This article appeared in a journal published by Elsevier. The attached copy is
furnished to the author for internal non-commercial research and educational
use, including for instruction at the author’s institution and sharing with
colleagues.
Other uses, including reproduction and distribution, or selling or licensing
copies, or posting to personal, institutional or third party websites are
prohibited.
In most cases authors are permitted to post their version of the article (e.g. in
Word or Tex form) to their personal website or institutional repository. Authors
requiring further information regarding Elsevier’s archiving and manuscript
policies are encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Applied Acoustics 73 (2012) 1–11
Contents lists available at ScienceDirect
Applied Acoustics
journal homepage: www.elsevier.com/locate/apacoust
Aircraft noise-monitoring according to ISO 20906: Evaluation of uncertainty
derived from the human factors affecting event detection
C. Asensio a,⇑, M. Ausejo a, K. Jambrosic b, J. Kang c, G. Moschioni d, R. Pagán a, I. Pavón a, J. Romeu e,
J.A. Trujillo a, G. Vigano f, M. Ruiz a, M. Recuero a
a
Universidad Politécnica de Madrid (CAEND), Spain
University of Zagreb, Croatia
University of Sheffield, United Kingdom
d
Politecnico di Milano, Italy
e
Technical University of Catalonia, Spain
f
Eco Progetti, Italy
b
c
a r t i c l e
i n f o
Article history:
Received 10 December 2010
Received in revised form 20 May 2011
Accepted 27 May 2011
Available online 20 July 2011
Keywords:
Aircraft noise monitoring
Monitoring
Uncertainty
E-comparisons
a b s t r a c t
One of the most important issues in aircraft noise monitoring systems is the correct detection and marking of aircraft sound events through their measurement profiles, as this influences the reported results. In
the recent ISO 20906 (unattended monitoring of aircraft sound in the vicinity of airports) this marking
task is split into: detection from the sound level time history, classification of probable aircraft sound
events, and the concluding identification of aircraft sound events through non-acoustic features.
An experiment was designed to evaluate the factors that influence the marking tasks and quantify their
contribution to the uncertainty of the reported monitoring results for some specific cases. Several noise
time histories, recorded in three different locations affected by flyover noise, were analyzed by practitioners selected according to three different expertise levels. The analysis was carried out considering three
types of complementary information: noise recordings, list of aircraft events and no information at all.
Five European universities and over 60 participants were involved in this experiment.
The results showed that there were no significant differences in the results derived from factors such as
the participant’s institution or the expertise of the practitioners. Nonetheless, other factors, like the noise
event dynamic range or the type of help used for marking, have a statistically significant influence on the
marking tasks. They cause an increase of the uncertainty of the reported monitoring and can lead to
changes in the overall results.
The experiment showed that, even when there are no classification and identification errors, the detection stage causes uncertainty in the results. The standard uncertainty for detection ranges from 0.3 dB for
those acoustic environments where aircraft are clearly detectable to almost 2 dB in more difficult
environments.
Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction
Noise monitoring is one of the most important tools in noise
management [1–5]. Its use has spread, especially for noise impact
assessment near airports, as noise is an extremely important issue
for airports and their surrounding communities [6–10]. Although
noise measurement instruments are becoming more advanced
day-by-day allowing an almost complete automation of measurements, the practitioner still has a crucial role in the process. This
leads to the final results having a relevant dependence on the
human factors deriving from the subjective interpretation of the
⇑ Corresponding author. Tel.: +34 915618806x304.
E-mail address: cesar.asensio@caend.upm-csic.es (C. Asensio).
0003-682X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apacoust.2011.05.013
standards, regulations or data themselves. In the case of airport
noise-measurements, the presence of a quite detailed structure of
standards does not seem to guarantee a user-independent interpretation of the phenomena and definitely not of the results.
International and other major airports usually face this task
using a fully integrated noise monitoring system which consists
of several permanent monitors installed in strategic locations
around the airport [11,12]. Exceptionally, it is necessary to integrate such measurements with specific short-term measurements
lasting a few days at certain locations.
Most of the factors that can affect the uncertainty of the results
are dealt with in ISO 20906 [1]: measuring instrumentation, residual sound, emission at the source, ground effect, etc. The standard
Author's personal copy
2
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
Table 1
Influencing factors.
