Docstoc

Deutsches Zentrum Luft und Raumfahrt DLR

Document Sample
Deutsches Zentrum Luft und Raumfahrt DLR Powered By Docstoc
					Deutsches Zentrum
für Luft- und Raumfahrt e.V.


Forschungsbericht 2004-07/E


Effects of Nocturnal Aircraft Noise
Volume 1
Executive Summary


M. Basner
H. Buess
D. Elmenhorst
A. Gerlich
N. Luks
H. Maaß
L. Mawet
E.-W. Müller
U. Müller
G. Plath
J. Quehl
A. Samel
M. Schulze
M. Vejvoda
J. Wenzel


Institute of Aerospace Medicine
Cologne


87 pages
25 figures
7 tables
30 references
                Effects of Nocturnal Aircraft Noise
                             Volume 1
                        Executive Summary




German Aerospace Center (DLR)
Institute of Aerospace Medicine
Flightphysiology Department
Linder Hoehe
51147 Cologne
Germany
flugphysiologie@dlr.de
www.dlr.de/flightphysiology



Cologne, July 2004



Director of the institute:        Authors:
Prof. Dr. med. R. Gerzer          M. Basner, H. Buess, D. Elmenhorst,
                                  A. Gerlich, N. Luks, H. Maaß,
                                  L. Mawet, E.-W. Müller, U. Müller,
                                  G. Plath, J. Quehl, A. Samel,
                                  M. Schulze, M. Vejvoda, J. Wenzel



Head of the department:
Dr. A. Samel
aircraft noise, sleep, EEG, actimetry, performance, concentration, stress, cortisol, adrenalin,
noradrenalin, annoyance, logistic regression

Mathias BASNER et al.
Institute of Aerospace Medicine of the German Aerospace Center (DLR), Cologne
Effects of Nocturnal Aircraft Noise (Volume 1): Executive Summary
DLR-Forschungsbericht 2004-07/E, 2004, 87 pages, 25 figures, 7 tables, 30 references

From 1999 until 2003, the DLR-Institute of Aerospace Medicine conducted extensive investigations
on human specific effects of nocturnal aircraft noise within the framework of the HGF/DLR-project
"Leiser Flugverkehr". 128 subjects were investigated for 13 consecutive nights in four representa-
tive laboratory studies. From 11 pm until 7 am between 4 and 128 aircraft noise events with maxi-
mum sound pressure levels between 45 and 80 dB(A) were played back in a realistic fashion. The
following electrophysiological variables were continuously sampled during the night: EEG, EOG,
EMG, EKG, respiratory movements, airflow, finger pulse amplitude, position and actimetry. Con-
centrations of the stress hormones adrenalin, noradrenalin and cortisol were analyzed in all night
urine samples. Subjective assessments of strain and annoyance were collected with standardized
questionnaires. Computer assisted performance tests were performed every evening and morning
by the subjects. 64 residents of Cologne-Bonn Airport were investigated in their own homes for
nine consecutive nights with methods identical to those in the laboratory. Here, sound pressure
levels outside and inside the bedroom (at the sleeper's ear) were sampled continuously. In total,
2,240 study nights were polysomnographically investigated. The simultaneous recording of electro-
physiological and acoustic data allowed for an event related analysis with a resolution of 125 ms.
The age of both male and female subjects was between 18 and 65 years. Subjects did not suffer
from intrinsic sleep disorders and had normal hearing thresholds. This volume gives an overview of
study design and methods and summarizes the most important findings. For more detailed descrip-
tions and analyses please refer to the research reports DLR-FB-2004-08/E to DLR-FB-2004-11/E.


Fluglärm, Schlaf, EEG, Aktometer, Leistung, Konzentration, Stress, Cortisol, Noradrenalin, Adrena-
lin, Belästigung, logistische Regression
                                                              (in englischer Sprache veröffentlicht)
Mathias BASNER et al.
Institut für Luft- und Raumfahrtmedizin des DLR, Köln
Nachtfluglärmwirkungen (Band 1): Zusammenfassung
DLR-Forschungsbericht 2004-07/E, 2004, 87 Seiten, 25 Bilder, 7 Tabellen, 30 Literaturstellen

