra are shown in Figure 55 for BarCom03 and -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All ULG Groups -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All CNG Groups

68 NoBarCom05 for one of the five-minute periods in the Calm Inversion meteorological class. The increase in levels in the mid-range and higher-range frequencies is similar to that observed for the I-24 and I-90 locations. Figure 55. Sample sound pressure level spectra for BarCom03 and NoBarCom05, MD-5, Calm Inversion Group CIG-3-4, 23:15 (Leq(5min), dBZ). Figure 56 shows the level difference averages for all of the Calm Inversion classes at the MD-5 location. The results for early-evening Calm Neutral class evening periods are similar. The graph shows the higher BarCom03 levels in the bands from 200 Hz to 500 Hz (with the maximum at 5 dB higher at 250 Hz and 315 Hz). The 4 kHz band is 6 dB higher at NoBarRef02 than BarRef01 due to loud, localized frog noise. A possible explanation for the barrier effect being prominent in the low frequency range (250 to 500 Hz) for BarCom03 at I-90 (Figure 54) and MD-5 (Figure 56) is that direct and reflected sound take different propagation paths. The direct sound at both the Barrier and No Barrier sites is likely experiencing ground effects/wave interference that cause a dip in sound level in that frequency range, as is illustrated in the sample spectra in Figure 52 and Figure 55. The reflected sound at the barrier site is experiencing a different propagation path than the direct sound, with different ground effects and wave 25 30 35 40 45 50 55 60 65 70 75 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k So un d Pr es su re Le ve l, dB Z 1/3 Octave Band Frequency, Hz BarCom03 NoBarCom05

69 interference with ground reflections; a dip in the 250 to 500 Hz range could be non-existent or diminished. As a result, the barrier effect would be pronounced in the 250 to 500 Hz range. Figure 56. Averages of the differences in Leq(5min) +/- one standard deviation (dB), all microphones, for all Calm Inversion groups, MD-5. Finding 9: Farther back from the road, but still within 100 ft, background level increases in the bands from 630 Hz up through 3.15 kHz, as evidenced at I-90 and MD-5, but not I-24. First, for I-24, Figure 57 presents the differences in L90(5min) and L99(5min) along with Leq(5min) for the A-weighted sound levels, computed as BarCom03 minus NoBarCom05. There is not much evidence of the elevated background level at BarCom03 than at BarRef01, not too unexpected given the dominance of the direct sound from the nearby vehicles. For I-90, Figure 58 presents these same differences. In this case, there is strong evidence of the elevated background level at BarCom03. While the Leq(5min) averages about 0.5 dB to 1 dB higher than NoBarCom05, the L90 at BarCom03 are 1 dB to 2 dB higher and the L99 at BarCom03 are 1 dB to 4 dB higher. For MD-5, Figure 59 presents the same differences, computed as BarCom03 minus NoBarCom05. There is evidence of the elevated background level at BarCom03 during the daytime hours. While the Leq(5min) averages about 0.5 dB to 1 dB higher than NoBarCom05, the L90(5min) at BarCom03 range from 9 dB lower to 10 dB higher than NoBarCom05, averaging approximately 2 dB higher. Almost all of the daytime L99(5min) are higher at BarCom03 than NoBarCom05, evidence of an increase in the background level due to reflected sound off the barrier. As noted at the other locations, none of these levels have been edited for contaminating sounds. One possible reason for the background level increase at the I-90 and MD-5 Barrier microphones and not at the I-24 Barrier microphone is that the barrier at MD-5 and I-90 sits just at the edge of the shoulder, while the I-24 back is set back nearly 100 ft. As a result, there is more likelihood of a sustaining of the pass-by signal due to the reflected sound at the close-in barriers, which elevates the background level. In the evening at the MD-5 location, the clear trend was for the L90(5min) and L99(5min) at NoBarCom05 to grow larger relative to BarCom03 as the evening got later. This trend is a clear result of the increased level and constancy of frog and insect noise. -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All CIG Groups

70 Figure 57. Differences in A-weighted 5-min L90, L99 and Leq, I-24, BarCom03 and NoBarCom05. Figure 58. Differences in A-weighted 5-min L90, L99 and Leq, I-90, BarCom04 and NoBarCom06. -6 -4 -2 0 2 4 6 8 10 13 :1 3 13 :2 3 13 :3 3 13 :4 3 13 :5 3 14 :0 3 14 :1 3 14 :2 3 14 :3 3 14 :4 3 14 :5 3 15 :0 3 15 :1 3 15 :2 3 15 :3 3 15 :4 3 15 :5 3 16 :0 3 16 :1 3 16 :2 3 16 :3 3 16 :4 3 16 :5 3 17 :0 3 17 :1 3 So un d Le ve l d iff er en ce , d B Time L90 L99 Leq -6 -4 -2 0 2 4 6 8 10 13 :0 0 13 :1 0 13 :2 0 13 :3 0 13 :4 0 13 :5 0 14 :0 0 14 :1 0 14 :2 0 14 :3 0 14 :4 0 14 :5 0 15 :0 0 15 :1 0 15 :2 0 15 :3 0 15 :4 0 15 :5 0 16 :0 0 16 :1 0 16 :2 0 16 :3 0 16 :4 0 16 :5 0 17 :0 0 So un d Le ve l D iff er en ce , d B Time L90 L99 Leq