Factor
Influence
Acoustic environment
Manual/automatic marking
Automatic detection parameters: threshold,
duration, pre-trigger, post-trigger
Radar/technician notes versus audio recordings
Institution
How the acoustic environment influences the detection of an aircraft event, masking it totally or partially
How the selectivity of human detection can be compared to automatic detection
How the choice of parameters for automatic detection changes the results
Experience of practitioners
The specific influence of aiding factors, compared to the efforts and duration of the analysis
How the different cultural approaches of nominally similar institutions can change the interpretation of
phenomena (for instance, the presence of local regulation could bias the interpretation of the standard)
How the experience and technical and scientific education influence the interpretation of the phenomena
also describes some of the possibilities for minimizing or avoiding
the influence of these factors.
The main objective of this investigation consists in evaluating
and quantifying the influence of the data processing on the reported results. This influence seems to be mainly present in the
detection, classification and identification tasks. Keeping apart
the contribution to uncertainty of misidentification (false positives,
false negatives), the authors have set the focus on the influencing
factors summarized in Table 1. These data will be very useful for
quantifying the uncertainty in these kinds of measurements, by
providing tools to evaluate the compared weight of each factor,
and providing keys for making decisions on how to implement a
full cost-effective monitoring system in an airport. The quantification of the standard uncertainty derived from each factor will complement the analysis of the uncertainty provided in ISO 20906.
2. Methodology
2.1. Standard measurement method
Fig. 1 shows the aircraft events identification schema as defined
in ISO 20906. The monitor records the A-weighted sound level for
every 1-s interval, in terms of equivalent sound level (LAeq,1s) or
sound pressure level with time weighting SLOW (LAS). The recorded
time history, Lp(t) is used for the detection of noise events.
The main sources of uncertainty in the results can be summarized as follows, depending on the type of measurement:
(1) In the case of annual equivalent noise level continuous measurements, the measurements, which have by default and
standards no selectivity, include factors related to atmospheric conditions, airport operating conditions and the
noise that comes from the environment surrounding the
microphone. Thus, the main contribution to the overall
uncertainty of single event level are the measurement
instrument, which has been widely studied [1,13], and the
background noise, both beyond the scope of this work.
(2) In the case measurements for incomplete periods, the results
are not usually considered as representative for the longterm condition due to the variability of the source along the
year. If for any reason they are to be considered for the assessment of the long-term condition, the analysis of the uncertainty must include other aspects, like airport operating
modes, number of operations, and atmospheric conditions.
These aspects have also been widely studied [1,5,14–16],
and they are also beyond the scope of this research.
Using the time history profile of the measurements, the detection process must extract a list of sound events based on acoustic
criteria. This is usually done by applying level and time thresholds
[17–19], so that if the measurements remain for a fixed time-interval over the threshold, it is considered as a sound event (see Fig. 2,
from ISO 20906).
Afterwards, the sound events are classified as ‘‘aircraft sound
events’’ or ‘‘non-aircraft sound events’’. This classification is based
primarily on acoustic knowledge applied to the measurements. It
can include several aspects such as duration of the sound event,
maximum level, and slopes. Finally, non-acoustic data (radar tracking, technicians, recordings, etc.) are used for the complete identification of the sound event.
Depending on the implementation, event detection and event
classification can be combined into one stage, as they both act on
the measurements. The identification process could be considered
as a validation that is made by radar tracking, by listening to audio
recordings, or other means.
Despite all these processes, uncertainty in the results will arise
for the following reasons:
– There are some aircraft sound events that cannot be detected
because of the high residual noise level (background noise).
– There is always the possibility of misclassification and/or
misidentification of the sound events.
– There is a rate of sound events that is not detected because of
the detection algorithm, or its configuration and customization by users (misdetection).
Fig. 1. Aircraft events identification scheme.
Author's personal copy
3
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
Fig. 2. Event detection method suggested by ISO 20906.
– Even in the case of a correct identification, the duration of the
sound events is also affected by the detection algorithm.