Im Zeitraum von 1999 bis 2003 führte das DLR-Institut für Luft- und Raumfahrtmedizin im Rahmen
des HGF/DLR-Projekts "Leiser Flugverkehr" umfangreiche Untersuchungen zu humanspezifischen
Wirkungen nächtlichen Fluglärms durch. 128 Versuchspersonen wurden in vier repräsentativen La-
borstudien über 13 aufeinanderfolgende Nächte untersucht. Zwischen 23:00 und 07:00 Uhr wur-
den zwischen 4 und 128 Fluggeräusche mit Maximalpegeln zwischen 45 und 80 dB(A) über Laut-
sprecher realitätsnah eingespielt. Folgende elektrophysiologische Variablen wurden kontinuierlich
erfasst: EEG, EOG, EMG, EKG, Atmungsbewegungen, Atemfluss, Fingerpulsamplitude, Position und
Aktometrie. Die Stresshormone Adrenalin, Noradrenalin und Cortisol wurden im nächtlichen Sam-
melurin bestimmt. Die subjektive Einschätzung der Belastung und Belästigung wurde mit standardi-
sierten Fragebögen erfasst. Computergestützte Leistungstests wurden jeweils abends und morgens
durchgeführt. 64 Anrainer des Köln-Bonner Flughafens wurden mit identischer Technik in ihrer ge-
wohnten Umgebung in neun aufeinanderfolgenden Nächten untersucht, wobei der Schallpegel
außen und innen (am Ohr des Schläfers) kontinuierlich gemessen wurde. In den Labor- und Feld-
studien wurden somit insgesamt 2.240 Probandennächte polysomnografisch untersucht. Die simul-
tane Aufzeichnung von elektrophysiologischen und akustischen Parametern erlaubte eine ereignis-
korrelierte Auswertung mit einer Auflösung von 125 ms. Die Versuchspersonen beiderlei Ge-
schlechts waren zwischen 18 und 65 Jahre alt, altersentsprechend schlafgesund und normalhörend.
In diesem Band werden neben dem Studiendesign und der Methodik die wichtigsten Ergebnisse
zusammenfassend dargestellt. Eine umfassendere Darstellung erfolgt in den Forschungsberichten
DLR-FB-2004-08/D bis DLR-FB-2004-11/D.
List of Abbreviations

Abbreviation   Meaning

AGARD          Advisory Group for Aerospace Research and Development
AMSAN          isolation facility (Arbeitsmedizinische Simulationsanlage)
dB             decibel, physical unit of the sound pressure level
dB(A)          physical unit of the A-weighted sound pressure
DIN            Deutsches Institut für Normung e.V.
EBF            Strain and Recreation Questionnaire
               (Erholungs- und Belastungsfragebogen)
ECG            electrocardiogram
EEG            electroencephalogram
EMG            electromyogram
EOG            electrooculogram
FA             entry questionnaire
FAT            Fatigue check-list
FNL            Aircraft Noise Questionnaire
HGF            Helmholtz-Gemeinschaft deutscher Forschungszentren
kHz            kilohertz, physical unit of frequency
LAS            A-weighted sound pressure level
               measured with time-weighting "slow"
LAS,eq         A-weighted equivalent continuous sound level
               measured with time-weighting "slow"
LAS,eq_event   A-weighted equivalent continuous sound level for the event
               measured with time-weighting "slow"
LAS,max        A-weighted maximum sound pressure level
               measured with time-weighting "slow"
LRA            logistic regression analysis
MDBF           Multidimensional Mood Questionnaire
               (Mehrdimensionaler Befindlichkeitsfragebogen)
m              meter(s)
min            minute(s)
ms             millisecond(s)
MST            Memory Search Task
Abbreviation     Meaning

ng               nanogram(s)
NREM-sleep       sleep stages 1 to 4
REM-sleep        rapid-eye-movement sleep
sd               standard deviation
sec              second(s)
S1, S2, S3, S4   sleep stages S1, S2, S3, S4
SPT              sleep period time
SRT              Single Reaction Task
STRAIN           Study on Human Specific Response to Aircraft Noise
STRES            Standardized Tests for Research for Environmental Stressors
TST              total sleep time
UTT              Unstable Tracking Task
Index



1           Introduction ................................................................................. 1

2           Study objectives .......................................................................... 2

3           Study design and methods ......................................................... 4

    3.1     Study design................................................................................... 4

    3.2     Sampling of study subjects ............................................................. 6

    3.3     Description of the study sample ...................................................... 9

    3.4     Acoustics...................................................................................... 12

      3.4.1 Acoustics in the laboratory studies ............................................ 12

      3.4.2 Acoustics in the field studies ..................................................... 17

    3.5     Electrophysiological signals ........................................................... 18

    3.6     Biochemical methods.................................................................... 19

    3.7     Performance................................................................................. 20

    3.8     Psychology ................................................................................... 21

    3.9     Data analysis ................................................................................ 21

4           Acoustics – results and discussion of the data of the field
            studies ........................................................................................ 22

5           Sleep ........................................................................................... 27

    5.1     Influence of nocturnal aircraft noise on sleep ................................ 30

    5.2     Methods....................................................................................... 31

    5.3     Analysis of the influence of noise on sleep stage distribution ........ 32
    5.4     Event correlated analysis ............................................................... 38

      5.4.1 What is a suitable descriptor for noise induced sleep
               disturbances?............................................................................ 38

