Assessing Community Annoyance of Helicopter Noise (2017)

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.

58 For reasons previously described, helicopter noise exposure levels were estimated by both measurement and modeling at the Long Beach and Las Vegas sites, and by noise modeling alone in Washington, D.C. 5.1 Comparison of Measurement and Modeling Estimates of Exposure Levels at Long Beach and Las Vegas Survey Sites The Long Beach and Las Vegas survey areas were fully developed residential areas, with sub- stantial background noise. DNLs associated with helicopter operations were therefore estimated for measurement sites within each of these two survey areas by cumulating measured sound exposure level (SEL) values for each helicopter flyover during the week prior to interviewing. An analysis was then conducted on a flyover-by-flyover basis to determine whether noise levels recorded during flyovers represented helicopter-produced noise exposure or noise exposure produced by other noise sources. 5.1.1 Measured DNLs The times of the closest point of approach (CPA) of helicopter flights to each monitoring site were entered into a database. The database also included all 1 second Leq data (A-weighted, C-weighted, and 1⁄3 octave band) for a period of 1 minute prior to and 1 minute after the CPA time. Signal-to-noise ratios of flyovers were adequate to distinguish helicopter noise emissions from ambient noise near CPA times, but were difficult to unambiguously distinguish from background noise at greater distances and times before and after CPA. SEL values as a function of distance for both A- and C-weighted SEL values were accord- ingly examined more closely. The examination showed that any noise event associated with a helicopter flight track that passed within a 3,000-foot radius of a monitoring point had a maximum A-weighted noise level (Lmax) ≥ 55 dB, lasted at least 3 seconds, and could be attrib- uted to a helicopter overflight. Measured noise levels that met these criteria were accumulated to compute daily, helicopter-only DNL values for each site. (C-weighted Lmax values were not used for this purpose, because the background noise included substantial C-weighted noise.) 5.1.2 Modeled DNLs Operational information and radar data recorded during the survey were then used to model DNL at each measurement site. This was used to compare modeled to measured DNL values. Several iterations of the model were completed so that at each site the modeled noise matched the C H A P T E R 5 Analyses of Noise Exposure Measurements and Interview Findings

Analyses of Noise Exposure Measurements and Interview Findings 59 measured DNL values. Locations of dispersed flight tracks and numbers of operations assigned to dispersed tracks in the modeling software were modeled to measured estimates of DNL values. Table 5-1 shows the numbers of helicopter operations by aircraft type and time of day. Table 5-2 compares measured with modeled noise levels at the Long Beach and Las Vegas survey sites. Differences between measured and modeled DNL values were less than 2 dB, except at Site 4 in Long Beach.17 Differences of this magnitude are well within (1) the overlapping uncertainty of measurement, (2) uncertainty in noise modeling, (3) the uncertainty inherent in the measurement system for SEL (approximately + 0.8 dB, per ISO 20906, Annex B), and (4) the sampling uncertainty for a short-term measurement period. 5.1.3 Relation of A-Weighted to C-Weighted SELs A- and C-weighted SEL differences were computed for each flight in each study area using measurement data. Figures 5-1 and 5-2 plot A- and C-weighted SELs against each other at the two study sites. The two noise metrics are highly correlated at each site, despite the scatter about the regression line of about ±5 dB. The difference between A- and C-weighted SEL is greater at Las Vegas (approximately 10 dB difference) than at Long Beach (approximately 5 dB difference). This is almost certainly due to the absence of smaller Robinson rotorcraft from the Las Vegas fleet. The 1 second Leq thresholds at LGB and LAS were 55 and 50 dB, respectively. The difference in threshold was due to ambient noise levels. The result was that most events correlated had Type of Helicopter Average Daily Overflights Day Night Total Long Beach B206L 0.4 0.1 R22 1.8 0.4 R44 2.4 0.5 S76 1.7 0.3 SA350D 9.2 1.8 15.5 3.1 18.6 Las Vegas EC130 103.4 11.4 SA350D 31.6 3.4 135.0 14.8 149.8 Washington, D.C. A109 3.4 0.0 B212 5.0 0.0 S61 1.6 0.0 S70 3.6 0.0 SA365N 4.6 0.0 18.2 0.0 18.2 Table 5-1. Helicopter operations data. DNL Study/Estimation Method Site 1 Site 2 Site 3 Site 4 LGB Measured 47.1 48.8 47.9 44.7 LGB Modeled 46.0 47.5 49.7 49.3 Difference* -1.1 -1.3 1.8 4.6 LAS Measured 52.0 50.6 48.7 52.9 LAS Modeled 53.0 49.1 46.8 52.9 Difference* 1.0 -1.5 -1.9 0.0 *Positive numbers indicate that the modeled DNL was greater than the measured DNL. Table 5-2. Comparison of measured with modeled DNL values.

60 Assessing Community Annoyance of Helicopter Noise y = 1.0598x + 3.2351 R² = 0.9209 50 60 70 80 90 100 110 50 60 70 80 90 100 C- W ei gh te d SE L (d B) A-Weighted SEL (dB) LGB Figure 5-1. Comparison of measured A-weighted and C-weighted SELs of helicopter overflights in Long Beach interviewing area. y = 1.0814x + 6.4062 R² = 0.9376 50 60 70 80 90 100 110 50 60 70 80 90 100 C- W ei gh te d SE L (d B) A-Weighted SEL (dB) LAS Figure 5-2. Comparison of measured A-weighted and C-weighted SELs of helicopter overflights in Las Vegas interviewing area. SELs above 60 and 55 dB, respectively. The handful of events with SELs below these levels were events that exceeded the threshold, but had very short durations. Section 5.6 analyzes C-weighted and low-frequency exposure estimates in greater detail at the three survey sites. 5.2 Disposition of Contact Attempts A total of 10,562 contact attempts (7,684 to land line telephones and 2,878 to wireless telephones) were made in Long Beach, and 7,803 contact attempts (4,668 to land line telephones and 3,135 to wireless telephones) in Las Vegas. For Washington D.C. 4,224 (2,873 landline and 1,351 wireless) contact attempts were made. Table 5-3 summarizes the outcomes of these

Analyses of Noise Exposure Measurements and Interview Findings 61 interview contact attempts. The “non-sample” category includes disconnects, businesses and other non-residential telephone numbers, fax machines, modem lines, wrong addresses, changed numbers, and non-English speaking households. “Noncontacts” includes busy signals, no answer, call blocked, and answering machines after fifteen attempts to contact. The completion rates are calculated as {completed interviews/[total - (non-sample + noncontact)]}, while the refusal rates are calculated as {refused interviews/[total - (non-sample + noncontact)]}. 5.3 Locations of Respondents’ Residences The locations of households that completed interviews are shown in Figures 5-3, 5-4, 5-5, and 5-6 as green dots, enlarged sufficiently to preserve confidentiality of individual respondents. These figures also show the approximate locations of households in which respondents were highly annoyed by helicopter noise (and in the case of Washington, D.C., interviews, by fixed- wing aircraft noise.) Households completing interviews were generally well dispersed geographically throughout the study areas, as were highly annoyed respondents. In Long Beach, some clustering of highly annoyed respondents was observed along the Redondo corridor, and along the northern section of the coastal route. Much less clustering was observed in Las Vegas along Tropicana Avenue, and in the Washington, D.C., area. 5.4 Analysis of Interview Responses 5.4.1 Tabulation of Responses Table 5-4 displays responses to individual questionnaire items for the three interviewing sites, both separately and combined. (Percentage values may sum to less than 100 because invalid responses were omitted.) The reported results do not differentiate between respondents contacted by home landline and wireless telephones. Table 5-5 shows similar information for mean estimated helicopter noise exposure levels and distances from flight corridors. 5.4.1.1 Narrative Account of Responses to Questionnaire Items This sub-section summarizes responses to individual questionnaire items across sites in general terms. More detailed accounts of the findings are presented in the following subsections. Sample Disposition Long Beach Las Vegas Washington, D.C. Landline Wireless Landline Wireless Landline Wireless Total Sample Released for Dialing 7,684 2,878 4,688 3,135 2,873 1,351 Non-Sample 3,511 419 2,713 1,028 1,137 553 Noncontact 1,913 1,022 0 0 1,244 390 Non-Sample + Noncontact 5,424 1,441 2,713 1,028 2,381 943 Contacted Sample 2,260 1,437 1,975 2,107 492 408 Refused Interviews 1,466 432 1,348 1,973 152 306 Completed Interviews 794 295 607 134 340 102 Interview Completion Rate 35.1% 20.5% 30.7% 6.4% 69.1% 20.7% Interview Refusal Rate 64.9% 30.1% 68.2% 93.8% 30.9% 75.0% Table 5-3. Interview completion and refusal rates by site and type (landline/wireless) of telephone service.