71 Figure 59. Differences in broadband A-weighted 5-min L90, L99 and Leq, MD-5, BarCom03 and NoBarCom05. Figure 60 shows an example of the 1/3 octave band Ln differences for BarCom03 and NoBarCom05 at I-90 as an example. Similar graphs for the other locations are in Appendix B. The brown color in the 250- 500 Hz bands indicate an increase in all of the Ln descriptors means the Barrier levels are higher than the No Barrier levels, which was discussed in Finding 8 (the blue color means the No Barrier levels are higher). The vertical brown streaks on the right sides of the data columns in the frequency bands from 630 Hz up through 3.15 kHz indicate that the BarCom03 background levels are higher than the NoBarCom05 background levels. Figure 60. I-90 Differences in Ln(5min) by 1/3 octave frequency bands: BarCom03 and NoBarCom05. -6 -4 -2 0 2 4 6 8 10 So un d Le ve l D iff er en ce , d B Time L90 L99 Leq

72 Finding 10: Back at 400 ft from the road, the BarCom04 levels are typically 1 dB to 4 dB higher than at the No Barrier site, as evidenced at SR-71. Figure 61 shows the differences in the unweighted and A-weighted levels for the BarCom04 and NoBarCom06 microphone pair at SR-71. These microphones were the most distant from the road in the study, at 400 ft. For virtually all of the running five-minute Leq periods, the BarCom04 levels, both unweighted and A-weighted, are higher than the NoBarCom06 levels. The unweighted levels range from 0 dB to 5.5 dB higher than NoBarCom06. The A-weighted levels range from 1 dB to 3.7 dB higher. For both unweighted and A-weighted cases, the average difference was 2.1 dB higher at BarCom04. Figure 61. Differences in running Leq(5min), SR-71, BarCom04 minus NoBarCom06. All of the groupings of five-minute periods that were judged equivalent for the reference Leq and average speeds at the SR-71 location fell into one meteorological class: Downwind Neutral. Figure 62 presents the sound pressure level spectra for BarCom04 and NoBarCom06. The BarCom04 levels are higher in all bands except 100 Hz to 200 Hz where NoBarCom06 is higher. Note that terrain differences between the two sites can affect results below 500 Hz. Differences below 500 Hz are likely attributable to a combination of terrain differences and barrier effects. A simplified FHWA TNM analysis showed that, for some of the frequencies below 500 Hz, the BarCom04 sound levels should be lower due to ground effects. Please refer to the spectrogram results for Site SR-71 in Appendix B for more information. Figure 63 shows the averages of the differences in the distant BarCom04 and NoBarCom06 levels for all of the Downwind Neutral groups. The levels in the frequency bands from 20 Hz up through 80 Hz were 2 dB to 4 dB higher at BarCom04 compared to NoBarCom06. Then, from 315 Hz through 8 kHz, the BarCom04 levels are 1.5 dB to 3 dB higher than NoBarCom06. In the range of 100 Hz through 250 Hz, the NoBarCom06 levels range from 0 dB to 3 dB (at 200 Hz) higher than the BarCom04 levels. -3 -2 -1 0 1 2 3 4 5 6 9: 00 9: 10 9: 20 9: 30 9: 40 9: 50 10 :0 0 10 :1 0 10 :2 0 10 :3 0 10 :4 0 10 :5 0 11 :0 0 11 :1 0 11 :2 0 11 :3 0 11 :4 0 11 :5 0 12 :0 0 12 :1 0 12 :2 0 12 :3 0 12 :4 0 12 :5 0 13 :0 0 13 :1 0 13 :2 0 Di ffe re nc e in le ve l, dB Time dBA dBZ

73 Figure 62. Sample sound pressure level spectra for BarCom04 and NoBarCom06, SR-71, Downwind Neutral group DNG-3-2, 11:38-11:43 (Leq(5min), dBZ). 25 30 35 40 45 50 55 60 65 70 75 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k So un d Pr es su re Le ve l, dB 1/3 Octave Band Frequency, Hz BarCom04 NoBarCom06