The Universidad Politécnica de Madrid (UPM) prepared a full
experiment oriented towards extracting information from the
factors previously described, regarding the influence of the human
factors on the event marking process and subsequently on the
reported noise monitoring results. After an agreement on the general terms, the participants from the rest of the research groups
and Universities started the collaboration by just participating in
the data processing. Many university students, some researchers,
professors and consultants with different acoustic-specific education and backgrounds participated in this experiment. The bases
of the experiment were not revealed to the participants until the
results had been gathered, thereby trying not to bias the conclusions of the experiment. After their participation, the leaders from
every institution were informed about the full process, so that their
comments for improving the quality of the study could be received.
Although this experiment cannot be strictly considered as an
interlaboratory comparison, it was designed following the main
basis of a proficiency test, described in the ISO/IEC Guide 43-1
[20]. A partial-process scheme was applied (as defined in ISO/IEC
Guide 43-1) to create some data transformation exercises where
the laboratories were furnished with sets of data and were
required to manipulate them to provide further information. The
coordinator of this experiment was the UPM.
The following sections provide an in-depth description of the
reference material, the test items preparation process and all the
issues related to this experiment.
Every participant filled in a form with information concerning
their experience and how it was related to environmental noise
measurements, and aircraft noise monitoring. According to this
information, the participants were classified in A, B and C classes,
where A meant no experience in environmental noise measurements, B meant some experience in environmental noise measurements, and C meant a lot of experience in environmental noise
monitoring.
The institutions involved in these tests were the following:
Politecnico di Milano (IT).
Universidad Politécnica de Madrid (ES).
University of Zagreb (HR).
Universidad Politécnica de Cataluña (ES).
University of Sheffield (UK).
In order to make the results anonymous, every institution was
assigned a number that was not known to the others.
Table 2 summarizes the number of participants per institution,
classified according to their experience.
2.3. Measurement locations
Three different noise environments in the proximity of Madrid
airport were tested, according to the following description:
Table 2
Participants and institutions.
Institution
2.2. The participants
Only two people from the UPM were involved in the primary
design of the experiment from the very beginning. The remaining
participants were considered as mere participants until they provided their own, independent, single results.
1
2
3
4
5
Total
Participants
Class A
Class B
Class C
9
38
1
0
0
48
4
3
0
0
0
7
3
2
1
1
1
7
Author's personal copy
4
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
– Aircraft sound events easily detectable (from measurements
and audio files).
– Aircraft sound events hard to detect in the measurement files,
but clearly audible.
– Aircraft sound events very difficult to detect, and the presence of other sound events.
Measurements and audio files had been previously recorded by
the UPM in the locations called MEJ, MOL and LOE, which are described below. For all the cases, the microphone was placed at a
height of 4 m above the ground. MEJ was selected in Mejorada
del Campo, at approximately 12 km south-east of the airport. The
road traffic residual noise level was lower than 50 dBA, and had
many neatly distinguishable sound events (Fig. 3). Most of them
were produced by aircraft. MOL was selected in El Molar, at
approximately 20 km north of the airport. At this location the
sound events are harder to detect from the measurements
(Fig. 4). LOE is located in Loeches, at approximately 15 km south
east of the airport. This location is quite apart from the flight
routes, so aircraft noise levels are lower than in MEJ. The higher
sound events detectable in the measurements profile were not
caused by aircraft (Fig. 5).
Fig. 6 shows a map (from Google Maps) where the reader can
locate the measurement points with respect to the airport.
2.4. The measurements and recordings
Three different test files were selected, according to the defined
scenarios and locations.
The measurements and recordings used in this project were
carefully selected from more than 200 h of environmental noise
recordings carried out in different sites around Madrid-Barajas
airport throughout 2009. All the measurements (LAeq,1s and LAF)
were made using a Brüel & Kjaer 2250 sound level meter. Its headphones output (audio signal from the preamplifier) was recorded.
In order to make the exercise easy for the participants while trying to keep their attention for the whole practice, it was decided to
use files no longer than 1 h. Every test set consisted of an audio file
and a measurement file. The duration of the files was 54 min and
22 s.
2.5. The design of the experiment
In order to obtain information on the detection techniques, it
was decided that the three files had to be analyzed three times
in different conditions:
– With no further information than measurement time history.
– With measurement time history and a list of the aircraft
sound events.
– With measurement time history and its related audio file.
For a better reliability of uncertainty assessments on the analyses, the same file, properly modified, was used for the three modes.