62 Assessing Community Annoyance of Helicopter Noise Figure 5-3. Approximate locations of Long Beach respondents (in green), and those highly annoyed by helicopter noise (in red). Figure 5-4. Approximate locations of Las Vegas respondents (in green), and those highly annoyed by helicopter noise (in red). LAS Helo Survey Results

Survey Responses Green-Survey Respondents Red-Highly Annoyed (Helicopter) 1 mi N © 2016 Google l Figure 5-5. Approximate locations of Washington respondents (in green), and those highly annoyed by helicopters (in red). Survey Responses Green-Survey Respondents Red-Highly Annoyed (Fixed wing) 1 mi N © 2016 Goolgle Figure 5-6. Approximate locations of Washington respondents (in green), and those highly annoyed by fixed-wing aircraft noise (in red).

64 Assessing Community Annoyance of Helicopter Noise Table 5-4. Questionnaire response percentages and frequencies. several times or more per hour 1.4% (15) 6.1% (45) 31.4% (139) 8.8% (199) 9A Annoyed by aircraft other than helicopters not at all (from Item 9) 86.5% (942) 90.6% (671) 50.5% (223) 80.8% (1,836) Slightly 2.2% (24) 2.3% (17) 5.2% (23) 2.8% (64) Moderately 2.8% (31) 1.6% (12) 14.7% (65) 4.8% (108) Very 1.1% (12) 0.8% (6) 7.9% (35) 2.3% (53) Extremely 1.8% (20) 1.2% (9) 15.2% (67) 4.2% (96) 10A Degree of annoyance with helicopter thumping or slapping sounds not at all (from Item 10) 79.5% (866) 87.2% (646) 75.8% (335) 81.3% (1,847) Slightly 5.5% (60) 4.2% (31) 4.5% (20) 4.9% (111) Moderately 3.6% (39) 3.7% (20) 4.8% (21) 3.5% (80) Very 2.0% (22) 0.9% (7) 3.2% (14) 1.9% (43) Extremely 2.0% (22) 1.3% (10) 3.2% (14) 2.0% (46) 11A Annoyed by helicopter buzzing not at all (from Item 11) 77.6% (845) 87.0% (645) 79.6% (352) 76.9% (1,747) Slightly 6.2% (67) 23.2% (24) 2.5% (11) 4.2% (95) Moderately 4.8% (52) 3.0% (22) 4.1% (18) 4.5% (102) Very 1.5% (16) 0.7% (5) 1.8% (8) 4.0% (92) Extremely 2.1% (23) 1.3% (10) 2.5% (11) 1.9% (44) 12A Annoyed by helicopter whining or tonal not at all (from Item 12) 83.6% (910) 90.7% (672) 80.1% (354) 85.2% (1,936) Slightly 2.8% (31) 1.6% (12) 3.2% (14) 2.5% (57) Moderately 2.0% (22) 1.8% (13) 2.9% (13) 2.1% (48) Very 1.7% (19) 0.7% (5) 1.6% (7) 1.4% (31) Extremely 1.7% (18) 0.9% (7) 2.0% (9) 1.5% (34) 13A Annoyed by helicopter vibrations or rattling not at all (from Item 13) 76.4% (832) 87.4% (648) 74.9% (331) 79.9%(1,811) Slightly 5.5% (60) 4.0% (30) 6.1% (27) 5.1% (117) Moderately 4.6% (50) 1.6% (12) 4.3% (19) 3.6% (81) Very 2.9% (32) 0.9% (7) 2.7% (12) 2.2% (51) Extremely 3.4% (37) 1.6% (12) 3.4% (15) 2.8% (64) 14 Frequency of notice of vibration or rattling noises once a week or less 60.7% (661) 48.6% (360) 55.2% (244) 55.7% (1,265) once a day 7.8% (85) 4.7% (35) 5.9% (26) 6.4% (146) several times a day 5.6% (61) 4.0% (30) 6.3% (28) 5.2% (119) 15A Frequency of complaint never (from Item 15) 96.2% (1,048) 98.1% (727) 92.5% (409) 96.1% (2,184) Once 0.5% (5) 0.1% (1) 1.1% (5) 0.5% (11) a few times 0.7% (8) .05% (4) 1.6% (7) 0.8% (19) many times 0.4% (4) 0.8% (6) 1.1% (5) 0.7% (15) QUESTIONNAIRE ITEM CODING LONG BEACH % (count) N = 1,089 LAS VEGAS % (count) N = 741 WASHINGTON % (count) N = 442 COMBINED SITES % (count) N = 2,272 1 Duration of residence less than one year 2.6% (28) 2.7% (20) 2.7% (12) 2.6% (1,573) at least 1 year but less than 2 years 5.8% (63) 3.2% (24) 2.0% (9) 4.2% (98) 2 to 5 years 23.0% (250) 19.8% (147) 18.6% (82) 21.2% (475) 5 to 10 years 52.5% (572) 59.2% (439) 67.9% (300) 57.7% (1,311) more than 10 years 16.2% (176) 15.0% (111) 8.8% (39) 14.3% (326) 4 Characterization of neighborhood as quiet or noisy Quiet 68.1% (742) 84.1% (623) 47.1% (208) 69.2% (1,573) quiet except for aircraft 4.2% (32) 6.1% (45) 31.9% (141) 10.2% (232) Noisy 24.4% (266) 8.6% (63) 3.4% (15) 17.9% (407) 4A Judged noisiness of neighborhood quiet (from Item 4) 68.1% (742) 84.1% (623) 47.1% (208) 69.2% (1,573) Slightly noisy 2.8% (31) 1.8% (13) 1.1% (5) 2.2% (49) Moderately noisy 13.5% (147) 3.9% (29) 5.6% (38) 9.4% (214) Very noisy 4.2% (46) 1.6% (12) 5.0% (22) 3.5% (80) Extremely noisy 3.5% (38) 1.2% (9) 2.7% (12) 2.6% (59) 5A Annoyance of street traffic noise not at all (from Item 5) 71.7% (781) 85.7% (635) 83.0% (367) 78.5% (1,783) Slightly 8.3% (90) 5.8% (43) 4.8% (21) 6.8% (154) Moderately 10.9% (119) 4.3% (32) 7.5% (33) 8.1% (184) Very 3.9% (43) 1.8% (13) 0.9% (4) 2.6% (60) Extremely 4.1% (45) 1.2% (9) 2.7% (12) 2.9% (66) 6A Frequency of notice of helicopter noise not noticed (from Item 6) or less than once a day 36.9% (402) 51.4% (381) 47.5% (210) 43.7% (993) about once a day 17.0% (294) 19.3% (143) 18.6% (82) 22.8% (519) a few times a day 19.1% (208) 16.5% (122) 16.5% (73) 17.7% (403) several times or more per hour 4.5 (49) 6.3% (46) 5.0% (22) 5.1% (117) 7A Judged annoyance of helicopter noise not at all (from items 6 and 7) 67.6% (736) 85.4% (633) 71.8% (317) 74.2% (1,686) Slightly 6.2% (67) 2.3% (17) 4.5% (20) 4.6% (104) Moderately 7.5% (82) 3.5% (26) 7.7% (34) 6.3% (142) Very 3.9 (42) 1.5% (11) 2.5% (11) 2.8% (64) Extremely 5.2 (57) 1.9% (14) 4.5% (20) 4.0% (91) 8A Frequency of notice of other aircraft noise not noticed (from Item 8) or less than once a day 69.3% (755) 66.5% (493) 24.0% (106) 59.6% (1,354) once a day 14.2% (154) 12.8% (95) 12.2% (54) 13.3% (303) a few times a day 6.3% (69) 9.0% (67) 23.3% (103) 10.5% (239)

Analyses of Noise Exposure Measurements and Interview Findings 65 Duration of Residence (Item 1). All of the neighborhoods in which interviewing was con- ducted were characterized by stable residential populations. Fewer than 3% of the respondents at any of the interviewing sites had lived at their current addresses for less than 6 months prior to the conduct of the present study, while half or more of the respondents had lived at their current addresses for 5 to 10 years. The populations of the interviewing sites were thus thoroughly familiar with helicopter noise exposure. Characterization of Neighborhood as Quiet or Noisy (Item 4). Large majorities of respon- dents in Long Beach and Las Vegas described their neighborhoods as quiet. Nearly half of the respondents in Washington did as well. Nonetheless, nearly a quarter of the respondents in Long Beach described their neighborhood as noisy, and nearly a third of the respondents in Washington described their neighborhood as “quiet, except for aircraft noise.” Only small percentages of respondents at all sites described their neighborhoods as “highly” (“very” or “extremely”) noisy: 7.7% in Long Beach and Washington, and 2.8% in Las Vegas. These figures closely resembled the percentages of respondents highly annoyed by traffic noise in Long Beach and Las Vegas (8.0% in Long Beach and 3% in Las Vegas), but were only about half (3.6%) of the percentage describing their neighborhoods as very or extremely (“highly”) noisy in Washington. Frequency of Notice of Helicopters (Item 6). Figure 5-7 shows how often respondents reported noticing helicopters in Long Beach, Las Vegas, and Washington, respectively. Only small minorities reported noticing helicopters more than a few times a day, and responses in the three survey areas were quite similar. This finding was unexpected because respondents at LAS were exposed to ten times as many helicopter operations as LGB. Association between Helicopter Noise Annoyance and Interviewing Method. A 2 × 2 Chi-square analysis revealed no significant difference in reports of high annoyance by helicopter noise and the respondent’s form of telephone subscription (wireless or landline) in the combined data from the three interviewing sites, p = .561. Likewise, no statistically significant differences in the prevalence of high annoyance were observed at any of the three data collection sites individually, p > .17. Annoyance with Specific Characteristics of Helicopter Noise (Items 10–12). Blade Slap Roughly 80% of all respondents indicated in questionnaire Item 10 that they were not annoyed in any degree by main rotor impulsive noise (“thumping or slapping”). Only about 4% of respon- dents across sites described themselves as highly annoyed by such sounds. Tail Rotor/Sideline Noise A similar percentage of respondents indicated in questionnaire Item 9 that they were not at all annoyed by “buzzing” noises (of the sort often created by tail Table 5-5. Means and standard deviations of respondents’ helicopter noise exposure levels and distances from flight corridors. MEASURE Long Beach Mean (SD) N=1,089 Las Vegas Mean (SD) N=741 Washington Mean (SD) N=442 Combined Sites Mean (SD) N=2,272 Mean DNL Due to Helicopters (standard deviation of DNL) 40.3 (6.4) 43.8 (5.5) 43.3 (4.8) 42.0 (6.1) Mean Distance from Flight Corridor, in Decimal Nautical Miles (standard deviation of distance from center of corridor) 0.42 (0.3) 0.49 (0.3) 0.42 (0.2) 0.44 (0.3) SD = standard deviation.