      5.4.2 Methods ................................................................................... 41

      5.4.3 Results of the laboratory studies................................................ 42

      5.4.4 Results of the field studies......................................................... 47

      5.4.5 Comparison of the results of laboratory and field studies .......... 53

    5.5     Awakening duration..................................................................... 55

    5.6     Falling asleep again ...................................................................... 57

    5.7     Importance of the results for the discussion of aircraft noise
            protection criteria ......................................................................... 59

    5.8     Transfer of the results to a German airport ................................... 61

6           Psychological effects ................................................................. 63

7           Stress hormones ........................................................................ 68

8           Performance............................................................................... 73

9           Summary .................................................................................... 78

10          Literature.................................................................................... 81
1           Introduction

            Alexander Samel



Within the framework of the HGF/DLR-Project “Leiser Flugverkehr“, the
German Aerospace Center (DLR) developed a catalogue for the improve-
ment of technical, operational and legal measures for the reduction of air-
craft noise. Within this project, the task of work package #1 "Effects of
Nocturnal Aircraft Noise on Humans" was to create in extensive studies a
solid basis for an improved understanding of the effects of nocturnal air-
craft noise on sleep.

An essential reason for this research activity is the lack of sufficiently large
studies, in which sleep was polysomnographically investigated. So far, this
deficiency only allows a very limited and unreliable assessment of aircraft
noise-induced sleeping disturbances and their acute effects. Although there
have been several proposals for limiting values, also considering principles
of preventive medicine, these values mainly rely on data from small primary
studies conducted in the past. Additionally, these studies partly draw con-
tradictory conclusions and also show considerable deficits in their methods,
execution, analysis and/or interpretation. Consequently, the proposals so
far are based on research results which were attained in small sample sizes
with different methods and therefore consistently lead to fierce discussions
amongst noise-effects researchers and other interested groups.

In this situation, between 1999 and 2003, the DLR conducted several stud-
ies in which identical methods with large sample sizes were used in order to
expand the scientific knowledge. As the investigations took place in the
laboratory as well as in the field, i.e. in the homes of those affected by air-
craft noise, a direct comparison between simulated and real conditions was
possible.




                                       1
2        Study objectives

         Alexander Samel



The most important objective of the project was to develop criteria for as-
sessing the effects of nocturnal aircraft noise on humans. A second very
important goal was the establishment of a huge database of acoustical,
physiological and psychological functions and properties derived from an as
much as possible representative sample survey of people who are affected
by nocturnal aircraft noise. By them, a broad scientific basis shall be pro-
vided for determined strategies of noise abatement by technical progress
and operational procedures, as well as for planning purposes.


                          increasing noise intensity




                  ?            ?
                                                           primary
                                                       during the night




             I         II             III
                                                        secondary
                                                       on the next day




 Figure 2.1: Development of assessment criteria derived from primary and secondary
 sleep disturbances.




                                        2
The development of criteria can be performed by the determination of
dose-effect relationships and threshold levels for primary and secondary re-
actions of different dimensions, if they are existent and are detectable in
proper laboratory or field experiments.

It is important whether dose-response relationships and thresholds are de-
tectable for primary or secondary sleep disturbances: primary disturbances
are those which occur as immediate reactions to aircraft noise during the
night (e.g. vegetative arousals, awakenings, elevated excretion of stress
hormones); secondary sleep disturbances are characterized as reactions visi-
ble only on the following day as a consequence of sleep disturbances ex-
perienced during the previous night (e.g. intensified fatigue, reduced per-
formance, enhanced annoyance). In the least noisy area I (see Figure 2.1), it
can be expected that noise will not lead to measurable reactions, and thus,
a threshold or dose-response effects will not be found. Above a certain
threshold, reactions during sleep will presumably occur – with inter-
individual variations of the threshold (primary reactions). These reactions
can probably be compensated and will not lead to alterations on the next
day (area II of Figure 2.1). It can be expected that any further increment of
nightly noise (event related or as a continuous noise) will produce primary
sleep disturbances that cannot be compensated and thus, will also cause
secondary reactions. Again, thresholds and dose-effect relations might dif-
fer inter-individually (area III). As secondary reactions may manifest over
long lasting periods of many years, it cannot be excluded that tertiary ef-
fects can occur finally (e.g. health impacts), as a cause among many others
according to a multi-factorial etiology.

The occurrence of dose-effect relations and thresholds, respectively, of pri-
mary and secondary sleep disturbances can be used to develop criteria for
the evaluation of noise effects on humans. Furthermore, it can also be as-
sumed, that the probability of health effects caused by noise (tertiary sleep
disturbances) will be extremely low, if appropriate primary and secondary
sleep disturbances do not occur or can be compensated.


                                       3
3            Study design and methods

             Mathias Basner



3.1          Study design


Data sampling for the study commenced in September 1999 and ended in
June 2003. Table 3.1 shows names, time periods and types of the different
study parts.

