A fundamental goal of aircraft noise regulation is control of the prevalence of noise-induced annoyance in airport communities. A common regulatory strategy is to identify values of longterm, time-weighted average aircraft noise exposure that may not be exceeded in the vicinity of airports without certain consequences. Noise exposure per se is neither the sole cause nor a perfect predictor of the annoyance of aircraft noise, however. Regulatory limitation of noise levels to certain values of favored noise metrics may therefore provide the appearance, rather than the substance, of a solution to problems of community reaction to aircraft noise. Response bias, as identified by Green and Fidell (1991), is a factor that exerts about as much influence on the observed prevalence of annoyance in communities as cumulative noise exposure. The importance of formal consideration of response bias in assessments of the adequacy of regulatory enforcement levels is addressed.
|How to cite this article:|
Fidell S. Assessment of the effectiveness of aircraft noise regulation. Noise Health 1999;1:17-26
| Introduction|| |
Like other forms of regulation, aircraft noise regulation seeks to balance competing societal interests. Conferring a degree of protection on public health and welfare from adverse effects of the noise of aircraft operations is generally understood to be the primary purpose of aircraft noise regulation. Protection of public health and welfare is not the only purpose, however. Aircraft noise regulation is also intended to provide certain protections for aircraft operators and airport proprietors from unfavourable community reactions to noise exposure, and to maintain efficient and affordable air transportation services for society at large.
Logically defensible regulation of aircraft noise requires unambiguous, quantitative goals. In principle, regulation could be intended to protect specified proportions of the residential population in certain age groups from awakening due to aircraft noise more than a certain number of times per night; or to minimise speech interference for a stated percentage of the time; or to limit the prevalence of a consequential degree of long-term, noise-induced annoyance to a some level. In practice, neither the nature nor degree of regulatory protections for various parties, nor the proportion of the noise-exposed population expected to benefit from regulatory protections, are well defined.
Standard approach to aircraft noise regulation
The standard approach to aircraft noise regulation in the vicinity of airports worldwide is to identify a value of an acoustically measurable quantity - typically, some form of cumulative, time-weighted average sound level - that may not be exceeded within certain geographic areas without certain consequences. Focusing attention on regulatory enforcement levels rather than on avoidance of adverse noise effects encourages unending debate about minutiae of aircraft noise metrics. It also helps many to forget that the primary purpose of regulation is to protect people from annoyance, not sound level meters from noise exposure.
The strategy of specifying seemingly precise, one-size-fits-all regulatory thresholds to provide unspecified degrees of mitigation of noise effects further encourages "reverse engineering" of such threshold values for tacit purposes other than protection of public health and welfare. For example, a narrow focus on establishing policy points for tolerable noise exposure levels permits regulatory agencies to avoid forthright explanations of the intended benefits of regulation, and to escape frank explanation of the unavoidably non-technical underpinnings of their policy preferences.
It is nonetheless generally understood that regulatory noise levels reflect not merely the primary purpose of aircraft noise regulation, but also the regulatory agency's charter, not to mention political, economic, and other pragmatic pressures on the agency. Detailed justifications for selection of regulatory noise exposure thresholds are rarely offered, and questions of how well regulation based on a threshold value of a noise metric accomplishes the fundamental goal of limiting adverse consequences of noise exposure are rarely addressed.
For example, controversy about noise exposure levels that are "compatible" with certain land uses begs the question of whether a compelling case can be constructed to define compatibility in units of decibels in the first place. The concept of compatibility (of neighbouring communities' land uses with airport operations) is not bidirectional, and is only tenuously linked to fundamental goals of minimising adverse effects of noise on people. Focusing attention on noise level thresholds (rather than on noise effects) for access to funds to mitigate noise exposure in areas of "incompatible" l and use further diverts attention from underlying regulatory goals. Not surprisingly, threshold values of long-term noise exposure are sometimes inappropriately cited in cases involving needs to mitigate speech interference (e.g., for acoustic isolation of classrooms from aircraft noise intrusions), or to limit sleep interference (cf. Pearsons et al., 1995).
Evolution of aircraft noise regulation
The framework for aircraft noise regulation evolved in the 1960s and 1970s, at a time of limited understanding of aircraft noise effects. Decisions about tolerable noise levels were based on incomplete information about the consequences of selecting particular values of noise metrics as thresholds of adverse effect. Given that quantitative goals for aircraft noise regulation were not explicitly stated, and that agencies charged with regulating noise had to reach decisions on the basis of uncertain information, it is hardly surprising that regulation does not always function as well as anticipated. When enough people believe that the benefits of existing regulation are not commensurate with its costs, attention eventually reverts to the fundamental purposes and strategies of aircraft noise regulation.