66 Assessing Community Annoyance of Helicopter Noise rotors or interactions of the tail rotor with the main rotor wake). Only about 6% of respondents across sites described themselves as highly annoyed by such sounds. Whining or Tonal Noise Slightly higher percentages of respondents (85%) at all sites indicated that they were not at all annoyed by whining or tonal noise (presumably jet engine inlet noise). Only about 3% of respondents across sites described themselves as highly annoyed by whining or tonal sounds. Annoyance Due to Helicopter-Induced Vibration and Rattling (Items 13–14). About 80% of all respondents were not annoyed in any degree by helicopter-induced vibration and rattling sounds in their homes. Five percent of all respondents described themselves as highly annoyed by vibration or rattling. A one-way analysis of variance conducted on responses made by Long Beach and Las Vegas respondents revealed a statistically significant difference in distance to flight track between respondents who were and were not annoyed to any degree by in-home vibration and rattling, F(1, 1718) = 6.17, p = .013. The absolute difference was quite small, however, h2 = .004, with a 95% confidence interval (CI) extending from <.001 to .011. Those who reported no annoyance lived farther from the flight track (M = 0.45 nm, SD = 0.27) than those who lived closer to the flight track (M = 0.41 nm, SD = 0.27). Frequency of Complaint (Item 15). Only about 4% of all respondents overall reported that they had complained about helicopter noise. Only in Washington did more than 1% of respondents report having complained more than once. 5.4.1.2 Evidence Relevant to Hypotheses Identified During Planning for the Current Study Seven hypotheses were identified in Chapter 2 of this report. Evidence concerning these hypotheses is discussed below. Hypothesis 1. Decibel for decibel, rotary-wing aircraft is more annoying than fixed-wing aircraft. Washington was the only interviewing site at which respondents were exposed to Figure 5-7. Frequency of notice of helicopters at interviewing sites.

Analyses of Noise Exposure Measurements and Interview Findings 67 appreciable amounts of cumulative noise due to both helicopter and fixed-wing overflights. Figure 5-8 plots (a) percentages of respondents highly annoyed by helicopter and fixed-wing aircraft noise, and (b) percentages of respondents annoyed to any degree in Washington. Note that cumulative exposure to aircraft noise was greater for fixed-wing aircraft than heli- copters in Washington, D.C., and that the expected relationship between noise and annoyance is more evident for fixed-wing aircraft. Note also that only 4 of the 442 respondents reporting high annoyance were exposed to fixed-wing aircraft noise levels in the 45–50 dB range, calling into question the reliability of the 0% high annoyance data point. In only one range of cumulative noise levels (~50–55 dB) did substantial numbers of respon- dents report high annoyance to both fixed-wing aircraft and helicopters. The rates of 21% high annoyance for fixed-wing aircraft and 7% for helicopters were substantially different. The rates for annoyance to any degree appear to be quite similar for helicopters and fixed-wing aircraft in the 45–50 dB range, but higher for fixed-wing aircraft other than helicopters in the 50–55 dB range at 43% and 18%, respectively. The question of whether fixed-wing or helicopter noise was the more annoying at the Washington, D.C., interviewing site was addressed by comparing aircraft noise source responses to helicopter responses (note the higher noise level of fixed-wing aircraft noise in Figure 5-8). Of the 398 cases available for analysis, 44 cases had missing values on one or both of the annoyance measures. The two measures of annoyance were logarithmically transformed prior to inferential analysis due to strong positive skewness. The fixed-wing aircraft generated greater annoyance, as described in detail in the next paragraph. 0 5 10 15 20 25 25-29.9 30-34.9 35-39.9 40-44.9 45-49.9 50-54.9 55-59.9 P e rc e n t H ig h ly A n n o y e d DAY-NIGHT AVERAGE SOUND LEVEL (dB) Helicopter Fixed Wing 25-29.9 30-34.9 35-39.9 40-44.9 45-49.9 50-54.9 55-59.9 0 5 10 15 20 25 30 35 40 45 50 P e rc e n t A n n o y e d t o A n y D e g re e DAY-NIGHT AVERAGE SOUND LEVEL (dB) Helicopter Fixed Wing (a) (b) Figure 5-8. Proportion of respondents (a) highly annoyed and (b) annoyed to any degree by helicopter and fixed-wing aircraft noise at Washington study site.

68 Assessing Community Annoyance of Helicopter Noise Repeated-measures analysis of variance with varying covariates revealed significantly greater annoyance due to fixed-wing aircraft noise (after adjusting for fixed-wing DNL) than helicopter noise (after adjusting for helicopter DNL), F(1, 396) = 23.70, p < .001, partial h2 = .06 with 95% confidence limits from .02 to .11. On an original scale in which 0 = not at all annoying or not noticing noise source to 4 = extremely annoying, mean annoyance for helicopter noise (log) was 0.108 (SD = 0.217) and the mean annoyance for fixed-wing aircraft noise (log) was 0.255 (SD = 0.292). The greater annoyance reported for exposure to fixed-wing aircraft, although statistically significant, was small, at less than one standard deviation difference in annoyance between the two noise sources. This evidence is both meager and inconclusive. It could well be more the product of recent changes in the fixed-wing flight patterns than differences in perceived annoyance relative to helicopter noise. (Changes in noise exposure associated with the changes in flight tracks were fully accounted for in the noise modeling done for this analysis.) 5.4.1.3 Annoyance by Helicopter Versus Fixed-Wing Aircraft Noise at Long Beach and Las Vegas A similar analysis was conducted at the Long Beach and Las Vegas interviewing sites by once again adjusting for the frequency of noticing helicopter noise and fixed-wing aircraft noise as covariates to determine which aircraft type was more annoying. Of the 1,507 cases available for analysis, 323 cases were missing values on one or more of the four measures. Repeated-measures analysis of variance with varying covariates revealed significantly greater annoyance due to helicopter noise (after adjusting for frequency of noting helicopter noise) than fixed-wing aircraft noise (after adjusting for frequency of noticing fixed-wing or helicopter noise), F(1, 1505) = 31.04, p < .001, partial h2 = .04 with 95% confidence limits from .02 to .06. On an original scale in which 0 = not at all annoying or not noticing noise source to 4 = extremely annoying, the mean annoyance for helicopter noise (log) was 0.087 (SD = 0.20) and mean annoyance for fixed- wing aircraft noise (log) was 0.021 (SD = 0.11). These findings assume no difference in actual loudness of the two types of aircraft noise in the two locations beyond differences in frequency of noticing them. 5.4.1.4 Dosage-Response for High Annoyance The three panels of Figure 5-9 show proportions of respondents highly annoyed by helicopter noise within seven categories of DNL at all three interviewing sites. Binary logistic regression analysis showed a statistically significant relationship between high annoyance (very or extremely annoyed by helicopter noise) and the sound level to which respondents were exposed in Long Beach, but not in the Las Vegas or Washington, D.C., data collection sites. Among the 1,089 Long Beach respondents, 1,050 were at home during the week before data collection and 99 of them were highly annoyed by helicopter noise (Table 5-4 and Table 5-6). A small (Nagelkerke R2 = .019) but significant dosage-response relationship was observed, c2(1, N = 1,050) = 9.28, p = .002. The odds ratio (Be) was 1.107, with 95% confidence limits from 1.061 to 1.327. The dosage-response relationship was not statistically significant at Long Beach, p = .538 or in Washington, p = .143. 5.4.1.5 Annoyance to any Degree due to Helicopter Noise A 2 × 3 (annoyance to any degree by data collection site) analysis of variance predicting helicopter DNL revealed statistically significant main effects of annoyance and site, but not their interaction (Figure 5-10). Helicopter noise exposure was greater for those reporting being at least slightly annoyed (M = 43.47, SE = 0.339) than those who were at home but reported no annoyance (M = 42.26, SE = 42.26), F(1, 2191) = 10.83, p = .001. However, the relationship accounted for little variance in noise exposure, partial h2 = .005 with 95% confidence limits from .001 to .012. Data collection site also predicted differences in noise exposure, F(2, 2191) = 50.97, p < .001, partial h2 = .044 with 95% confidence limits from .012 to .063. Noise exposure differences are presented in Table 5-5.