Name                 Study period                                Type of study

STRAIN I             September until November 1999               Laboratory

STRAIN II            May until July 2000                         Laboratory

STRAIN III           February until April 2001                   Laboratory

STRAIN V             September 2001 until May 2002               Field

STRAIN VI            May 2002 until November 2002                Field

STRAIN IV            March until June 2003                       Laboratory


 Table 3.1: Study periods of the different parts of the study STRAIN (STudies on human
 specific Response to Aircraft Noise).


In the laboratory studies, 128 subjects were investigated for 13 consecutive
nights, whereas in the field studies 64 volunteers were observed for nine
consecutive nights. For comparative reasons, both the laboratory and the
field studies commenced on a Monday evening.

In the laboratory studies, the simulation facility of the DLR-Institute of Aero-
space Medicine allowed for the simultaneous investigation of eight sub-
jects. The first of the 13 observation nights served as adaptation, the sec-
ond as baseline and nights 12 and 13 as recovery. All of these nights were
noise-free. A control group of 16 subjects was used to investigate the influ-


                                           4
ence of the laboratory situation on otherwise undisturbed sleep and there-
fore the subjects did not receive any noise at all. The experimental group
consisting of the remaining 112 subjects received between 4 and 128 air-
craft noise events (ANEs) per night with differing maximum sound pressure
levels (SPL) during nine consecutive nights (nights 3 to 11). Lights were
turned off at 11 pm and on again at 7 am, which allowed for a maximum
sleep period time of 8 hours. In total, 1072 nights containing aircraft noise
and 592 nights without aircraft noise (adaptation, baseline, recovery, con-
trol) were investigated.

In the field studies, the homes of residents living in the vicinity of Co-
logne/Bonn Airport were selected in a way that the exposure to aircraft
noise was high on one hand, but the exposure to other kinds of traffic
noise, especially road traffic noise, was as low as possible on the other
hand. Because flight paths change due to alternating weather conditions,
and the frequency of planes taking off and landing depends on the week-
day, the study period consisted of nine consecutive nights, including week-
ends. Beside noise levels outside and inside the bedroom, exactly the same
data as in the laboratory studies (see below) were collected in the field. In
contrast to the laboratory studies, subjects participating in the field studies
were allowed to individually choose sleep period times with the require-
ment that sleep period times included the time period between midnight
and 6 am. In total, 64 subjects were investigated in 576 nights during the
field studies.

In total, 2,240 subject nights were investigated in both laboratory and field
studies together. 20 volunteers participated in both, laboratory and field
studies.

The study protocol was approved by the ethics commission of the Medical
Association of the district North Rhine. Subjects were instructed according
to the Helsinki declaration, participated voluntarily and were free to discon-
tinue their participation at any time without explanation.



                                      5
Study subjects received an allowance amounting to          75,- (field) and     55,-
(laboratory) per observation night. Training of computer-assisted perform-
ance tests prior to the start of the study was reimbursed with up to          350,-.


3.2       Sampling of study subjects


Study subjects were sampled in a multi-stage selection process. Each appli-
cant had to pass the following steps of the selection process:

• Fill out and return entry questionnaire FA

• participate in a detailed presentation of the study and fill out the FPI-
   questionnaire [9]

• participate in the medical check-up (medical history, medical examina-
   tion, ECG, blood and urine samples, hearing threshold test)

• constant performance by the end of 40 training sessions of the com-
   puter-assisted performance tests (see chapter 8)

A detailed description of the multi-stage selection process can be found in
the report DLR-FB 2004-08/E: "Effects of Nocturnal Aircraft Noise Volume
2: Study Design and Methods, Acoustics".

The applicants had to fulfill certain criteria for study eligibility. The exclusion
of certain applicants was intended to increase the validity of the results for
the study sample (internal validity). Additionally, the study group was aimed
to be as representative as possible in order to be able to extend the findings
of the study to a larger population (external validity).

The most important eligibility criteria will be described and explained in de-
tail below. The selection of subjects was performed in such a way that rec-
ommendations based on the study results were in favor of airport residents
affected by nocturnal aircraft noise, i.e. the selection process guaranteed




                                        6
that the effects of nocturnal aircraft noise were rather overestimated than
underestimated.

The age of the study subjects was restricted to the range from 18 to 65
years. An investigation of children for nine or 13 consecutive nights was
ethically not passable. Applicants older than 65 would not have to be ex-
cluded inevitably. But with a limited number of study subjects a wider age
range would have caused a decrease in precision for single age categories
because of the smaller number of subjects per category. Aside from this, it
is not expected that the sensitivity to aircraft noise of subjects older than 65
increases suddenly and disproportionately, which is why trends found up to
the age of 65 can be extrapolated until new evidence is available.