Although it is easy to lose sight of the fact, the de facto basis for aircraft noise regulation in the United States is limitation of the prevalence of noise-induced annoyance. If aircraft noise did not annoy people so greatly, and if tenths of decibels were not worth millions of dollars, nobody would go to the trouble of measuring aircraft noise with the loving exactitude routinely lavished on the enterprise 2 . The fact that annoyance is at the root of civil aircraft noise measurement is an awkward matter for some to acknowledge. Engineers often take greater interest in the details than in the purposes of aircraft noise measurement; researchers and consultants may be reluctant to confess the limited extent of contemporary understanding of noise-induced annoyance; acousticians and regulators are generally more comfortable discussing decibels than social policy; airport proprietors, airlines, and aircraft manufacturers may fear challenges to existing regulatory frameworks; and so forth.
Nonetheless, annoyance remains the fundamental rationale for aircraft noise regulation 3 . The U.S. Environmental Protection Agency (EPA, 1978) believes "For the purposes of identifying protective noise levels, annoyance is quantified by using the percentage of people who are annoyed by noise. This is felt to be the best estimate of the average general adverse response of people, and in turn, is viewed as reflecting activity interference and the overall desire for quiet." More recently, the U.S. Federal Interagency Committee on Noise (FICON, 1992) has reiterated that "Annoyance is a summary measure of the general adverse reaction of people to noise that generates speech interference ... or sleep disturbance or simply interferes with the desire for a tranquil environment." FICON (1992) further asserts that "The dose-effect relationship, as represented by DNL (the Day-Night Average Sound Level) and 'Percent Highly Annoyed' (%HA), remains the best available approach for analysing overall health and welfare impacts for the vast majority of transportation noise analysis situations."
The formal position of the U.S. regulatory acoustic establishment for the last two decades has therefore been (1) that noise-induced annoyance is the primary effect of environmental noise exposure in residential settings, and (2) that analysis of environmental noise effects is best accomplished by means of a descriptive function based on a particular noise metric. In other words, DNL (a form of long-term, timeweighted average sound level) is preferred as a descriptor of environmental noise not because it is easily defined, nor readily manipulated, nor optimal for some other purpose, but simply because it is held to be "the best available" predictor of annoyance. If a demonstrably superior predictor of the prevalence of annoyance were available, it would presumably be embraced by regulatory agencies.
When a regulatory agency identifies a particular value of DNL as an enforceable threshold of effect, the agency is effectively asserting its belief that the corresponding prevalence of annoyance represents an acceptable balance of the competing societal interests in regulation. Thus, when a group of U.S. federal agencies constructs a particular dosage-response relationship (FICON, 1992) to justify selection of Ldn = 65 dB as a regulatory threshold, it is asserting that it is acceptable in the opinion of the agencies for 12.3% of the exposed residential population to be highly annoyed by aircraft noise. This is an expressly non-technical opinion.
Rationale for regulation based on dosageresponse analyses
The standard approach to assessing community response to aircraft noise is an expedient one for which no theoretical rationale is offered or even believed necessary. The dosage-response approach evolved rapidly following publication of Schultz's (1978) meta-analysis. This seminal work "synthesised" (in Schultz's terms) an arbitrary fitting function to summarise the observed relationship between paired field observations of noise exposure and the prevalence of annoyance. Subsequent refinements of Schultz's work (e.g., those of Fidell et al., 1991, FICON, 1992, and Miedema and Vos, 1998), based on additional information and somewhat less arbitrary fitting functions, are generally viewed as reinforcing Schultz's pragmatic approach.
Schultz's approach to quantifying community response to environmental noise was initially greeted with scepticism and controversy. His approach was so convenient and such an obvious improvement over less systematic approaches, however, that it rapidly gained acceptance. A number of factors have contributed to the popularity of atheoretical dosage-response functions for predicting community response to environmental noise. These include a strong engineering bias in psychoacoustic tradition; the seeming improvements in accuracy, precision, and objectivity over prior assessment methods; the apparent lack of systematic alternatives; the perennial imbalance between readily estimated costs of noise control and its much less tangible benefits in residential settings; and irresistible pressures for expedient rather than fully considered social policy decisions.
It has long been recognised that individual and community reactions to aircraft and other environmental noise exposure are not fully determined by acoustic variables (cf. Rosenblith and Stevens, 1953). Efforts since those of Schultz (1978) to derive a dosage-response relationship based on use of DNL as a predictor variable still fail to account for a good deal of the variance in the relationship between cumulative noise exposure and the prevalence of a consequential degree of annoyance in communities.