Data Collection Site: Long Beach (a) Data Collection Site: Las Vegas (b) Figure 5-9. Proportions (with 95% CIs) of respondents highly annoyed by helicopter noise within (a) Long Beach and (b) Las Vegas. (continued on next page)

70 Assessing Community Annoyance of Helicopter Noise Data Collection Site: Washington, D.C (c) Figure 5-9. (Continued) Proportions (with 95% CIs) of respondents highly annoyed by helicopter noise within (c) D.C. data collection sites. Noise Type Site HA/Na B Standard Error Wald df p Nagelkerke R2 Odds Ratio (Be) 95% CI for Odds Ratio Lower Upper In-home vibration/ rattling LGB 69/1050 0.059 0.066 0.811 1 .368 .002 1.061 0.933 1.206 LAS 19/728 0.081 0.145 0.315 1 .575 .002 1.085 0.817 1.440 DCA 27/419 0.194 0.132 2.148 1 .143 .013 1.214 0.937 1.573 Thumping and Slapping LGB 44/1050 0.145 0.083 3.08 1 .079 .010 1.156 0.983 1.359 LAS 17/728 0.138 0.356 0.79 1 .374 .006 1.148 0.846 1.558 DCA 28/419 0.188 0.138 1.84 1 .174 .012 1.207 0.920 1.583 Buzzing LGB 39/1050 0.360 0.092 8.03 1 .005 .030 1.297 1.084 1.553 LAS 18/728 -0.051 0.157 0.10 1 .748 .001 0.951 0.698 1.295 DCA 19/419 0.219 0.168 1.69 1 .192 .014 1.244 0.896 1.728 Whining LGB 37/1050 0.069 0.088 0.61 1 .436 .002 1.071 0.901 1.274 LAS 12/728 0.199 0.190 1.10 1 .294 .010 1.221 0.841 1.771 DCA 16/419 -0.037 0.176 0.045 1 .832 <.001 0.963 0.683 1.360 aHA = Number of respondents highly annoyed; N = Number of valid responses; B = the customary symbol for slope; "Wald" = the value of a Wald test for the significance of the slope; "df" = the usual abbreviation for degrees of freedom; p = the customary symbol for significance; "Nagelkerke R2" is an adjusted coefficient of determination; the odds ratio is a measure of an association of exposure and an outcome; CI = confidence interval. Table 5-6. Summary of logistic regression analyses of proportion highly annoyed by various helicopter noises for three data collection sites.

Analyses of Noise Exposure Measurements and Interview Findings 71 Hypothesis 2. The prevalence of annoyance due to rotary-wing noise is most appropriately predicted in units of A-weighted cumulative exposure. At only one of the three interviewing sites was there a good correlation between annoyance and the A-weighted decibel. Neither the C-weighted nor the helicopter-adjusted LFSL exhibited any greater correlation with annoyance. At the Las Vegas and Washington, D.C., interviewing sites, annoyance was unrelated to dose, as measured by the A-weighted, C-weighted, or the helicopter-adjusted LFSL. At the Washington, D.C., interviewing site, a controversy over relocated fixed-wing tracks may have obscured any dependence of annoyance on dose.18 The low doses of helicopter noise for the three studies cannot be ignored, however. It would have been advantageous to have surveyed a community with higher helicopter noise dose (greater than 60 DNL). To do that, a survey would have had to occur around a military facility. The research panel restricted the surveys to civil helicopter routes thus limiting the noise dose to DNL below 60 dB. See Section 5.6 for details on the low-frequency analysis. Hypothesis 3. Main rotor impulsive noise controls the annoyance of helicopter noise (and hence requires an impulsive noise “correction” to A-weighted measurements). Noise measurements included A- and C-weighted impulsive noise levels. The difference between these and non-impulsive A- and C-weighted levels differed only by constants. However, the civil helicopters measured in this study do not produce the main rotor impulsive noise levels that military helicopters can produce in certain flight regimes. That is not to say there were none, but that the levels were not as pronounced as with heavier helicopters. This hypothesis would be better tested where there were heavy military helicopter operations so that the impulsive noises were more pronounced. Therefore, no clear conclusion could be drawn from these surveys. Figure 5-10. Prediction of helicopter DNL by reported annoyance due to helicopter noise and data collection site.

72 Assessing Community Annoyance of Helicopter Noise (a) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y V ib ra ti o n o r R a tt li n g Helicopter DNL Data Collection Site: Long Beach Proportion Lower limit Upper limit (b) 0 0.05 0.1 0.15 0.2 0.25 0.3 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y V ib ra ti o n o r R a tt li n g Helicopter DNL Data Collection Site: Las Vegas Proportion Lower limit Upper limit (c) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y V ib ra ti o n o r R a tt li n g Helicopter DNL Data Collection Site: Washington, D.C. Proportion Lower limit Upper limit Figure 5-11. Proportion (with 95% CIs) of respondents highly annoyed by helicopter in-home vibration or rattling within (a) Long Beach, (b) Las Vegas, and (c) D.C. interviewing sites. Asymmetric CIs were calculated using the Clopper-Pearson method. Hypothesis 4. The prevalence of annoyance due to helicopter noise is heavily influenced by indoor secondary emissions (rattle and vibration) due to its low-frequency content. Binary logistic regression analyses were conducted for high annoyance due to in-home vibration/ rattling as well as other helicopter sounds: BVI (thumping or slapping), buzzing, and whining. Table 5-6 summarizes these analyses. No statistically significant relationship was observed between annoyance due to in-home vibration and rattling and annoyance due to noise level alone. The dosage-response relationship between helicopter noise exposure and annoyance due to “buzzing” noises differed significantly from chance in Long Beach, but not in Las Vegas or Washington, D.C. Figures 5-11 through 5-14 show proportions of reports of high annoyance for each of the specific noise types.

Analyses of Noise Exposure Measurements and Interview Findings 73 (a) (b) (c) 0 0.04 0.08 0.12 0.16 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y T h u m p in g o r S la p p in g Helicopter DNL Data Collection Site: Long Beach Proportion Lower limit Upper limit 0 0.05 0.1 0.15 0.2 0.25 0.3 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.29 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y T h u m p in g o r S la p p in g Helicopter DNL Data Collection Site: Las Vegas Proportion Lower limit Upper limit 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y T h u m p in g o r S la p p in g Helicopter DNL Data Collection Site: Washington Proportion Lower limit Upper limit Figure 5-12. Proportion (with 95% CIs) of respondents highly annoyed by helicopter thumping and slapping (BVI) noise at (a) Long Beach, (b) Las Vegas, and (c) D.C. interviewing sites. Asymmetric CIs were calculated using the Clopper-Pearson method.

74 Assessing Community Annoyance of Helicopter Noise (a) (b) (c) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y B u zz in g Helicopter DNL Data Collection Site: Long Beach Proportion Lower limit Upper limit 0 0.05 0.1 0.15 0.2 0.25 0.3 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y B u zz in g Helicopter DNL Data Collection Site: Las Vegas Proportion Lower limit Upper limit 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y B u zz in g Helicopter DNL Data Collection Site: Washington Proportion Lower limit Upper limit Figure 5-13. Proportion (with 95% CIs) of respondents highly annoyed by helicopter buzzing noise within (a) Long Beach, (b) Las Vegas, and (c) D.C. interviewing sites. Asymmetric CIs were calculated using the Clopper-Pearson method.