The participants were not aware that they were performing the
analyses on the same data: the original files were modified by pretending they had been taken in different locations, for different
dates and time intervals. Then, an offset was added to the files (different bias for LAeq and LAF). Finally, each file was split into small
parts and merged back in a different order (Fig. 7). Only the material summarized in Table 3 was supplied to the participants, with
no other help or indications other than ISO 20906.
Fig. 3. Measurement time history for location MEJ.
Author's personal copy
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
Fig. 4. Measurement time history for location MOL.
Fig. 5. Measurement time history for location LOE.
5
Author's personal copy
6
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
Fig. 6. Measurement locations.
The procedure of identification requires the detection of aircraft
sound-events from the measurement profiles. When the aircraft is
detected, it is necessary to mark the sound event, defining its
beginning and its end (see Fig. 2). After marking the events of
the whole file, all the sound events are used for the calculation of
Laircraft,D (equivalent noise level for the full reference period D).
The measurement files were supplied in a Brüel & Kjaer Evaluator Type 7820 format, in order to make the viewing and marking
processes easier, and every participant had to provide the processed files and fill in the form shown in Table 4.
3. Results
3.1. Descriptive statistics
The information supplied by the participants was checked and
put together, and then StatGraphics [21] was used for the statisti-
cal analysis. Most of the data in the form contain redundant
information:
– The duration of the measurements (Column A) and the overall
equivalent noise level for the full reference period (Column B)
do not depend on the practitioner operation.
– The duration of aircraft noise events (Column E) and the their
equivalent noise level referred to this duration (Column F) are
combined to show the practitioner operation in column D,
which is the parameter considered in this research (Laircraft,D).
Fig. 8 shows a box plot for Laircraft,D, showing the data obtained
for each of the nine files. After biasing the files back, the results
from files 1, 4 and 7 concentrate around a very similar mean value,
as the three files were obtained by just modifying the original file.
The same thing can be noticed concerning files 2, 5 and 8, and with
files 3, 6 and 9.
Author's personal copy
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
Fig. 7. Modifications applied to original measurement files.
7
Author's personal copy
8
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
Table 3
Files used for the comparison.
Analyzed file
Location
Material supplied to practitioners
File
File
File
File
File
File
File
File
File
LOE
MOL
MEJ
LOE
MOL
MEJ
LOE
MOL
MEJ
X
X
X
X
X
X
X
X
X
Measurement files
9
8
7
6
5
4
3
2
1
Comments
List of aircraft
Audio recordings
X
X
X
X
X
X
Original measurements and recording
Original measurements and recording
Original measurements and recording
Variation of file 9
Variation of file 8
Variation of file 7
Variation of file 9
Variation of file 8
Variation of file 7
Table 4
Form submitted by participants.
File
Full reference period
D Duration of
measurements
(s) (Column A)
Leq,D A-weighted Overall
equivalent noise level
(dB) (Column B)
Duration of aircraft events
Lresidual,D A-weighted
Residual equivalent noise
level (dB) (Column C)
Laircraft,D A-weighted
equivalent noise level, only
aircraft (dB) (Column D)
Daircraft Aircraft
events duration
(s) (Column E)
Laircraft,Daircraft A-weighted
equivalent noise level, only
aircraft (dB) (Column F)
1
2
3
4
5
6
7
8
9
errorij ¼ Laircraft;ij TV j
Fig. 8. Distribution of observed Laircraft,D for every file.
In order to check the statistical significance of the factors, and
for the posterior estimation of the uncertainty, we used the following model (based on ISO 20906):
Laircraft;ij ¼ TV j þ dsim þ dresidual þ dident þ ddetect
ð2Þ
where, errorij is the difference between the true value for the environment j(TVj), and the reported value i in that environment
j(Laircraft,ij).
By convention, the true value for each acoustic environment
was referenced by the mean value reported by the acoustic experts
when they used marking help, recordings and notes (Fig. 9).
Afterwards, it was checked if the new variable, error, followed a
Normal distribution, so that a parametric approach could be applied for the analysis. The probability distribution of the residuals
was analyzed using the Chi-square and the Kolmogorov–Smirnoff
tests, and a Normal Probability Plot, It was evidenced that data
were not distributed according to a Gaussian distribution, so a
non-parametric approach had to be used. Fig. 10 shows the Normal
probability plot for the residuals.