74 Figure 63. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarCom04 minus NoBarCom06, for all Downwind Neutral groups, SR-71. Finding 11: Back at 400 ft from the road, all of the Ln descriptors were higher at the BarCom04 site, not just the background levels, as evidenced at SR-71. Figure 64 presents the differences in L90(5min) and L99(5min) along with Leq(5min) for the A-weighted sound levels, computed as BarCom04 minus NoBarCom06. Both of these figures start 45 minutes into the measurement period. The was a great deal of what turned out to be roofing nail gun noise that was audible at NoBarCom06 during this period, Rather than trying to edit it all out of the data, this period was just deleted from this Ln analysis. The results are different for this microphone pair than most of the other pairs across the various study locations because these microphones are the farthest from the road. The BarCom04 Leq(5min) ranged from 2.5 dB to 3.8 dB higher than the NoBarCom06 level for the first 23 minutes of the period shown on the figure. During this time, the meteorological class was Calm Neutral and the L90(5min) and L99(5min) differences ranged from 2 dB to 5 dB higher at BarCom04 than at NoBarCom06. During the last three hours, the Leq(5min) difference became more variable – 0.5 dB to 2.5 dB higher at BarCom04. During this period, L90(5min) differences also became more variable, being 0 to 3.5 dB higher at BarCom04. The L99(5min) became even more variable, with the BarCom04 values ranging from 1 dB lower than those at NoBarCom06 to 5.4 dB higher. During this time period, the meteorological class was Downwind Neutral. On average over the full measurement period, the BarCom04 Leq(5min), L90(5min) and L99(5min) were 1.7 dB, 2.0 dB and 2.1 dB higher than at NoBarCom06. These results, taken together, suggest that the overall levels from the traffic noise are higher at the Barrier site, but because the traffic is 400 ft away, there is less overall rise and fall to the levels compared to being in close to the road. As a result, there is little chance for lulls in the noise under the studied traffic flows. Perhaps nighttime measurements when the flow is much lower might show that elevating of the background level at a distant site across from a barrier. Figure 65 presents the spectral Ln differences for BarCom04 and NoBarCom06. A pattern can be seen of higher broadband A-weighted levels at BarCom04 in the low and mid-to-upper bands, with higher NoBarCom06 levels in the 100 Hz to 250 Hz bands as well as the highest frequency bands. (Note that terrain differences between the two sites can affect results below 500 Hz; please refer to the SR-71 spectrogram results in Appendix B for more information.) This pattern applies across most of the Ln descriptors, not just L90(5min) and L99(5min). -8 -6 -4 -2 0 2 4 6 8 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All DNG Groups

75 Figure 64. Differences in broadband A-weighted 5-min L90, L99 and Leq, SR-71, BarCom04 and NoBarCom06. Figure 65. SR-71 Differences in Ln(5min) by 1/3 octave frequency bands: BarCom04 and NoBarCom06. Finding 12: The increase in levels due to reflections decreased by 1 dB to 2 dB going from a lower-height microphone to a higher microphone, as evidenced at I-24, I-90, and MD-5. At I-24, BarCom03 and NoBarCom05 were 5 ft above the roadway plane. BarCom04 and NoBarCom06 were 15 ft above the roadway, at the same distant back as the lower microphones. Figure 66 shows the differences in level between the microphone pairs for an average of all of the Upwind Lapse groups. The top graph shows the differences in levels between BarCom03 and NoBarCom05. In general, the BarCom03 levels are equal to or slightly greater than the NoBarCom05 levels over most of the frequency range up through 4 kHz. The increase is less than 1 dB from 31.5 Hz to 250 Hz, and on the order of 1 dB to 2 dB in the bands from 315 Hz to 1 kHz. Above 4 kHz, the levels at NoBarCom05 are -2 -1 0 1 2 3 4 5 6 9: 45 9: 55 10 :0 5 10 :1 5 10 :2 5 10 :3 5 10 :4 5 10 :5 5 11 :0 5 11 :1 5 11 :2 5 11 :3 5 11 :4 5 11 :5 5 12 :0 5 12 :1 5 12 :2 5 12 :3 5 12 :4 5 12 :5 5 13 :0 5 13 :1 5 13 :2 5 So un d Le ve l D iff er en ce , d B Time L90 L99 Leq