Analyses of Noise Exposure Measurements and Interview Findings 75 Figure 5-14. Proportion (with 95% CIs) of respondents highly annoyed by helicopter whining noise within (a) Long Beach, (b) Las Vegas, and (c) D.C. data collection sites. Asymmetric CIs were calculated using Clopper-Pearson method. (a) (b) (c) 0 0.04 0.08 0.12 0.16 0.2 0.24 0.28 0.32 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y W h in in g Helicopter DNL Data Collection Site: Long Beach Proportion Lower limit Upper limit 0 0.05 0.1 0.15 0.2 0.25 0.3 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y W h in in g Helicopter DNL Data Collection Site: Las Vegas Proportion Lower limit Upper limit 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 27.2-32.00 32.01-35.37 35.38-38.73 38.74-42.10 42.20-45.47 45.48-48.84 48.85-52.2 9 5 % C I P ro p o rt io n H ig h ly A n n o y e d b y W h in in g Helicopter DNL Data Collection Site: Washington Proportion Lower limit Upper limit

76 Assessing Community Annoyance of Helicopter Noise The logistic regression of buzzing noises on helicopter noise exposure was the only one that was unlikely to have arisen by chance alone, but even it accounted for very little variance in the relationship between annoyance and exposure. In the apparent absence of any strong association between helicopter noise exposure and annoyance at the low exposure levels that were available for study, it is likely that nonacoustic factors may have controlled community response to helicopter noise at the study sites. Hypothesis 5. The prevalence of annoyance due to helicopter noise is heavily influenced by nonacoustic factors. It is clear from the differences in response in the Long Beach and Las Vegas communities that nonacoustic factors strongly influence community response in these communities. Las Vegas had approximately ten times the number of flights, albeit at a higher alti- tude, and yet a substantially reduced fraction of the population were highly annoyed. The higher altitude effect on DNL (in the range of a 3 to 4 dB reduction) was nowhere near the effect of the higher number of operations on DNL (plus 10 dB). Aircraft fleet mix cannot account for the difference either. In Washington, D.C., the public concern over moved fixed-wing flight tracks negated the dosage-response effect for both fixed-wing and helicopter noise. No acoustic factors can account for these differences. Note that the literature (Fidell et al. 2011) discusses a myriad of nonacoustic factors that can contribute to people’s attitude to noise. The primary nonacoustic factors are fear and distrust. Certainly, the low altitudes of helicopters could be contributing to fear. Other factors that may be playing a role are expectations, invasion of privacy, the apparent need for the helicopter operations, or a perception that not enough is being done to control helicopter noise. Predicting Helicopter Noise Annoyance from Annoyance Due to Other Noise Sources. As an indication of possible individual differences and/or response bias, a binary multiple logistic regression examined whether high annoyance by fixed-wing aircraft and high annoyance by traffic noise (very or extremely annoyed) predicted whether an individual was highly annoyed by helicopter noise. Site was included as a predictor, along with interactions between site and the other two sources of noise annoyance to account for differences among sites. Data for this analysis were provided by 2,197 of the 2,272 respondents (others reported not being at home during the period in question). The eight-predictor model showed prediction of high annoyance that was significantly better than would be expected by chance, c2 (8, N = 2,197) = 178.59, p < .001. The fit of the model to the data was very good, Hosmer-Lemeshow c2 (2, N = 2,197) = 0.531, p = .767 (where p = 1.0 indicates perfect fit); the variance in high annoyance due to helicopter noise is accounted for moderately well, Nagelkerke R2 = .20. Table 5-7 shows the results of the logistic regression. Variable B Standard Error Wald df p Odds ratio (Be) 95% CI for OR Lower Upper LGB vs. DCA 1.406 0.429 10.74 1 .001 4.078 1.760 9.452 LAS vs. DCA 0.143 0.487 0.09 1 .789 1.153 0.444 2.996 Fixed-wing aircraft 2.685 0.465 33.36 1 <.001 14.658 5.894 36.457 Traffic 2.159 0.701 9.49 1 .002 8.667 2.193 34.243 LGB vs. DCA by fixed-wing aircraft -0.354 0.506 0.34 1 .560 0.702 0.214 2.303 LAS vs. DCA by fixed-wing aircraft 0.023 0.816 <0.01 1 .978 1.023 0.207 5.064 LGB vs. DCA by traffic -0.825 0.758 1.19 1 .276 0.438 0.099 1.936 LGB vs. DCA by traffic 0.643 0.897 0.51 1 .474 1.902 0.328 11.036 Constant -4.000 0.410 B = the customary symbol for slope; "Wald" = the value of a Wald test for the significance of the slope; "df" = the usual abbreviation for degrees of freedom; p = the customary symbol for significance; the odds ratio is a measure of an association of exposure and an outcome; CI = the confidence interval; OR = odds ratio. Table 5-7. Logistic regression analysis of high annoyance due to helicopter noise as a function of high annoyance due to other noise sources, data collection site, and interactions.

Analyses of Noise Exposure Measurements and Interview Findings 77 Reporting high annoyance with helicopter noise is predicted by reports of high annoyance by traffic and fixed-wing noise sources, and by whether respondents lived in Long Beach versus Washington, D.C. Respondents who were highly annoyed by fixed-wing aircraft noise were almost fifteen times more likely to be highly annoyed by helicopter noise than those who were not highly annoyed by fixed-wing aircraft noise. Respondents who were highly annoyed by traffic noise were more than 8.5 times as likely to be annoyed by helicopter noise. Residents of Long Beach were about four times more likely to be highly annoyed by helicopter noise than were residents of D.C. (An earlier analysis, not shown, indicated that residents of Long Beach were about three times as likely to be highly annoyed by helicopter as those living in Las Vegas, p < .001.) None of the interactions between site and noise type differed significantly from chance, p > .05. Thus, the analysis suggests fairly strong individual differences in reporting high annoyance due to different noise sources. In other words, a respondent who reported high annoyance to any other noise source was much more likely to be annoyed by helicopters. This adds to the common belief in varying levels of noise sensitivity, but it does not rule out that this may be associated with nonacoustic variables such as expectations. Hypothesis 6. The prevalence of annoyance due to helicopter noise is heavily influenced by proximity to helicopter flight paths. Binary logistic regression analysis was used to determine whether proximity to the flight track influences a high degree of annoyance due to helicopter flight paths. At Long Beach, the dosage-response relationship was small (Nagelkerke R2 = .018) but statistically significant, c2(1, N = 1,050) = 8.70, p = .003. Odds ratio (Be) was 0.279 (indicating a negative relationship between distance and annoyance) with 95% confidence limits from 0.117 to 0.662. The dosage-response relationship failed to approach statistical significance at Long Beach, p = .664. Thus, proximity to flight path is as good a predictor of high annoyance as noise level. The relationship of annoyance to distance is discussed further in Section 5.5.2. Hypothesis 7. Complaints lodged about helicopter noise are more reliable predictors of the prevalence of annoyance than measures of exposure to helicopter noise or proximity to helicopter flight paths. Complaints by Annoyance. Only a very few respondents (2.6%) indicated that they had ever registered complaints (Item 15). However, a Chi-square analysis of whether respondents complained by whether they were at least slightly annoyed by heli- copter noise revealed a statistically significant relationship, c2(1, N = 2,167) = 73.70, p < .001, Cramer’s V = .19. Among the 1,937 respondents who reported no annoyance by helicopter noise, 1.3% complained; of the 330 respondents who reported at least slight annoyance by helicopter, 9.4% registered complaints. Thus, a reasonably clear relationship was found between the preva- lence of annoyance (in any degree) and complaint behavior. Complaints by Noise Exposure. A 2 × 2 analysis of variance examined whether complaining (yes or no) was related to noise exposure or site or their interaction. There was no statistically significant difference in noise exposure for those who did and did not complain, p = .722, nor was there a significant interaction with the site p = .649. The difference between sites was statistically significant, F(2, 2155) = 5.36, p = .005 but small, h2 = .005. Note that this correlation analysis refers to noise complaints as those provided in the survey response, i.e., did the responder file a noise complaint. This analysis is not referring to the noise complaints filed with the airports. Unfortunately, the noise complaints collected by the airports either did not segregate helicopter complaints from fixed-wing, were not geocoded and available for GIS analysis, or both. 5.4.1.6 Additional Relationships with Helicopter Noise Exposure Were Helicopters Noticed? A between-subjects two-way (site by notice of helicopters) analysis was conducted of noise exposure. Statistically significant main effects for both site and frequency category were observed, but no interactions were noted, as seen in Figure 5-15.

78 Assessing Community Annoyance of Helicopter Noise Helicopter noise exposure was greater for those who noticed helicopters (M = 43.27, SE = 0.192) than for those who did not notice helicopters (M = 41.73, SE = 0.191), F(1, 2107) = 32.17, p < .001, but the relationship was weak, partial h2 = .02, with 95% confidence limits from .01 to .03. Figure 5-9a, b, and c show that the ranges of helicopter noise exposure levels, from the low 30 dB to low 50 dB range, was similar at all three sites. Frequency of Notice of Helicopters. Categories of frequency of noticing helicopters are in Table 5-3. A between-subjects two-way (site by frequency category) analysis was conducted of noise exposure, with planned trend analysis. Statistically significant main effects were observed for both site and frequency category, but no interactions were noted. The relationship between noise exposure and categories of frequency of noticing helicopters was statistically significant, F(3, 2020) = 17.34, p < .001, but moderate, partial h2 = .025, with 95% confidence limits from .01 to .04. Linear and cubic trends were statistically significant, with p < .001 and .013, respectively. As seen in Figure 5-16, the trend is at least speculatively consistent with a sigmoidal dosage-response function. In any event, over a small dynamic range notice- ability increased with increasing DNL. 5.5 Relationships Among DNL, Distance, and Percent Highly Annoyed This section examines two relationships observed in the data. The first shows the relation- ship between the modeled DNL during the week prior to interview and the distance from the flight corridor centerline. The second shows the relationship between annoyance and DNL during the week prior to interview. DNL and distance from a noise source are obviously highly correlated, but annoyance could conceivably be more closely related to proximity to direct overflights. Frequency of noticing helicopters H el ic o p te r- o n ly D ay -N ig h t A ve ra g e S o u n d L ev el , d B Figure 5-15. Plot of noticing helicopters as a function of DNL.