By applying the non-parametric Kruskal–Wallis test [22,23], the
influence of several factors on the data distribution was studied.
ð1Þ
where TVj is the true value at that location, dsim is a quantity to allow
for any uncertainty in the measuring instrumentation, dresidual
allows for any uncertainty due to the influences of residual sound,
dident considers the influences from the identification and classification tasks, and ddetect stands to consider the variability of the results
derived from the detection task.
According to the design of the experiment there is no variability
in data derived from the instrument or the residual noise, and also
the identification component can be neglected (especially for
marking method 2). Then, the model can be widely simplified,
and the data reported by the participants can be transformed into
error terms as follows:
Fig. 9. Distribution of Laircraft,D observed by experts, for every acoustic environment.
Author's personal copy
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
9
tions. . . had no statistical influence on the results. So it was
possible to avoid using this factor for the rest of the study.
However, it should be borne in mind that all the participants
have a common engineering basis despite very different personal
expertise.
3.3. The expertise factor
For each factor, the test was applied and it was attempted to confirm or reject the null hypotheses: ‘‘Ho: All the samples come from
the same statistical distribution’’.
Every participant was classified according to their experience
regarding acoustic measurements, environmental noise assessing
and airport noise. Three categories were established at a first stage:
A for students, C for experts in environmental acoustic and noise
assessment, and B for those people having some experience in
acoustics.
The group of non-experts showed a higher percentage of
outliers, and some other precautions had to be considered. But,
concerning the marking tasks, the statistical analysis showed that
there are no significant differences in the results derived from
the experience of the participant.
3.2. The institution factor
3.4. The marking method factor
The procedures were supplied using a website created for this
purpose, and no extra clarifications were made by the organizers
of the comparison exercise.
Only the two institutions that provided a representative
amount of 10 participants were included in the Kruskal–Wallis
test, which determined that there were no significant statistical
differences between the samples concerning the institution (for a
confidence level of 95%). As presumed, the quality or depth of
the explanations given by the different institutions to their participants, or other stimuli like the interpretation of local regula-
In this experiment, the detection and classification of probable
aircraft sound events (see Fig. 1) can be made manually or automatically using thresholds. Afterwards, three different possibilities
for the identification of aircraft sound events (marking) were used.
Marking method 3 involves using audio recordings, marking method 2 involves using a list of sound events (coming from field notes,
radar tracking. . .), and marking method 1 involves using only the
measurement profiles.
The statistical analysis showed that using additional information for the identification task has an influence on the results
Fig. 10. Normal probability plot for the residuals.
Fig. 11. Events range.
Author's personal copy
10
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
(mainly related to classification/identification). But, depending on
the acoustic environment, there are also differences derived from
the type of help used (recordings or list of sound events).
3.5. The event range factor
Apart from the affect of background noise on the measurement,
the detection task (not classification or identification) might be
quite influenced by the residual noise, compared to the maximum
level of the sound event. The event range determines the difference
between the LAeq,1s at the top, and the residual noise level (see
Fig. 11).
The statistical analysis showed that there are significant differences derived from this factor, so the contribution of the range factor has been studied separately for the three environments under
study.
In order to translate these coverage intervals into terms consistent with the approach used in [1,25], it was decided to estimate
the standard uncertainty (udetect) for each case. Assuming the most
conservative scenario (uncertainty might be slightly overestimated), the furthest limit of each coverage interval was used for
the estimation of the expanded uncertainty (Udetect), and a uniform
distribution was assumed so that the coverage factor could be considered K = 2, with infinite degrees of freedom.
U detect ¼ Maxfjlimit j; jlimitþ jg
ð3Þ
U detect U detect
¼
K
2
ð4Þ
udetect ¼
Table 5
Coverage intervals (95%) for error.
Marking method
3.6. Uncertainty calculations
After defining the main factors that contribute to the uncertainty of the measurements, this section describes a methodology
for quantifying the contribution to the uncertainty derived from
the influence of the human factors affecting the event detection
during the data processing of aircraft noise monitoring.