76 higher than the levels at BarCom03. The difference was caused by insects in the vegetation behind the NoBarCom05 microphone that were not present near the BarCom03 site. The lower graph compares the levels at BarCom04 and NoBarCom06. The results show that the BarCom04 levels in the frequency bands from 20 Hz up through 1.25 kHz were equal to or slightly higher than at NoBarCom06. At 31.5 Hz to 63 Hz, they were approximately 1 dB higher than NoBarCom06. At 1.6 kHz and above, the NoBarCom06 levels were higher than the BarCom04 levels ranging from a fraction of a decibel at 1.6 kHz to 2.5 dB in the 6.3 kHz band. The higher levels at NoBarCom06 in the high-frequency bands are attributed to insect noise in some vegetation behind this microphone. Figure 66. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarCom03 and NoBarCom05 (top) and BarCom04 and NoBarCom06 (bottom), for all Upwind Lapse groups, I-24. For I-90, Figure 67 shows the averages of the differences in the Barrier and No Barrier microphones’ levels for all of the Calm Neutral groups. The upper graph compares the levels at BarCom03 and NoBarCom05, the lower-height microphones, both of which were 69 feet from the center of the near travel lane and 10.4 feet above the roadway surface. The levels in the frequency bands from 20 Hz up through 80 Hz were 0.5 dB to 1 dB higher at BarCom03. For 1 kHz and higher, the BarCom04 levels were approximately 1 dB to 2 dB than those at NoBarCom05. The most noticeable differences were in the 250 Hz to 500 Hz bands, where the levels were 2.5 dB to 5 dB (at 400 Hz) higher. -4 -2 0 2 4 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All ULG Groups -4 -2 0 2 4 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All ULG Groups

77 The lower graph compares the levels at BarCom04 and NoBarCom06, both of which were 93 feet from the center of the near lane and 17 feet above the roadway surface. The levels in the frequency bands from 31.5 Hz up through 250 Hz were 0.5 dB to 1 dB higher at BarCom04. From 1.25 kHz to 3.15 kHz, the BarCom04 levels were approximately 0.5 dB higher. The most noticeable differences were in the 315 Hz to 630 Hz bands, where the levels were 1.5 dB to 3 dB (at 400 Hz) higher. Figure 67. Averages of the differences in Leq(5min) +/- one standard deviation (dB), BarCom03, BarCom04, NoBarCom05 and NoBarCom06, for all Calm Neutral groups, I-90. Similar results were seen at MD-5 where the lower microphones were 5 ft above the roadway plane. Graphs for that data are in Appendix B, as well as graphs and details of the Ln analysis at I-24, I-90 and MD-5. Although this idea requires further consideration, one possible conclusion that can be drawn from this finding is that barrier reflection effects may be more pronounced closer to the ground due to sound- reducing propagation effects. Closer to the ground, shielding from median barriers/vehicles and ground effects can reduce sound levels at various frequencies. When an opposing barrier is added, the reflected noise likely has energy at those same frequencies, and this energy can become exposed. For receivers -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All CNG Groups -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz All CNG Groups

78 higher above the ground, the noise-reducing propagation effects are decreased, so the barrier-reflected noise may be partially masked or diminished. Finding 13: No effect on the sound level differences was seen as a function of traffic volume, as evidenced at all microphone pairs. For all sites and among all groups within a meteorological class, there was little correlation found between changes in traffic volumes and changes in the differences in Leq(5min). Although, the Calm Inversion group actually showed a roughly 300% change in Factored Hourly traffic volume across all of the equivalent groups, meaningful conclusions about correlations between traffic volumes and the differences in Leq(5min) could not be established. Also, in general, the range in speeds for each class was too small to address any relationship between speed sound level difference. The ranges in volumes and speeds in the studied five-minute equivalent periods are shown Table 13. Table 13. Ranges in volumes and speeds in the studied five-minute equivalent periods. Location Range of Two-Way Factored Hourly Volume*, vph Range in Average Speeds, mph I-24 5,700 to 8,212 67 mph to 72 mph for Upwind Lapse groups 68 mph to 72 mph for Calm Lapse groups 69 mph to 71 mph for Calm Neutral groups I-90 4,779 to 5,488 66 mph to 71 mph for Downwind Lapse 68 mph to 70 mph for Calm Neutral groups SR-71 3,628 to 3,764 66 mph to 76 mph for Downwind Neutral groups MD-5 400 to 2,936 58 mph to 63 mph for Downwind Lapse groups 58 mph to 64 mph for Downwind Neutral groups 58 mph to 63 mph for Calm Neutral groups 58 mph to 64 mph for Downwind Neutral groups *Total two-way volume averaged across the periods in that group and factored up to one hour. Finding 14: There are slight differences in the sound level differences for different meteorological classes, however there are no clear trends, as evidenced at I-90, I-24 and MD-5. Data collected at greater distances from the road might tell more. As shown earlier in Table 11, data were collected under a fair range of meteorological classes across all of the locations. There are too many differences from one location to another, such that comparison of meteorological class results from one location to another would not be meaningful. However, comparison of difference meteorological class results at the same location has more validity. The following comparisons were made: • Calm Neutral and Downwind Lapse at I-90 • Upwind Lapse, Calm Lapse and Calm Neutral at I-24 • Downwind Lapse, Downwind Neutral, Calm Neutral and Calm Inversion at MD-5 The equivalent-period data at the SR-71 location all fell in the same Downwind Neutral meteorological class, not allowing for comparisons across classes at the most distant microphones in the study. However, the less rigorous comparison of the broadband A-weighted levels shown earlier in Figure 64 under Finding 11 gives evidence of the differences between the BarCom04 and NoBarCom06 levels being 1.5 dB to 2 dB greater during the early Calm Neutral periods than the later Downwind Neutral periods.