Analyses of Noise Exposure Measurements and Interview Findings 79 5.5.1 DNL Versus Distance Relationships Figures 5-16 through 5-18 show DNL versus distance relationships for Long Beach, Las Vegas, and Washington D.C., respectively. They show orderly reductions in SELs with distance. The Long Beach data has the greater variance likely due to a much greater dispersion of flight tracks within a corridor and the existence of two corridors affecting the survey area, the Cherry Avenue corridor, and the split in the Redondo corridor into a westbound and eastbound leg at the coast- line. For those respondents living directly under the corridor (i.e., within 0.1 nm of the centerline) the sound exposure does not change appreciably with distance. At 1 nm from centerline, DNL dropped by 19 and 17 dB, respectively, for Long Beach and Las Vegas. 5.5.2 Dosage-response Relationships The following paragraphs describe dosage-response relationships between SELs and the prevalence of annoyance. 5.5.2.1 Washington, D.C. Figures 5-19 through 5-22 show relationships between (A-weighted) DNL and the preva- lence of high annoyance observed among respondents at the Washington, D.C., interview site. Separate relationships are shown for fixed- and rotary-wing aircraft. The first relationship, for fixed-wing aircraft, shows the percent highly annoyed in Figure 5-19 and the number of respon- dents in Figure 5-20. Figure 5-21 (for helicopters) shows the percent highly annoyed. Figure 5-22 shows the number of respondents for helicopters. Figure 5-23 shows the annoyance of exposure to helicopter noise as a function of reciprocal distance [20 Log (1/distance), where distance is in nautical miles]. Thus, 0 dB on the logarithmic scale indicates 1 nautical mile. The multiplier of 20 was chosen because at distances greater that 25 30 35 40 45 50 55 0.01 0.10 1.00 D ay -N ig ht A ve ra ge S ou nd L ev el (d B) Distance from Corridor Centerline (nm) LGB Figure 5-16. INM-generated DNLs for each respondent at LGB as a function of respondent distance from two flight corridor centerlines.

80 Assessing Community Annoyance of Helicopter Noise Figure 5-17. INM-generated DNLs for each respondent at LAS as a function of respondent distance from flight corridor centerlines. 25 30 35 40 45 50 55 0.01 0.10 1.00 D ay -N ig ht A ve ra ge S ou nd L ev el (d B) Distance from Corridor Centerline (nm) LAS Figure 5-18. INM-generated DNLs for each respondent at DCA as a function of respondent distance from flight corridor centerline.

Analyses of Noise Exposure Measurements and Interview Findings 81 Figure 5-19. Percent of respondents highly annoyed at the Washington, D.C., interview site as a function of A-weighted DNL for fixed-wing aircraft. 0 5 10 15 20 25 30 25 30 35 40 45 50 55 60 65 Pe rc en t H ig hl y A nn oy ed Day-Night Average Sound Level (dB) CTL: 0.00 dB Number of data points: 6 Maximum likelihood curve t. Number of observations: 442 Asymptote: 15.16 percent 0 20 40 60 80 100 120 140 160 25 30 35 40 45 50 55 60 65 N u m b er o f O b se rv at io n s in B in Day-Night Average Sound Level (dB) Number of data points: 6 Number of observations: 442 Figure 5-20. Number of respondents in each fixed-wing noise exposure category at the Washington, D.C., interview site (Bin = histogram bin).

82 Assessing Community Annoyance of Helicopter Noise Figure 5-21. Percent of respondents highly annoyed at the Washington, D.C., interview site as a function of A-weighted DNL for helicopters. 0 5 10 15 20 25 30 25 30 35 40 45 50 55 60 65 P er ce n t H ig h ly A n n o ye d Day-Night Average Sound Level (dB) CTL: 0.00 dB Number of data points: 9 Maximum likelihood curve fit. Number of observations: 442 Asymptote: 4.75 percent Figure 5-22. Number of respondents in each helicopter noise exposure category at the Washington, D.C., interview site. 0 20 40 60 80 100 120 140 160 25 30 35 40 45 50 55 60 65 N u m b er o f O b se rv at io n s in B in Day-Night Average Sound Level (dB) Number of data points: 9 Number of observations: 442

Analyses of Noise Exposure Measurements and Interview Findings 83 Figure 5-23. Percent of respondents highly annoyed at the Washington, D.C., interview site as a function of distance from helicopter corridor. 0 5 10 15 20 25 30 -5 0 5 10 15 20 25 30 35 40 45 50 P er ce n t H ig h ly A n n o ye d 20 Log [1 / distance (nm)] (dB) CTL: 0.00 dB Number of data points: 9 Maximum likelihood curve fit. Number of observations: 442 Asymptote: 4.75 percent a few hundred feet, most INM noise-power-distance (NPD) curves drop off at that rate when SEL is plotted as a function of log [distance]. Figure 5-24 shows the number of interviews at each distance. 5.5.2.2 Las Vegas Both A- and C-weighted measurements of DNL were available for analysis at the Las Vegas interviewing site. Wind-related, low-frequency noise measurement artifacts were less severe at LAS 0 20 40 60 80 100 120 140 160 180 200 220 -5 0 5 10 15 20 25 30 35 40 45 50 N u m b er o f O b se rv at io n s in B in 20 Log [1 / distance (nm)] (dB) Number of data points: 9 Number of observations: 442 Figure 5-24. Number of respondents in each helicopter noise exposure category at the Washington, D.C., interview site.

84 Assessing Community Annoyance of Helicopter Noise than at LGB. Figure 5-25 plots the prevalence of percent highly annoyed against (A-weighted) DNL. Figure 5-26 shows the number of survey respondents in each exposure bin. No obvious trend of increasing annoyance with increasing noise level was observed: annoyance is nearly constant at all noise exposure levels. If there is a sigmoid function to the data for Las Vegas, the increase in annoyance with dose must occur at much higher noise levels than were encountered in LAS. The result is that all of the data are on the asymptote. This asymptote is at about 2 percent highly annoyed independent of noise exposure. Significantly, the asymptote does not go to zero at low noise exposure levels. Figure 5-25. LAS, percent highly annoyed as a function of A-weighted DNL for helicopters. 0 5 10 15 20 25 30 25 30 35 40 45 50 55 60 65 P er ce n t H ig h ly A n n o ye d Day-Night Average Sound Level (dB) CTL: 0.00 dB Number of data points: 11 Maximum likelihood curve fit. Number of observations: 741 Asymptote: 2.02 percent Figure 5-26. LAS, number of respondents for each helicopter survey point. 0 20 40 60 80 100 120 140 160 180 200 220 25 30 35 40 45 50 55 60 65 N u m b er o f O b se rv at io n s in B in Day-Night Average Sound Level (dB) Number of data points: 11 Number of observations: 741

Analyses of Noise Exposure Measurements and Interview Findings 85 Figure 5-27 shows the percent highly annoyed as a function of the C-weighted DNL. The C-weighting includes low-frequency noise far more effectively than does the A-weighting. Figure 5-28 shows the number of survey respondents for each survey bin. The C-weighted DNL response curve is similar to the A-weighted DNL, or in other words, flat. The asymptote shows a flat 2% highly annoyed independent of noise exposure, even accounting for the low-frequency noise. In the hypothesis that annoyance response is a function of acoustic and nonacoustic parameters, nonacoustic parameters must be the dominating response. 0 5 10 15 20 25 30 40 45 50 55 60 65 70 75 80 P er ce n t H ig h ly A n n o ye d Day-Night Average Sound Level (dB) CTL: 0.00 dB Number of data points: 6 Maximum likelihood curve fit. Number of observations: 741 Asymptote: 2.02 percent Figure 5-27. LAS, percent highly annoyed as a function of C-weighted DNL for helicopters. 0 20 40 60 80 100 120 140 160 180 200 220 40 45 50 55 60 65 70 75 80 N u m b er o f O b se rv at io n s in B in Day-Night Average Sound Level (dB) Number of data points: 6 Number of observations: 741 Figure 5-28. LAS, number of respondents for each helicopter C-weighted survey point.

86 Assessing Community Annoyance of Helicopter Noise Figure 5-29 shows response as a function of reciprocal distance in the same manner as for Washington, D.C. Other than a singular point, there is no clear trend of increasing annoyance with decreasing distance to the helicopter corridor. Figure 5-30 shows the number of survey respondents for each survey bin. 5.5.2.3 Long Beach Dosage-response graphs for Long Beach are shown in Figures 5-31 and 5-33 for the A-weighted DNL and reciprocal distance, respectively. Figures 5-32 and 5-34 show the number of respondents for each survey point. Figure 5-29. LAS, percent highly annoyed as a function of distance from helicopter corridor. 0 5 10 15 20 25 30 -5 0 5 10 15 20 25 30 35 40 45 50 P er ce n t H ig h ly A n n o ye d 20 Log [1 / distance (nm)] (dB) CTL: 0.00 dB Number of data points: 22 Maximum likelihood curve fit. Number of observations: 741 Asymptote: 2.02 percent Figure 5-30. LAS, number of respondents for each helicopter distance. 0 20 40 60 80 100 120 140 160 180 200 220 -5 0 5 10 15 20 25 30 35 40 45 50 N u m b er o f O b se rv at io n s in B in 20 Log [1 / distance (nm)] (dB) Number of data points: 22 Number of observations: 741

Analyses of Noise Exposure Measurements and Interview Findings 87 Figure 5-31. LGB, percent highly annoyed as a function of A-weighted DNL for helicopters. 0 5 10 15 20 25 30 25 30 35 40 45 50 55 60 65 P er ce n t H ig h ly A n n o ye d Day-Night Average Sound Level (dB) CTL: 68.93 dB Number of data points: 13 Maximum likelihood curve fit. Number of observations: 1089 Asymptote: 4.37 percent 0 20 40 60 80 100 120 140 160 180 200 220 25 30 35 40 45 50 55 60 65 N u m b er o f O b se rv at io n s in B in Day-Night Average Sound Level (dB) Number of data points: 13 Number of observations: 1089 Figure 5-32. LGB, number of respondents for each helicopter survey point.