The first step consisted of the calculation of a true value for each
acoustic environment. I was used for transforming the reported
data into error terms, according to Eq. (2). Afterwards, data were
split into nine blocks according to the influence factors detected
(three marking methods, three acoustic environments). Then, the
probability distributions (Fig. 12) and the coverage intervals (Table
5) were estimated on the basis of [24], using Matlab.
1 (no help)
2 (list of events)
3 (recordings)
Dynamic range of events (dBA)
>20
10–20
<10
[0.5, +0.2]
[0.5, +0.2]
[0.4, +0.2]
[3.2, +0.8]
[2.9, +0.9]
[1.2, +0.4]
[4.3, +1.1]
[3.0, +3.1]
[3.6, +2.6]
Table 6
Standard uncertainty, udetect.
Marking method
1 (no help)
2 (list of events)
3 (recordings)
Fig. 12. Cumulative distribution functions for blocked error data.
Dynamic range of events (dBA)
>20
10–20
<10
0.3
0.3
0.2
1.6
1.5
0.6
2.2
1.6
1.8
Author's personal copy
C. Asensio et al. / Applied Acoustics 73 (2012) 1–11
The estimated standard uncertainty for each case is quantified
in Table 6. According to the assumptions described, this factor assumes a uniform distribution, infinite degrees of freedom and its
sensitivity coefficient equals 1 (cdetect = 1, from Eq. (1)).
Whatever method is used for marking, the standard uncertainty
linked to detection remains very low for the acoustic environment
with high event ranges (approximately 0.3 dB). Where aircraft
event ranges are high enough, the main contributions to uncertainty might be caused by the misclassification and misidentification of sound events.
11
– The process has shown that interlaboratory comparisons in
acoustics can be carried out through the use of fit-for-purpose
virtual measurements. The Internet makes it possible to dispense samples and collect results from laboratories all over
the world.
Acknowledgments
The authors would like to express their gratitude to all the participants in this survey.
4. Conclusions
References
An experiment was designed to analyze the influence of the
human factors affecting events detection on the reported aircraft
noise monitoring results according to ISO 20906. This paper
describes the methodology applied for the design of the experiment, the statistical analysis of the observations, the hypothesis
testing of the influencing factors and the quantification of the
uncertainty caused by these factors.
The outlier rate was higher among students, and although it
seemed that the variability of the results reported by the experts
group is lower, regarding the detection tasks, no statistically significant difference on the results derived from the expertise or the
institution factors was appreciated. From the statistical analysis
of the results it was concluded that only the acoustic environment
and the method used for marking had a significant influence on the
detection task. Therefore, the standard uncertainty was quantified
for nine single cases, matching the combination of three marking
methods and three acoustic environments.
The experiment was designed to minimize the influence of classification and identification tasks as much as possible, so that the
influence of the detection task could be analyzed independently.
Accordingly, marking method 2 reflects the situation of fully unattended monitoring (no recordings) and preserves the results from
any interference derived from the classification and identification
tasks. In this case, the coverage intervals are larger as the acoustic
environment becomes more complicated (reaching 1.6 dB for environment 3). This is the general trend for all the marking methods.
Whatever method is used for marking, the standard uncertainty
linked to detection remains very low for the acoustic environment
with high event ranges. This component of uncertainty, applied for
cumulative sound events in a reference interval, is much lower
(approximately 0.3 dB) than other contributions. For instance, the
contribution derived from the measurement instrument ranges
around 0.8 dB for every single event (referred as uslm in ISO
20906). Where aircraft event ranges are high enough, the main
contributions to uncertainty might be caused by the misclassification and misidentification of sound events. Under these circumstances, radar tracking or other identification tools [26,27] are
needed to guarantee an overall quality of monitoring results.
It must be noted, that for Environment 2 (range or 10 to 20 dB)
the reported variability of results is much lower for marking method 3 (recordings) than for marking method 2. There might be several causes to explain this: the detection of overlapping sounds
through the recordings, the use of linear charts (versus the logarithmic scale of measurement profiles),. . .
Besides the direct application of these results to the quantification of the overall monitoring uncertainty quantification, this
research has set the basis for future research lines:
– The reported results have provided a set of ‘‘calibrated’’ recordings and measurements that can be used for testing, training or
fine-tuning automatic noise detection units.
[1] ISO, ISO 20906:2009. Acoustics – unattended monitoring of aircraft sound in
the vicinity of airports; 2009.