Study subjects had to have a healthy sleep according to age. The goal of
the study was to investigate the influence of nocturnal aircraft noise on
sleep. If a subject suffers from an intrinsic sleep disorder, it is impossible to
differentiate whether secondary sleep disorders (tiredness, performance
decrements, etc.) observed on the next day are caused by aircraft noise or
by the intrinsic sleep disorder itself, especially as the severity of intrinsic
sleep disorders may fluctuate unsystematically over several observation
nights.

Frequent spontaneous awakenings caused by an intrinsic sleep disorder
lead to a systematic underestimation of the fraction of awakenings induced
by aircraft noise. Aircraft noise induced awakenings are defined as the
amount of awakenings solely attributable to aircraft noise. They are calcu-
lated as the difference of "awakenings occurring under the influence of air-
craft noise" minus "spontaneous awakenings" (see chapter 5). The sleep
pressure of patients with an intrinsic sleep disorder is frequently increased
because of the perpetual disruption of normal sleep structure. Thus, the
probability of reactions induced by aircraft noise my be reduced compared
to subjects with healthy sleep according to age. There is no doubt that traf-
fic noise represents an additional strain for patients with intrinsic sleep dis-



                                       7
orders. Nonetheless, recommendations based on samples containing these
patients would underestimate the influence of nocturnal aircraft noise on
sleep, be anti-conservative and not in favor of airport residents affected by
nocturnal aircraft noise.

Loudly snoring participants were excluded from the laboratory studies, as
there was the possibility that the sleep of subjects in neighboring sleep cab-
ins was disturbed by these snoring sounds. In total, two participants had to
be exchanged for so called backups because of snoring and sleep related
breathing disorders in the laboratory studies. In order to prevent interfer-
ence of snoring sounds with aircraft noise events recorded in the field stud-
ies, sound level and noise events were recorded for one night prior to the
selection of participants for the field studies and loudly snoring applicants
were excluded.

There were some requirements according to the sleep habits of the appli-
cants. Applicants should not nap regularly during the day because daytime
naps were not allowed during the laboratory studies. Usual personal sleep
period times should at least be six and no longer than ten hours, because
otherwise habituation to the strict sleep period time from 11 pm until 7 am
in the laboratory studies would have been difficult. Regularly going to bed
before 9 pm or after 1 am as well as shift work during the night lead to ex-
clusion from the application process. Additionally, applicants should not
regularly need to go to the restroom for more than two times during the
night, because each visit to the restroom represents a potential interference
of the other subjects' sleep, aside from the disturbance of the subject's
own sleep.

All participants of the study needed to have normal hearing thresholds.
Depending on age the hearing loss on the weaker ear was restricted to
10% (18 to 33 years), 15% (34 to 49 years) and 20% (50 to 65 years).




                                      8
Applicants with arrhythmias were excluded because the analysis of pa-
rameters related to heart frequency or pulse amplitude would have been
too complicated or even impossible.

The consumption of drugs with depressant effects on the central nervous
system were not allowed, as these drugs may elevate awakening thresholds
and influence the results of the computer-assisted performance tests.

The Freiburger Persönlichkeitsinventar (FPI, see below) was used to test for
psychological eligibility, an important feature for group integrity during
the laboratory studies. The criteria were evaluated according to the progno-
sis of social compatibility as well as to the psychological and physical stabil-
ity. Applicants with the tendency to answer in terms of "social desirability"
were excluded, because otherwise interpretation of questionnaire results
would have been doubtful.

Applicants had to be able to fluently speak and write German. Contagious
diseases were not allowed. Applicants with alcohol or drug addictions were
excluded. As sleep cabins in the underground sleep facility are relatively
small, subjects with claustrophobia were excluded from the laboratory stud-
ies.

It is emphasized that other diseases than the ones described above did not
lead to an exclusion from the study, as long as those diseases did not inter-
fere with the experiment.


3.3        Description of the study sample


Both in the laboratory and in the field studies a few more women than men
were investigated (see Table 3.2), though there were more male applicants
(52,1%) in the laboratory and more female applicants (50,6%) in the field
studies.




                                       9
                                                   Gender

                                                   Male             Female

             Laboratory subjects
                                                   41.4%            58.6%
             (n=128)
             Field subjects
                                                   45.3%            54.7%
             (n=64)

 Table 3.2: Gender distribution in laboratory and field studies.


Table 3.3 shows the age distributions of participants of both laboratory and
field studies.

                 Age range (years)                                             Age

                 18-25        26-33   34-41        42-49    50-57      58-65   M (SD)

  Laboratory
  subjects       19.5%        27.3%   14.1%        14.8%    13.3%      10.9%   38 (13)
  (n=128)
  Field
  subjects       25.0%        14.1%   20.3%        23.4%    9.4%       7.8%    37 (13)
  (n=64)

 Table 3.3: Age distribution over six categories with a category width of eight years in
 laboratory and field studies (M = mean, SD = standard deviation).