Normative approach to modelling annoyance
A decade after Schultz's dosage-response relationship was published, Fidell, Schultz and Green (1988) noted the inherent inability of any exclusively acoustic variable to serve as a fully satisfactory predictor of self-reported annoyance. Fidell et al. (1988) identified response bias - in psychophysical decision theory (Swets, 1964), the willingness to respond independently from any signal-related information - as an inevitable confounding factor in self-reports of annoyance. Fidell et al. (1988) suggested a systematic approach to identifying the separate influences of acoustic and nonacoustic factors on self-reported annoyance. Reasoning from first principles rather than from curve-fitting exercises, they suggested consideration of a one-parameter, sigmoidal function as a dosage-response relationship relating cumulative noise exposure to the prevalence of annoyance in a community.
Green and Fidell (1991) subsequently elaborated this approach, documented the additional variance accounted for by taking response bias into formal consideration, and estimated the errors of prediction associated with both acoustic measurement and social survey data. More recently, Baird, Harder and Preis (1997) have proposed another theory-based approach to modelling the prevalence of noise-induced annoyance. The model of Baird et al. (1997) draws attention to constraints on interpretations of self-reports of annoyance associated with the nature of response scales provided to survey respondents.
Relative precision of estimation of exposure and annoyance
Regulation must be both simple and understandable to be effective. Aircraft noise exposure (particularly as expressed in units of DNL) lends itself to regulatory use in part because it is a simple and well defined concept that can be conveniently manipulated and graphically expressed as intuitively appealing,
source-based emission contours 4 . The ease with which aircraft noise exposure can be modelled and represented graphically, in conjunction with the logarithmic nature of decibel notation, lead to the widespread mis-impression that aircraft noise exposure is routinely measured with accuracy and precision greater than is possible in measurement of the prevalence of annoyance.
In fact, the reverse is closer to the truth: on comparable scales, errors of estimation of aircraft noise exposure are often greater than errors of estimation of the prevalence of annoyance. For example, the California Department of Transportation tolerates ±1.5 dB of uncertainty (or equivalently, a range of 3 dB, or a difference of 100%) in the labelling of noise exposure contours for its regulatory purposes. A range of three decibels in aircraft noise exposure is not merely a difference of three measurement units, but rather a net factor of two among any of the variables that affect time-weighted average noise levels: twice as many airplanes in the sky, or half the overflight duration, or twice the number of engines per aircraft, etc.
In contrast, confidence intervals on estimates of population proportions highly annoyed by aircraft noise developed from representative random samples of a few hundred survey respondents are generally on the order of ± 5% or less. If the goal of aircraft regulation were to limit the prevalence of consequential noise- induced annoyance to, say, a third of the population, as much as two-thirds of the population could be highly annoyed if the same tolerance were accorded to errors of measurement of annoyance (expressed on a linear scale) as to errors of measurement of noise exposure (expressed on a logarithmic scale).
Accounting for variability in predicted prevalence of annoyance
Notwithstanding the precision with which the prevalence of annoyance can be measured, variability in the observed prevalence of annoyance among communities with the same aircraft noise exposure is greater than implied by dosage-response analysis with a tolerance of ±1.5 dB for estimated noise exposure levels. [Table - 1] shows values of the prevalence of high annoyance predicted by FICON's (1992) dosageresponse relationship at noise exposure values 1.5 dB lesser and greater than nominal values. Thus, for instance, the range of predicted annoyance prevalence rates for a nominal value of 65 dB extends from that associated with a DNL value of 63.5 to that associated with a DNL value of 66.5 dB. Because the slope of FICON's predictive function is only about 1 - 2% highly annoyed per decibel in the range of regulatory interest, the tabled range of predicted percentages of the population highly annoyed is not very great.
[Table - 2] shows a range of ±1 about the means of the measured prevalence of high annoyance within 5 dB-wide noise exposure intervals in the aircraft noise data compiled by Fidell et al. (1991), augmented by data from several studies
conducted thereafter  . Each 5 dB noise exposure interval is centred on a multiple of 5 dB in DNL, so that (for instance) the interval shown at 65 dB encompasses the range from 62.5 dB Ldn 67.5 dB. [Figure - 1] presents the same information graphically, along with the dosage-response relationship recommended by FICON (1992) for all transportation noise.
It is apparent from the information summarised in [Table - 2] and [Figure - 1] that the FICON (1992) dosage-response relationship underestimates the prevalence of annoyance associated with exposure to aircraft noise in the range of exposure levels relevant to most regulatory purposes  . [Table - 3] summarises information similar to that contained in [Table - 1] for the predictive function of Green and Fidell, rather than FICON's. The slope of the Green and Fidell function is that of the rate of growth of effective loudness, while the D* parameter translates the function along the abscissa. A D* value is tantamount to a DNL threshold above which people describe themselves as highly annoyed for reasons independent from noise exposure per se. The value D* = 69.4 is the mean for the set of observations summarised in [Table - 2] and [Figure - 1].