79 At the I-90 location, the results show that the Calm Neutral differences were slightly greater than the Downwind Lapse differences at the community microphones. The I-24 data show that the Upwind Lapse average differences tend to be: • Both slightly less and slightly greater than the Calm Lapse average differences in the lower frequency bands, by a few tenths of a decibel; and • A few tenths of a decibel greater than the Calm Lapse average differences in the higher frequency bands (500 Hz to 4 kHz). At the MD-5 location (ignoring the frog noise at 4 kHz at the No Barrier microphones), for the lower community microphones (BarCom03 and NoBarCom05), the Calm Neutral differences are: • 1 dB to 1.5 dB greater than all three of other classes at 125 Hz; • 0.5 dB to 1.0 dB less than all three other classes at 200 Hz; • About 1 dB greater than the two Downwind cases at 250 Hz 1 dB to 1.5 dB less than the two Downwind cases at 400 Hz through 630 Hz; • About a half decibel less than the two Downwind cases at 1kHz through 3.15 kHz. For the MD-5 upper community microphones (BarCom04 and NoBarCom04), the Calm Neutral differences are: • 1 dB to 2.5 dB greater than all three of other classes at 125 Hz; • 0.5 dB to 1.0 dB less than all three other classes at 200 Hz; • About 1 dB less than the Calm Inversion cases at 63 and 100 Hz; • A half decibel or less different at the rest of the frequency bands compared to all three other meteorological classes. Supporting figures are shown in the following sections. I-90 Calm Neutral and Downwind Lapse Comparison Figure 68 compares the I-90 differences in level for the Downwind Lapse and Calm Neutral classes for BarCom03 vs. NoBarCom05 and BarCom04 vs. NoBarCom06. Again, the data values are the average Calm Neutral differences minus the average Downwind Lapse differences for each frequency band. The data show that the Calm Neutral average differences tend to be slightly greater than the Downwind Lapse average differences across the frequency spectrum. For both microphone pairs, the Calm Neutral differences are greater than the Downwind Lapse differences by a decibel or less up though 2.5 kHz. (with the exception at 25 Hz, where the difference is 2 dB). At 3.15 kHz and above, the Calm Neutral differences range from 0.5 dB to 2.0 dB higher.

80 Figure 68. Calm Neutral minus the Downwind Lapse average differences (Leq(5min), BarCom03 minus NoBarCom05 (top) and BarCom04 minus NoBarCom06 (bottom), I-90. I-24 Upwind Lapse, Calm Lapse and Calm Neutral comparisons Figure 69 compares the differences in level for the Upwind Lapse and Calm Lapse classes for the four community microphone positions for the I-24 location. The data values are the average Upwind Lapse differences minus the average Calm Lapse differences for each frequency band. -3 -2 -1 0 1 2 3 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - DLG -3 -2 -1 0 1 2 3 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - DLG

81 Figure 69. Differences in the Upwind Lapse average differences and the Calm Lapse average differences (Leq(5min) +/- one standard deviation, dB), BarCom03 minus NoBarCom05 (top) and BarCom04 minus NoBarCom06 (bottom), I-24. -3 -2 -1 0 1 2 3 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz ULG minus CLG -3 -2 -1 0 1 2 3 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz ULG minus CLG -3 -2 -1 0 1 2 3 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz ULG minus CLG

82 MD-5 Downwind Lapse, Downwind Neutral, Calm Neutral and Calm Inversion Comparison For the MD-5 data, the results are shown by microphone pair. The Downwind cases are in the afternoon measurement session and the Calm cases are in the evening session. In each figure, the top graph is for the differences in the Calm Neutral and Downwind Lapse average difference; the middle graph compares Calm Neutral to Downwind Neutral; and the bottom graph compare Calm Neutral to Calm Inversion. Figure 70 is for BarCom03 minus NoBarCom05 (the lower microphones in the field). Figure 71 is for BarCom04 minus NoBarCom06 (the upper microphones in the field).