88 Assessing Community Annoyance of Helicopter Noise Figure 5-33. LGB, percent highly annoyed as a function of distance from helicopter corridor. 0 5 10 15 20 25 30 -5 0 5 10 15 20 25 30 35 40 45 50 P er ce n t H ig h ly A n n o ye d 20 Log [1 / distance (nm)] (dB) CTL: 43.94 dB Number of data points: 18 Maximum likelihood curve fit. Number of observations: 1088 Asymptote: 4.51 percent Figure 5-34. LGB, number of respondents for each helicopter distance. 0 20 40 60 80 100 120 140 160 -5 0 5 10 15 20 25 30 35 40 45 50 N u m b er o f O b se rv at io n s in B in 20 Log [1 / distance (nm)] (dB) Number of data points: 18 Number of observations: 1088

Analyses of Noise Exposure Measurements and Interview Findings 89 The Long Beach dosage-response curve shows an increasing level of annoyance with increasing A-weighted DNL. This was the only survey site of the three sites where this clear trend is shown. Of note is the fact that the percent highly annoyed does not go to zero at lower noise exposures, but, in fact, the asymptote flattens out at about 4 percent highly annoyed no matter how low the DNL. Again, the hypothesis that annoyance response is composed of acoustic and nonacoustic response suggests that there are nonacoustic reasons that 4 percent of the population is highly annoyed with helicopters independent of noise dose. Figure 5-33, the relation of percent highly annoyed to the reciprocal of distance, also shows a trend of higher annoyance with closer distance, but with much higher unexplained scatter in the data at higher DNL. 5.5.3 Dosage-response Relationship for Combined Sites Figure 5-35 shows the dosage-response results for all three sites on the same plot. The solid lines represent the actual range of survey data and the dashed lines represent the curve developed from data extrapolated further out. Clearly each site is unique, indicating that each community has a unique response. The presence of residual annoyance as shown by the asymptote is a significant finding. It may indicate that the reason for apparent elevated helicopter complaints over those of fixed-wing has little to do with people’s differing sensitivity to noise levels from the two sources. However, whatever is underlying the observed residuals results in people being annoyed where similar levels from fixed-wing aircraft would likely result in zero high annoyance (mean- ing the helicopter annoyance is spread over a much larger geographic area than would otherwise have been predicted). Even a few percent highly annoyed over a vastly larger land area could add up to a “critical mass” of annoyed citizens. This is an unexpected but very real phenomenon. 0 2 4 6 8 10 12 14 16 18 20 25 30 35 40 45 50 55 60 65 P er ce n t H ig h ly A n n o ye d Day-Night Average Sound Level (dB) Long Beach Long Beach (data range) Las Vegas Las Vegas (data range) Washington D.C. Washington D.C. (data range) Maximum likelihood curve fits 4.4% 4.8% 2.0% Figure 5-35. Composite results for all three sites.

90 Assessing Community Annoyance of Helicopter Noise 5.6 Results of Low-Frequency Noise Analysis Low-frequency noise emissions of helicopters are of particular concern as identified in Chapter 1 of this report. Fixed-wing jet aircraft noise consists of broadband noise spread over the audio spectrum, but helicopter noise is characterized by distinct frequency characteristics. Most helicopter noise is concentrated at lower frequencies. The following sections describe the results of the low-frequency noise analysis. 5.6.1 Measuring Low-Frequency Helicopter Noise Most sound level meters include the ability to measure A- and C-weighted decibels, and using C-weighted decibels will capture the low-frequency components of helicopter noise. The downside to using the C-weighted decibel is that it does not identify if the noise is down in the range where rattle and vibration are induced which, as identified by the Low-Frequency Noise Expert Panel, is at frequencies below 80 Hz inclusive. A more advanced method of identifying LFSLs is by measuring noise in 1⁄3 octave bands. This produces not one measure of a sound level, but 36 individual measures of a sound level, one for each 1⁄3 octave band from 6 Hz to 20,000 Hz. An even more advanced method using narrow band analysis divides the spectrum into 400 narrow bands for even higher resolution. Another consideration in the measurement of helicopter noise is the time weighting. This is a complex topic that is difficult to simplify. Basically, the human response to a changing sound level is not instantaneous. In the days of sound level meters with a moving needle, the time averaging was done using a “slow” or a “fast” response that controlled how fast the needle moved. Slow response was generally used and was designed to approximate the human ear response to changing sound. Another weighting was developed for very short duration noise, such as a gunshot. This time weighting is called impulse weighting. With the advent of digital sound measurement devices, the slow and fast weightings are obsolete and instead a 1-second equivalent sound level is measured. This represents all of the acoustic energy contained within 1 second of time no matter how sudden the sound is. But short duration sounds such as gunshots or the impulsive noise of a helicopter noise is averaged into that 1 second. During LAS and LGB measurement programs the A- and C-weighted impulsive noise was also measured along with the 1 second equivalent sound level data. 5.6.2 Modeling the Low-Frequency Noise Level of Helicopters The INM and now AEDT include the capability to calculate both A-weighted and C-weighted noise levels as well as noise levels based on EPNL, a 1⁄3 octave band based metric that was devel- oped to reflect human perception of noisiness, not loudness, that includes penalties for pure tones. However, the database of aircraft noise levels built into INM and AEDT do not have data for frequencies below 50 Hz. One goal of this analysis is to determine if this deficiency precludes meaningful use of INM and AEDT for low-frequency studies of helicopter noise (note that there is no issue with the database as it is for A-weighted metrics). 5.6.2.1 Noise Measurement Data Collected for this Study Noise measurements were made during the LAS and LGB studies. The measurement systems, described earlier, included the measurement of the A- and C-weighted decibel and the 1⁄3 octave band data from 6 Hz to 20,000 Hz. The impulse A- and C-weighted sound pressure level was also recorded. A special discussion of measuring low-frequency noise is warranted here. Sound measurement systems consist of a microphone and windscreen combination connected by cable to the recording sound level meter. The windscreen is designed to remove the sounds of the

Analyses of Noise Exposure Measurements and Interview Findings 91 wind passing over the microphone grid. In general, and what was used for this study, a 4-inch windscreen made of open cell foam is used. As wind speed increases, the noise of the wind over the windscreen increases, especially at low frequencies. During the measurements at Las Vegas, wind was consistently very low and made a better dataset to test response to low-frequency noise. The Long Beach data, while having periods of calm was more generally windy, consistent with the coastal location: two weather fronts moved in through the study area during the survey. For this reason, the low-frequency response data were analyzed using the Las Vegas data. 5.6.2.2 Processing the LFSL Data The measurement system collected data for each 1 second of every day that include the afore- mentioned A-weighted, C-weighted, and 1⁄3 octave band data. These data were used to build a large database that included all the data for all four sites for the 7 days of measurement. The following method was used to analyze the data: 1. Aircraft radar data was obtained from the airport. Helicopter noise events were identified by matching the noise event time to the time of helicopter point of closest approach to the noise monitoring site. 2. A database was generated that included only helicopter noise events at each site. Each event consisted of the helicopter type and one record of data for each second of the noise event. The events were defined by the time at which the A-weighted level exceeded 55 dB and the time at which the event noise dropped below 55 dBA. This threshold allowed for isolating the helicopter noise from ambient noise as well as possible. Since all four measurement sites were in quiet residential areas, ambient noise levels were low with only passing cars as a significant intrusion. The database consisted of 110,821 1-second records in the helicopter event database. 3. For each 1-second record, the C-weighted sound pressure level was calculated using all of the available 1⁄3 octave bands and once again not using any 1⁄3 octave data below 50 Hz (to simulate the C-weighted data as would be computed by INM or AEDT). 4. For each 1-second record the LFSL was computed for that 1 second using the original definition of LFSL and expanding the definition of LFSL to include lower 1⁄3 octave bands. LFSL was recalculated with lower frequency bands down to and including 16 Hz, 10 Hz, and 6 Hz. 5. For each helicopter noise event at each site the SEL was computed using the A, C, and LFSL scale and using the A-weighted and C-weighted impulse scales. 5.6.3 Results of Low-Frequency Data Analysis An example of the sound spectrum in terms of 1⁄3 octave band sound pressure level is shown in Figure 5-36. The spectrum shown is for 1-second records with the highest LFSL, most strongly influenced by the high levels in the 20 and 25 Hz 1⁄3 octave bands. 5.6.3.1 Frequencies Used for LFSL Calculations The LFSL calculation was run using the original definition of 25 to 80 Hz as well as using lower frequency bands of 16 Hz, 10 Hz, and 6 Hz. There was significant difference between LFSL calculations based on 25 and 16 Hz lower bands, typically in the range of 5 dB. The difference between LFSL based on 16, 10, or 6 Hz was about 0.1 dB. Therefore, for the purposes of this study, LFSL was redefined as the arithmetic average of the 1⁄3 octave band sound pressure levels from 16 to 80 Hz and is labeled LFSL16. 5.6.4 Comparison of Low-Frequency Metrics to A-Weighted Metric Table 5-8 lists the various noise exposure metrics for the four measurement sites in Las Vegas. Included are the energy average SEL for all helicopter events in terms of the A, C, impulse A, and