Originally, a uniform age distribution for all categories was planned in order
to receive precise estimates for the whole age range investigated. For a uni-
form distribution each category should have consisted of about 17% of the
participants. Table 3.3 shows that the goal of uniformity was not reached
completely, yet sufficiently.

Aircraft noise annoyance represents an important control variable. There-
fore, it was accounted for in the selection process. It is emphasized at this
point that the subjective evaluation of aircraft noise annoyance must not be
equated with the actual physical exposure to aircraft noise. In fact, psycho-
logical research was able to show that only about one third of aircraft noise



                                              10
annoyance is caused by the actual aircraft noise load. Hence, it is possible
that someone feels strongly annoyed by aircraft noise although the actual
exposure to aircraft noise is low and vice versa.

Table 3.4 compares annoyance ratings of participants in laboratory and
field studies, applicants for laboratory studies STRAIN I to III, and a cross-
section representative for the German population in 2000 [18]. Both in the
last laboratory study and in the field studies subjective annoyance ratings
were collected only after the end of the study period with the FA-2 ques-
tionnaire. Therefore, data on annoyance ratings are only available for appli-
cants to STRAIN I to III.

                     Aircraft noise annoyance

                     not           little        medium   quite    very

Germany 2000
Ortscheid und Wen-   67.5%         17.7%         9.1%     3.7%     2.0%
de [18]
STRAIN I to III
Applicants           43.4%         33.2%         13.1%    7.5%     2.9%
(n=763)
Laboratory
Subjects             29.7%         38.3%         19.5%    8.6%     3.9%
(n=128)
Field
Subjects             3.1%          21.9%         39.1%    28.1%    7.8%
(n=64)

 Table 3.4: Aircraft noise annoyance – a comparison.


The cross-section of the German population is not uniformly distributed
over the five annoyance categories. One reason for this might be that many
people were interviewed in areas not influenced by aircraft noise. More
than two thirds of the interviewees did not feel annoyed by aircraft noise at
all. As many of the applicants for the laboratory studies STRAIN I to III were
living in the vicinity of Airport Cologne/Bonn, which is also situated near



                                            11
the German Aerospace Center (DLR) and the sleep lab, in this group only
43.4% did not feel annoyed by aircraft noise.

Taking into account the aircraft noise annoyance distribution of the appli-
cants, here, in contrast to age and gender, it was impossible to achieve a
uniform annoyance distribution in the subjects participating in the study:
Only 7.5% of the applicants felt quite annoyed, and only 2.9% felt very
annoyed. Due to the selection process in the laboratory studies, more par-
ticipants than applicants fell in the categories "medium", "quite" and
"very", at the expense of the categories "not" and "little". After all, two
thirds of the participants of the laboratory study felt annoyed by aircraft
noise to some extent. In the field studies, one third of the participants felt
"quite" or "very" annoyed. Here, only 3.1% (2 subjects) were not annoyed
by aircraft noise at all as opposed to 67.5% of the cross-section of the
German population.


3.4      Acoustics


The following sections describe how aircraft noise events were recorded for
and played back in the laboratory studies and how aircraft noise events
were sampled in the field studies.


3.4.1    Acoustics in the laboratory studies


During nights 3 to 11 between 4 and 128 noise events of starting or land-
ing planes with maximum sound pressure levels (SPL) from 50 to 80 dB(A)
were played back between 11 pm and 7 am. In the last laboratory study
STRAIN IV aircraft noise events (ANEs) with maximum SPLs of 45 dB(A) were
additionally played back. With a constant difference of 5 dB(A) between
events, 14 different aircraft noise events were applied in total.

The combinations of maximum SPL and number of ANEs per night that
were used during the laboratory studies is shown in Table 3.5. The combi-


                                      12
nations that the subjects received during a study period were randomly as-
signed. As there were more combinations than exposure nights per subject,
the study design may be described as an incomplete block cross-over de-
sign.

                                  Number of Noise Events starting     Number of Noise Events landing

                                  4     8     16    32    64    128   4    8     16    32    64    128

                             45                                                              32
 Maximum SPL LAS,max in dB




                             50               16    16    16    16               16    16    16    16
                             55   24    16    16    16    16    16    16   16    16    16    16    16
                             60   24    16    16    16    16          16   16    16    16    16
                             65   16    16    16    16    16          16   16    16    16    40
                             70   16    16    16    16                16   16    16    16
                             75   16    16    16                      16   16    16
                             80   16    16                            16   8

    Table 3.5: Combinations of maximum SPL at the sleeper's ear and number of aircraft
    noise events per night in the laboratory studies STRAIN I to IV (e.g. 24 subject nights
    with 4x60 dB(A) at the sleeper's ear). Frequencies other than 16 are highlighted in red
    color.