83 Figure 70. Calm Neutral average differences minus Downwind Lapse, Downwind Neutral and Calm Inversion average differences (Leq(5min), NoBarCom05 minus NoBarCom05, MD-5). -6 -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - DLG -6 -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - DNG -6 -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - CIG

84 Figure 71. Differences in the Calm Neutral average differences and Downwind Lapse, Downwind Neutral and Calm Inversion average differences (Leq(5min), BarCom04 minus NoBarCom06, MD-5. -6 -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - DLG -6 -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - DNG -6 -4 -2 0 2 4 6 dB A d BZ 2 0 2 5 3 1. 5 4 0 5 0 6 3 8 0 1 00 1 25 1 60 2 00 2 50 3 15 4 00 5 00 6 30 8 00 1 k 1. 25 k 1. 6k 2k 2. 5k 3. 15 k 4k 5k 6. 3k 8k 10 k Di ffe re nc e in Le ve l, dB 1/3 Octave Band Frequency, Hz CNG - CIG

85 Spectrogram Findings Finding 15: Barrier reflections cause sound levels to increase over a broad range of frequencies and cause higher sound levels to be sustained for a longer period of time, as evidenced by the spectrogram results at all locations. Spectrograms (spectral time histories) were generated for individual or groups of vehicle passby events, as well as samples of highway traffic noise for each of the measurement sites. Examples of the spectrograms are shown in this section. The spectrograms compare data collected at the Barrier site and at an equivalent position at the No Barrier site. This comparison allows for a visual examination of the effect of barrier-reflected noise. Examples of the vehicle passby events are shown first, followed by example time blocks of highway traffic noise. What can be seen in the spectrogram data when comparing the Barrier to the No Barrier sites is the following: 1. The hot spots (highest sound levels) get hotter (sound levels increase) when there is a barrier present. 2. The hot spots expand (taller and wider) when there is a barrier present. The presence of a noise barrier causes sound levels to increase over a broad range of frequencies and causes higher sound levels to be sustained for a longer period of time. It should be noted that the above observations apply to vehicles traveling on either side of the road, for a range of distances from the road and heights above the road, and for the vehicle types examined (autos, heavy trucks, and motorcycles). There is evidence that the barrier effect is more pronounced at farther distances from the road. It is assumed that the path length difference between direct and reflected sound is one of the variables controlling the strength of the effect seen from barrier reflections. At farther distances, the path length difference is comparatively smaller, allowing both the direct and reflected sound to contribute to the overall sound level. With larger path length differences, as is the case near the highway, the direct sound would be more dominant than the reflected, and therefore contribute more to the overall sound level, with the reflected sound contributing very little (since it has to travel so much farther than the direct sound). See Chapter 4 on suggested research for more information about this idea. The first vehicle passby spectrogram example is from the SR-71 site in California (Figure 72). For this example, results are being shown for the distant microphone pair, BarCom04 (top plot) and NoBarCom06 (bottom plot), which was located 400 feet from the road. The passby event is a motorcycle traveling southbound, adjacent to the community side, going by the Barrier site at around 12:10:25 and the No Barrier site around 12:10:50. The barrier effect can clearly be seen when comparing the two spectrograms. For the Barrier site, the hot spots are hotter and also wider and taller for a broad range of frequencies. It is particularly noticeable for frequencies from 250 Hz to 2.5 kHz. Because there were differences in the terrain over the long distance, a brief TNM analysis was conducted to determine how the terrain differences would affect the comparison. Based on the TNM analysis conclusions, the differences seen from 500 Hz to 2.5 kHz can be attributed to the barrier reflections. Below 500 Hz the differences may or may not be attributed to the barrier.

86 Figure 72. SR-71 spectrograms for motorcycle on southbound (community) side (approximate event times): Barrier site 12:10:25, No Barrier site 12:10:50. Top is BarCom04; bottom is NoBarCom06.

87 The second vehicle passby spectrogram example is from the MD-5 site in Maryland (Figure 73). For this example, results are being shown for the high microphone pair, BarCom04 (top plot) and NoBarCom06 (bottom plot), which was located 75 feet from the road. The passby event is a pickup truck traveling southbound, adjacent to the barrier side, going by the Barrier site at around 20:09:20 and the No Barrier site around 20:09:35. The barrier effect can clearly be seen when comparing the two spectrograms. The darkest red areas (highest sound levels) fill in more and become wider and taller with the barrier present. The red is centered around 800 or 1000 Hz. The same effect occurs in the surrounding frequency bands, stepping through various colors of the spectrum. The effect is clarified in Figure 74, where the highest levels from the previous figure are extracted and overlaid. The trace from the barrier site (in gray) is taller (broader frequency spread) and wider (more time duration) than the No Barrier site (in red). The intensifying and expanding hot spots indicates that the barrier is causing higher sound levels at frequencies which contribute most to the overall sound level and causing these levels to be sustained for a longer period. The third vehicle passby spectrogram example is from the I-90 site in Illinois (Figure 75). For this example, results are being shown for the microphone pair closest to the road, BarCom03 (top plot) and NoBarCom05 (bottom plot), which was located about 52 feet from the road. The passby event is a heavy truck traveling southbound, adjacent to the community side, going by the Barrier site at around 13:29:36 and the No Barrier site around 13:29:43. The event can be identified by Doppler Effect, with a distinct yellow/orange band (around 62 dBA) along time shifting from 160 Hz to 125 Hz. The barrier effect can clearly be seen when comparing the two spectrograms. For the Barrier site, the hot spots are wider and taller than for the No Barrier site for a broad range of frequencies. It can be seen that the tallest darkest red band (highest sound level band) centered around 1,000 Hz is both wider and taller, with the same effect occurring in the surrounding frequency bands, stepping through various colors of the spectrum. This difference indicates that the barrier is causing higher sound levels at frequencies which contribute most to the overall sound level and causing these levels to be sustained for a longer period for each vehicle passby event.