[Table - 4] presents analogous information to that contained in [Table - 2]. Whereas the standard deviation units in [Table - 2] are those of the available field data, the standard deviation units in [Table - 4] are those of the response bias parameter, D*. Note that the dispersal of the distribution of response bias among different communities is not very different from the dispersal of the distribution of observed annoyance prevalence rates in the various social surveys. This suggests that response bias exerts an influence on the prevalence of high annoyance comparable in magnitude to that of noise exposure per se.
The fact that acoustic and nonacoustic factors exert influences of similar magnitude on the observed prevalence of annoyance in communities indicates that aircraft noise regulation which does not systematically consider nonacoustic factors does not address a substantial part of the rationale supporting imposition of regulation in the first place. It also implies that one size does not fit all in the regulation of aircraft noise for purposes of limiting the prevalence of annoyance, and that different regulatory levels may be appropriate at different airports.
Given the current level of development of theoretical understandings of the origins of noise-induced annoyance, response bias can be determined at present only empirically (via social survey). Thus, if necessary, direct measurement of response bias can permit systematic assessment of the degree to which aircraft noise may be under- or over-regulated in particular communities. The difference between the observed D* value for a particular community and the grand average D* for all communities yields a direct index, interpretable in units of decibels, of the degree to which noise regulation under- or over-protects residential populations from aircraft noise-induced annoyance.
For example, for entirely nonacoustic reasons, some communities might tolerate 3 dB more or less noise exposure than others for the same apparent prevalence of annoyance. Explicit consideration of response bias in the regulatory framework could in the former instance justify a doubling of aircraft noise exposure, while requiring a halving of noise exposure in the latter instance.
| References|| |
|1.||Ades, H.W., Davis, H., Eldredge, D.H., von Gierke, H.E., Halstead, W.C., Hardy, J.D., Miles, W.R., Neff, W.D., Rudnick, I., Ward, A.A., and Warren, D.R. (1953). "BENOX Report: An exploratory study of the biological effects of noise," ONR Project NR 144079, Contract N6ori-020, Task Order 44, The University of Chicago. |
|2.||Baird, J.C., Harder, K., and Preis, A. (1997). "Annoyance and community noise: Psychophysical model of doseresponse relationships," J. Environ. Psych., 17, 333-343. |
|3.||FICON (1992). "Federal Agency Review of Selected Airport Noise Analysis Issues," Washington, D.C. |
|4.||Fidell, S., Barber, D., and Schultz, T. J. (1991). "Updating a Dosage-Effect Relationship for the Prevalence of NoiseRelated Annoyance," J. Acoust. Soc. Am. |
|5.|| Fidell, S., Schultz, T. J., and Green, D. M. (1988). "A Theoretical Interpretation of the Prevalence Rate of NoiseInduced Annoyance in Residential Populations," J. Acoust. Soc. Am., 84(6), 2109-2113. |
|6.||Green, D. M. and Fidell, S. (1991). "Variability in the Criterion for Reporting Annoyance in Community Noise Surveys,", J. Acoust. Soc. Am., 89(1), 234-243. |
|7.||Kryter, K.D., (1982). "Community annoyance from aircraft and ground vehicle noise", J. Acoust. Soc. Am., 72(4), 1222-1242. |
|8.||Miedema, H., and Vos, H. (1998). "Exposure-response relationships for transportation noise," J. Acoust. Soc. Am., 104(6), 3432-3445. |
|9.||Pearsons, K., Tabachnick, B., Barber, D., and Fidell, S. (1995). "Predicting Noise-Induced Sleep Disturbance," J. Acoust. Soc. Am., 97 (1), 331-338. |
|10.||Rosenblith, W. A., Stevens, K. N., and the Staff of Bolt Beranek and Newman Inc. (1953). "Handbook of Acoustic Noise Control, Vol. 2: Noise and Man," WADC TR52204, Wright Air Development Center, Wright-Patterson Air Force Base, Ohio. |
|11.||Swets, J. A. (1964). "Signal Detection Theory and Psychophysics," John Wiley and Sons, New York. |
|12.||United States Environmental Protection Agency, Office of Noise Abatement and Control (1978). "Condensed Version of Levels Document," EPA 550/9-79-100. |
BBN Technologies, A Unit of GTE Internetworking, 21128 Vanowen Street, Canoga Park, California 91303
Source of Support: None, Conflict of Interest: None
[Figure - 1]
[Table - 1], [Table - 2], [Table - 3], [Table - 4]