The combinations 4x50 dB(A) and 8x50 dB(A) were not used because rele-
vant reactions are not expected at this low level of exposure. On the other
hand, the combinations 128x60 up to 128x80 dB(A), 64x70 up to
64x80 dB(A), 32x75 dB(A), 32x80 dB(A) and 16x80 dB(A) were not played
back because such exposures are unrealistically high and might not have
been tolerated by the subjects, or might have caused subjects to discon-
tinue study participation ahead of schedule.

Each category was planned to consist of at least 16 subject nights, which
was accomplished for all categories but 8x80 dB(A) landing. The increased
number of nights with combinations 64x45 dB(A) landing and 64x65 dB(A)
landing resulted from a special sub-experiment, in which both combina-



                                                               13
tions were compared (see below). The number of nights with the combina-
tions 4x55 dB(A) und 4x60 dB(A) were increased because they were just be-
low the so called Jansen criterion, and thus should be covered more inten-
sively.

The ANEs played back during the night were recorded with class-1 sound
level meters (NC-10, Cortex Industries) in the vicinity of Düsseldorf Airport
with closed or tilted windows. The microphone was positioned near pillow
position or "at the sleeper's ear".

During a single study night always the same ANE was played back (e.g.
50 dB(A) starting only), i.e. there was no mixing of different ANE in one
single night. All eight subjects of one study period received the same noise
pattern, i.e. the same ANE was played back in all sleep cabins at the same
time. As sound insulation in the sleep cabins was not total, a temporal off-
set of playback of ANEs might have lead to the perception of ANEs from
neighboring sleep cabins.

From four to 128 ANEs were equidistantly played back between 11:15 pm
and 6:45 am. The distance between two ANEs was 120 minutes at four
events per night, 60 minutes at eight events per night, 30 minutes at 16
events per night, 15 minutes at 32 events per night, seven or eight minutes
at 64 events per night and 3 or 4 minutes at 128 events per night (see
Figure 3.1)

As the participants did not know of the equal distances between ANEs, an
anticipation of the time of occurrence of the next ANE was impossible.
Watches and alarm clocks were not allowed in the sleep cabins.




                                      14
 Number


   4x 20


   8x

        15

  16x


  32x
        10



  64x


  128x5

  64x
  real
         0
         39600   43200   46800   50400    54000     57600    61200    64800    68400
        23:00    0:00    1:00    2:00     3:00     4:00     5:00      6:00     7:00

                                          Time


 Figure 3.1: Playback pattern of ANEs depending on the number of ANEs per night. At
 the bottom a realistic pattern derived from the field studies, which was used in
 night 12 of STRAIN IV with maximum SPL of 65 dB(A), is shown.


Playback of ANEs was realized with an Acoustic Workstation CF85 (Cortex
Industries). Before each study period, every sleep cabin was acoustically
calibrated with class-1 sound level meters in order to guarantee realistic
playback of ANEs.

The SPL in each sleep cabin was recorded continuously during each study
night and allowed for the control of the correct playback of every ANE. Ad-
ditionally, it was possible to identify loudly snoring subjects.

At the beginning of the playback of each ANE a digital trigger signal was
put out and simultaneously sampled with the electrophysiological signals at
a sampling rate of 8 Hz. Thus, all signals were available on a single time
axis, which allowed for an event correlated analysis with a maximum resolu-
tion of 125 ms.


                                         15
The subjects were only informed that the first two study nights were noise-
free. They were otherwise blinded with respect to noise exposure, i.e. they
did not know when, how many and what kind of ANEs were played back
after the second night. In order to avoid subconscious manipulations, the
investigators were also blinded for the noise pattern of the specific night.
Only after the beginning of data sampling, i.e. after 11 pm, they were in-
formed about the noise pattern of the specific night, and thus were able to
monitor the correct playback of ANEs.

Altogether, 34,688 ANEs were played back in the laboratory studies. The
equivalent continuous sound level [DIN1] depending on the combinations
of maximum SPL and number of ANEs per night are shown in Table 3.6.
There was a constant background noise of about 30 dB(A) in the laboratory
studies caused by the air condition system.