[2] Bekebrede G, Hagenberg THM. Design of a flight track and aircraft noise
monitoring system. In: Proceedings of the 14th international council of the
aeronautical sciences 1984;2:1096–105.
[3] Branch MC, Man SG, Weber C. Monitoring community noise. J Am Plan Assoc
1974;40:266–73.
[4] European Parliament. Directive 2002/49/EC of the European Parliament and of
the council of 25 June, 2002 relating to the assessment and management of
environmental noise; 2002.
[5] IMAGINE WP2 partners. Determination of Lden and Lnight using
measurements. IMA32TR-040510-SP08; 2006.
[6] Black DA, Black JA, Issarayangyun T, Samuels SE. Aircraft noise exposure and
resident’s stress and hypertension: a public health perspective for airport
environmental management. J Air Trans Manage 2007;13:264–76.
[7] Morrell P, CH- Lu. Aircraft noise social cost and charge mechanisms – a case
study of Amsterdam Airport Schiphol. Transport Res Part D: Transport Environ
2000;5:305–20.
[8] Suau-Sanchez P, Pallares-Barbera M, Paül V. Incorporating annoyance in
airport environmental policy: noise, societal response and community
participation. J Transp Geogr 2011;19:275–84.
[9] Fidell S, Pearsons K, Tabachnick BG, Howe R. Effects on sleep disturbance of
changes in aircraft noise near three airports. J Acoust Soc Am
2000;107:2535–47.
[10] Franssen EAM, van Wiechen CMAG, Nagelkerke NJD, Lebret E. Aircraft noise
around a large international airport and its impact on general health and
medication use. Occup Environ Med 2004;61:405–13.
[11] Amsterdam Airport Schipol. Schipol Nomos website; 2010.
[12] Aena. Emisionsacusticas – Aeropuerto de Madrid-Barajas; 2010.
[13] Payne R. Uncertainties associated with the use of a sound level meter. NPL.
DQL-AC 002; 2004.
[14] De Muer T, Botteldooren D. Methods for quantifying the uncertainty in noise
mapping. In: Symposium: managing uncertainty in noise measurement and
prediction, Le Mans, France; 2005.
[15] Alberola Asensio J, Mendoza López J, Bullmore AJ, Flindell IH. Noise mapping:
uncertainties. Forum Acusticum, Sevilla, Spain; 2002.
[16] Farrelly FA, Brambilla G. Determination of uncertainty in environmental noise
measurements by bootstrap method. J Sound Vib 2003;268:167–75.
[17] Adams K. Aircraft noise events – the cornerstone of monitoring. In:
Proceedings of Acoustics 2004; 2004.
[18] Adams K. Aircraft noise event detection – the threshold problem. Internoise
2004;182.
[19] Jones David M, Matheson-Jones Eric H, Spillman Ronald R. A review of current
long-term environmental sound level measurement technologies. In:
Proceedings of the Alberta Energy And Utilities board conference; 2007.
[20] ISO, ISO/IEC Guide 43-1: 1997. Proficiency testing by interlaboratory
comparisons – Part 1: Development and operation of proficiency testing
schemes; 1997.
[21] Statistical Graphics Corp. Statgraphics plus, 5.1; 2000.
[22] Fisk DJ. Statistical sampling in community noise measurement. J Sound Vib
1973;30:221–36.
[23] Marques de Sá JP. Applied statistics using SPSS, Statistica, Matlab and R,
Springer; 2007.
[24] International Organization for Standardization. Working Group 1 of the joint
committee for guides in metrology, evaluation of measurement data –
Supplement 1 to the ‘‘Guide to the expression of uncertainty in
measurement’’ – propagation of distributions using a Monte Carlo method,
ISO. Geneva, Switzerland; 2008.
[25] International Organization for Standardization. Working Group 1 of the joint
committee for guides in metrology, evaluation of measurementdata – guide to
the expression of uncertainty in measurement, ISO. Geneva, Switzerland;
2008.
[26] Asensio C, Ruiz M, Recuero M. Real-time aircraft noise likeness detector. Appl
Acoust 2010;71:539–45.
[27] Genescà M, Romeu J, Pàmies T, Sánchez A. Real time aircraft fly-over noise
discrimination. J Sound Vib 2009;323:112–29.