92 Assessing Community Annoyance of Helicopter Noise impulse C scales. Also calculated and shown is the SEL for C-weighting using only the frequency data available in INM and AEDT, i.e., frequencies above 50 Hz inclusive. Note that the distance from the measurement site to the centerline of the helicopter corridor is also provided. The sites are not numbered in order of distance. The first observation is that the SEL computed using only the INM/AEDT frequencies differs substantially from the true C-weighted SEL. This is important because it means that INM or AEDT can not be used for analyzing low-frequency noise in the study. INM nor AEDT can be used to compute C-weighted DNL for the social survey data. However, the measurement data can be used to convert the A-weighted DNL data computed by the noise model into C-weighted DNL. Figure 5-37 shows the relation of A-weighted SEL to C-weighted SEL as a function of distance to the helicopter tracks. The attenuation of sound with distance is highly dependent on the sound frequencies. For example, if the air temperature is 15 degrees, the sound frequency is Energy Average Maximum of All Events Arithmetic Average of Max LFSL Close Ap ft A- weighted SEL C- A- weighted impulse SEL C- weighted impulse SEL 1 394 77.3 87.4 83.1 85.4 90.9 85.0 87.6 74.0 78.6 2 1,864 74.7 89.0 82.7 78.6 91.6 84.7 87.1 74.3 79.1 3 2,419 73.3 86.3 80.3 77.2 88.9 81.7 84.3 72.1 76.8 4 762 77.3 87.7 82.8 86.6 91.6 85.5 89.9 73.4 78.8 Note: SELcinm is a C-weighted SEL, as calculated by FAA’s INM (now AEDT) software. INM has no information about the acoustic energy of aircraft noise in frequency regions lower than the 50 Hz ¹⁄³ octave band. Site pr. weighted SEL SELcinm LFSL LFSL LFSL16LFSL16 Table 5-8. A-weighted and low-frequency metrics at four measurement sites in LAS. Figure 5-36. Sample spectrum for typical helicopter at the LAS interviewing site. 0 10 20 30 40 50 60 70 80 90 la la m ax i lc lc m ax i t6 t8 t1 0 t1 2 t1 6 t2 0 t2 5 t3 1 t4 0 t5 0 t6 3 t8 0 t1 00 t1 25 t1 60 t2 00 t2 50 t3 15 t4 00 t5 00 t6 30 t8 00 t1 00 0 t1 25 0 t1 60 0 t2 00 0 t2 50 0 t3 15 0 t4 00 0 t5 00 0 t6 30 0 t8 00 0 t1 00 00 t1 25 00 t1 60 00 t2 00 00 So un d Pr es su re L ev el , d B A, C, and 1/3 Octave Bands Spectrum at Max LFSL Helicopter EC 130

Analyses of Noise Exposure Measurements and Interview Findings 93 50 Hertz, and if the atmosphere relative humidity is 40% then the attenuation due to this atmo- sphere is 0.111 dB per km. For the same conditions but for 500 Hertz, the attenuation is 2.18 dB per km, and at 5,000 Hertz, the attenuation is 69.7 dB per km. So, in this example, the attenu- ation over any reasonable distance from a helicopter (e.g., 500 or 1,000 meters) at the lowest fre- quencies is essentially 0; at middle frequencies, it is a few dB; and at high frequencies, nearly all of the sound is eliminated. The A-weighted and C-weighted metric differs with distance because the atmosphere absorbs high-frequency sounds very efficiently and is very poor at absorbing low-frequency sounds. At larger distances, the low-frequency component of helicopter noise is heard more than the higher frequencies, which affect the A-weighted metric more, because the atmosphere has absorbed the high-frequency sounds. The difference between A-weighted and C-weighted SEL as a function of distance can be used to convert the social survey receptor A-weighted DNL to an estimate of C-weighted DNL. This was calculated, and then used to create a C-weighted dosage-response curve, shown in Figure 5-28. As evidenced from Table 5-8 and Figure 5-37, the C-weighted metric has a higher value than the A-weighted metric due to the concentration of low-frequency noise in the range of 16 to 80 Hz. Table 5-9 also shows the energy average SEL in terms of the A-weighted impulse scale and C-weighted impulse scale. Again, these values are also significantly higher than the normal A-weighted SEL. Figure 5-38 plots the A-weighted impulse SEL against the normal A-weighted SEL. Impulse weighting, even with the heavy discounting of low-frequency noise by the A scale, shows a significant increase in level. Lastly, the LFSL16 can be compared to the energy average A-weighted SEL. This is a bit of mixed comparison and is done with some caution. SEL is a measure of exposure, i.e., the acoustic energy y = 0.0019x + 9.348 R2 = 0.76812 Figure 5-37. The relation of A- and C-weighted SEL for the LAS measurement data. (Wtd = weighted.) Differences Relative to A-Weighted SEL Site 1 394 10.1 1.3 8.1 13.6 2 1,864 14.3 4.4 3.8 16.9 3 2,419 13.1 3.5 3.9 15.7 4 762 10.4 1.4 9.3 14.2 Average max LFSL16 - A-weighted SEL A-weighted impulse SEL - A-weighted SEL Close Appr. ft C-weighted SEL - A-weighted SEL C -weighted impulse SEL - A-weighted SEL Table 5-9. Differences in A-weighted and low-frequency metrics, LAS.

94 Assessing Community Annoyance of Helicopter Noise during an entire event. LFSL is defined as a value at the time of a maximum. LFSL was defined this way because rattle either occurs or does not occur and any attempt to average LFSL, energy or arithmetic, will blur the ability to predict rattle. Figure 5-39 compares A-weighted SEL with LFSL16. This comparison shows that the low-frequency components of helicopter noise have a significant potential to cause rattle that cannot be predicted from A-weighted SEL. 5.6.4.1 Summary of Low-Frequency Noise Analysis Figures 5-37 through 5-39 all have nearly identical slopes. That means that C-weighted, A-weighted impulse, and LFSL16 have nearly identical relationships to the A-weighted decibel. This means that understanding response to civil helicopter noise will not be enhanced by using special low-frequency or impulse metrics. Table 5-8 summarizes the differences between the various metrics. 5.7 Noise Complaint Data 5.7.1 Long Beach Helicopter Noise Complaints Table 5-10 shows the year 2015 helicopter noise complaints as recorded by the city of Long Beach. Of these 878 complaints, 89 occurred during the month of July (during which the survey was done). There is also another helicopter noise complaint database being built by the FAA as part of the LA Helicopter Initiative for all of the LA area. The City of Long Beach provides its Figure 5-38. Relation of A- and A-weighted impulse SEL (SELAi ) for LAS measurement data. y = 0.0019x + 9.348 R2 = 0.76812 Figure 5-39. Relationship of A-weighted SEL to LFSL16. y = 0.0014x + 0.6877 R2 = 0.78479

Analyses of Noise Exposure Measurements and Interview Findings 95 complaint data to the FAA, so the Long Beach data is a subset of the FAA database. The FAA database included 110 Long Beach complaints during the survey period. The Long Beach and the FAA databases include a field for address but it is often populated with a telephone number and not an address. For those 110 complaints in Long Beach during the social survey that did include at least a street name, the majority are in the study area. 5.7.2 Las Vegas Helicopter Noise Complaints Clark County Division of Aviation recorded 3,963 noise complaints during the year 2015, of which 59 were helicopter noise complaints. None of the noise complaints were in the survey area, although two were just outside the study area (Source: Memorandum, Department of Aviation, “October, November, December and Annual 2015 Noise Complaint Reports,” Clark County Division of Aviation, January 28, 2016). 5.7.3 Washington, D.C., Area Helicopter Noise Complaints The MWAA reported a total of 8,670 noise complaints for all aircraft in the year 2015. Of these, 343 were from Arlington and 7,930 were from NW Washington (Source: “2015 Annual Aircraft Noise Report,” MWAA, undated). The MWAA does not segregate noise complaints by fixed-wing or helicopter and there is not a way to recover which complaints were helicopter based. Month Helicopter Complaints January February March April May June July August September October November December 38 23 29 32 28 79 89 131 59 174 136 60 Source: LGB Airport Noise Office. Table 5-10. Long Beach helicopter complaints during the year 2015.