88 Figure 73. MD-5 spectrograms for a pickup truck on southbound (barrier) side (approximate event times: Barrier site 20:09:20, No Barrier site 20:09:35). Additional vehicle follows the heavy truck. Top is BarCom04; bottom is NoBarCom06.

89 Figure 74. Overlay of MD-5 pickup truck passby hot spots for levels greater than ~60 dBA: BarCom04 (hot spot now represented in gray/black) and NoBarCom06 (hot spot represented in orange/red). The next two examples show longer periods of traffic noise. In the first of these examples, a four- minute block of data starting at 9:49 for the SR-71 site in California is plotted in Figure 76. For this example, results are being shown for the distant microphone pair, BarCom04 (top plot) and NoBarCom06 (bottom plot), which was located 400 feet from the road. The spectrograms show a clear difference between the Barrier and No Barrier sites. As with the passby data, the clean data blocks show that hot spots are both wider and taller for a broad range of frequencies, particularly for 500 Hz and up, the range to which barrier effects can be attributed (based on the TNM analysis for terrain differences at this site). In the FHWA Method analysis of the overall A-weighted equivalent sound level, several clean data blocks were examined, and it was found that the difference between Barrier and No Barrier A-weighted equivalent sound levels ranged from 1.3 to 3.3 dB. The four-minute block at 9:49 shown in the spectrogram is the case where there was a 3.3 dB difference. (Note: The two “blips” in the spectrogram for BarCom04 at about 09:50:10 and 09:52:30 are due to vehicles on a side road passing closely by the microphone.) The next example of traffic noise shows a five-minute block of clean data starting at 15:56 for the I-24 site in Tennessee (Figure 77). For this example, results are being shown for two pairs of microphones, in order from top to bottom in the figure: BarRef01 (33 ft from road, barrier side), NoBarRef02 (33 ft from road, barrier side), BarCom04 (66 ft from road, high microphone), and NoBarCom06 (66 ft from road, high microphone). For the reference positions, the spectrograms show a clear indication that there are more occurrences of higher sound levels (dark red) for the barrier case compared to the No Barrier case. In addition, the higher sound level events are broader in frequency and time. Vehicles traveling eastbound (barrier side of road) dominate the sound levels, and during the five-minute block, single events can be tracked from the Barrier site to the No Barrier site about 15 to 20 seconds later. Across the road from the barrier, the high microphone also indicates that the higher sound level events are broader in frequency and time. Vehicles traveling westbound dominate the sound levels (community side of road), and single events can be tracked from the No Barrier site to the Barrier site about 15 to 20 seconds later. The barrier effect trends are not as obvious across the road from the barrier as for the reference positions, but they can be seen by focusing on a series of events and noticing that multiple consecutive events are more blended 10 seconds

90 together in the Barrier case than the No Barrier case. As the higher levels (hot spots) broaden, they blend together more. Figure 75. I-90 spectrograms for a heavy truck on southbound (community) side (approximate event times: Barrier site 13:29:36, No Barrier site 13:29:43). Top is BarCom03; bottom is NoBarCom05.

91

92 Figure 76. SR-71 spectrograms for 4-minute block of data in the morning at 09:49: top is BarCom04; bottom is NoBarCom06.

93 Figure 77. I-24 five-minute spectrograms; top to bottom: BarRef01, NoBarRef02, high mics (BarCom04 and NoBarCom06); for Calm Lapse group CLG-6-1, start time 15:56.

94 Psychoacoustics Findings Finding 16: Combined psychoacoustic metrics of Unbiased Annoyance and Psychoacoustic Annoyance yield similar results to one another, while Category Scale of Annoyance does not yield useful indications. As discussed above, a set of combined psychoacoustic metrics indicating annoyance were applied to the audio recordings for each microphone. Three such annoyance metrics were tested. Of these, the Unbiased Annoyance (UBA) and Psychoacoustic Annoyance (PA) consistently showed similar results. This is to be expected, as they derive from similar approaches in psychoacoustics research. Both of these metrics are dominated by Loudness (measuring total energy and accounting for masking) and Sharpness (an indicator of high-frequency spectral content). The Category Scale of Annoyance (CSA) was ineffective at indicating any differences at all sites and microphone locations. This is most likely due to its simplicity (a simple linear combination of psychoacoustic metrics) and the fact that it was derived from listening studies based on simplified product noise. This result is demonstrated in Figure 78, where the time series and histograms of UBA, PA, and CSA are shown for the upper community microphones. Finding 17: Annoyance metrics show differences between Barrier and No Barrier sites at moderate distances, but the results are contra-indicative. The psychoacoustic metrics applied to the audio recordings did not show positive correlation with higher annoyance at the Barrier sites. When simple descriptive statistics are applied to the resulting UBA and PA time series, there are cases where the mean values from the Barrier sites differed to an appreciable level of significance from those at the No Barrier sites. In those cases where the recording microphones were located close to the roadway, the statistics showed no difference between sites. The Loudness and Sharpness at these locations are dominated by direct sound from passing vehicles, and the annoyance metrics, which rely primarily on Loudness, are similarly dominated by direct sound. The cases where the mean values of annoyance differed to a statistically significant extent tended to occur for the higher microphones, and it was more pronounced for those microphones located at moderate distances from the roadway. This is demonstrated at CA SR-71, as shown in Figure 79. The microphones BarCom03 and NoBarCom05 are located 5 feet above and very close to the roadway, while BarCom04 and NoBarCom06 are 15 feet above and quite far from the roadway. Unfortunately, to the extent that the mean values of annoyance showed significant differences, they were contra-indicative: the annoyance metrics at Barrier sites tended to have lower values than those at the No Barrier sites. An explanation for why this tended to occur was not developed in this work. Finding 18: Annoyance metrics are less effective in heavy, constant traffic, but show differences in lighter traffic with separated passbys. The annoyance metrics computed from the recordings at MD-5 are of particular note. The recordings made in the afternoon, with continuous heavy traffic, do not show clear differences between the sites. However, the recordings made at night, when the sound signals consisted mostly of individual vehicle passby events, showed more and more differentiation as the traffic became lighter. This is demonstrated in Figure 80. This may indicate that the psychoacoustic metrics, as applied here, are more applicable to complexities of individual vehicle events than to the general sonic mash of heavy traffic. Note too that the annoyance metrics’ dependence on Loudness is revealed in the gradual decrease in traffic toward midnight.

95 Figure 78. Comparing UBA, PA, and CSA (top to bottom) at the upper community microphones, CA SR-71.

96 Figure 79. Comparing UBA computed for lower community microphones, close to the roadway, with upper community microphones, distant from the roadway, at CA SR-71.

97 Figure 80. Comparing PA at MD-5 for heavy traffic (above) and light, decreasing traffic (below). Applications FHWA Method The most immediate application of these results is the understanding by traffic noise analysis and abatement practitioners, that traffic sound levels and sound characteristics for receptors across from a proposed single reflective noise barrier can change after the installation of the barrier. The sound level increases can be evident in elevated broadband A-weighted sound levels and unweighted sound pressure levels that are caused by frequency-dependent changes. Also, the traffic noise background levels can be elevated, sustaining the sound of individual vehicle passage, as discussed below for spectrograms. This understanding can lead to the appropriate specification of sound-absorbing surfaces on these single barriers, especially for highway widenings where the roadway already exists—such as in a Type I widening project or a Type II retrofit noise abatement project (both as defined in current state highway agencies’ (SHA) noise policies). In many cases, the receptors across from a single barrier may have been determined to be impacted by a proposed project in the project’s noise study. However, they did not qualify for abatement based on acoustical or physical feasibility reasons or because they did not meet a SHA’s noise abatement reasonableness criteria in its traffic noise policy. The specification of sound-absorbing single barriers can be a proactive step to not make the situation worse for such impacted receptors.

98 In other cases, the receptors may be more distant and not be impacted; nonetheless, the results of this study suggest that such receptors can experience changes in the noise environment due to the introduction of the barrier on the opposite side of the road. While it might be difficult to justify making the barrier sound-absorbing in such a case, the practitioner at least has a better idea of the nature of the phenomenon. Spectrograms Spectrograms provide a color-coded visualization of frequency and temporal characteristics of highway traffic noise. Such visualizations can help to reveal differences when comparing a site with and without a noise barrier; this may be useful when trying to explain the effect of barrier reflections—there is a clear visual difference when a barrier is present. Spectrograms may also help to visualize barrier reflections with and without absorption (see recommendations); a comparison could be made among a site without an opposing barrier, one with a reflective barrier, and one with an absorptive barrier. This could help policy makers provide guidance on when it may be effective to use sound-absorbing barriers and help in showing the public what an absorptive barrier could provide. Psychoacoustics Because the derived annoyance metrics were either uncorrelated with site location or were contra- indicative for continuous flow traffic, their direct use (as applied in this work) cannot be recommended for indicating increased annoyance due to single barriers in the presence of heavy traffic. However, these metrics are useful in cases where the overall roadway sound is dominated by individual vehicle signatures. This is because time-varying annoyance metrics are dominated by the effects of short-term loud events (N5).