                                 Number of Noise Events Starting         Number of Noise Events Landing

                                 4     8     16    32    64        128   4     8     16    32    64       128

                            45                                                                   24,5
                                                                                                 31,0
                            50               28.0 31.0 34.0 37.0                     22.7 25.7 28.7 31.7
                                             32.1 33.5 35.4 37.7                     30.7 31.3 32.3 33.8
Maximum SPL LAS,max in dB




                            55   25.1 28.1 31.1 34.1 37.1 40.1 21.8 24.7 27.7 30.7 33.7 36.7
                                 31.2 32.1 33.6 35.5 37.8 40.5 30.6 31.1 32.0 33.3 35.2 37.5
                            60   31.7 34.7 37.7 40.7 43.7                26.7 29.5 32.5 35.5 38.6
                                 33.9 35.9 38.3 41.0 43.9                31.7 32.8 34.4 36.6 39.1
                            65   36.7 39.7 42.7 45.7 48.6                31.7 34.7 37.7 40.7 43.7
                                 37.5 40.1 42.9 45.8 48.6                33.9 36.0 38.4 41.1 43.9
                            70   41.1 44.1 47.1 50.2                     36.0 39.0 42.0 45.1
                                 41.4 44.3 47.2 50.2                     37.0 39.5 42.3 45.2
                            75   46.5 49.5 52.5                          42.1 45.1 48.1
                                 46.6 49.6 52.6                          42.3 45.2 48.2
                            80   51.5 54.5                               45.6 48.7
                                 51.5 54.5                               45.8 48.7

          Table 3.6: Equivalent continuous sound level LAS,eq(3) depending on the combinations
          of maximum SPL and number of ANEs per night (top and bold: aircraft noise only,
          bottom: aircraft noise plus constant background noise level of about 30 dB(A)).



                                                              16
3.4.2     Acoustics in the field studies


Both field studies STRAIN V and VI were conducted in the vicinity of Co-
logne/Bonn Airport. This chapter describes the acoustical setup in the field
studies. Results of the analysis of the acoustical signals sampled during the
field studies can be found in chapter 4.

In the field studies, solely the noise generated by air traffic at the Airport
Cologne/Bonn was sampled, i.e. no additional ANEs were presented via
loudspeakers, as was sometimes done in field studies by other investigators.
A sketch of the acoustical setup is shown in Figure 3.2.



                                                       Bedroom

                2 meters




                                         1                 2                 3
                                                 Trigger                     Auto-
                                                                            Trigger

                                                             Laptop


 Figure 3.2: Acoustical setup of the field studies STRAIN V und VI (schematically).


Three class-1 sound level meters (NC10, Cortex Instruments) were simulta-
neously used. One sound level meter (#1) recorded noise events outside the
bedroom with a distance of two meters to the windows, while two more



                                           17
sound level meters (#2 and #3) recorded noise events inside the bedroom
at the sleeper's ear.

The SPLs LAS and Llin were continuously sampled and stored during the
whole night. Once a certain background noise level (L90) was exceeded
(usually by at least 4 dB), #1 recorded the actual noise event with a sam-
pling rate of 24 kHz until the difference to the background noise level fell
again below 4 dB, but at least for 30 s. The single noise events were stored
as wav-files. Hence, the identification of the noise source (e.g. aircraft,
road, rail) was possible. Simultaneously, with the beginning of the re-
cording of the noise event outside, #2 was triggered and recorded the
noise event synchronously with #1, but now inside the bedroom. As re-
cords of the same noise event existed both outside and inside, conclusions
about dampening properties of windows and masonry were possible (see
chapter 4).

As in the laboratory studies, the trigger signal was synchronously recorded
with the electrophysiological signals at a sample rate of 8 Hz. In that way,
an event correlated analysis of acoustical and electrophysiological signals
with a resolution of up to 125 ms was possible in the field studies as well.

A third sound level meter (#3) recorded noise events inside the bedroom as
soon as a certain background noise level was exceeded (usually by at least
4 dB). In that way, it was possible to additionally identify noise events origi-
nating inside the bedroom or house (e.g. snoring) and that otherwise might
have been missed, as sound level meter #2 was triggered from the outside
sound level meter.


3.5       Electrophysiological signals


All electrophysiological signals were synchronously sampled during the
night, electronically amplified, converted from analog to digital signals and
sent via optical fibers to a personal computer (laboratory) or laptop (field)



                                      18
where they were reproduced on a computer screen for monitoring pur-
poses, and stored for subsequent analyses. The acoustical signals were syn-
chronously stored.

Scoring of sleep stages was performed according to the accepted standard
of Rechtschaffen and Kales [21] making use of the polysomnographic re-
cords of each night. The electroencephalogram (EEG) was derived with sil-
ver/silver chloride electrodes in the positions A1, A2, C3 und C4, the elec-
trooculogram (EOG) with electrodes placed near the edge of the right and
left eye and the electromyogram (EMG) with electrodes placed on the skin
above the muscles of the chin.

ECG and finger pulse amplitude were sampled continuously as well in order
to be able to detect vegetative arousals during sleep. Respiratory move-
ments of the thorax were measured with piezzo technique built into a tho-
racic strap. Air flow at mouth and nose were sampled with a thermistor.
Both signals were used to diagnose sleep related breathing disorders. The
thoracic strap also contained a position sensor in order to detect body
movements during sleep.

The simultaneous recording of electrophysiological and acoustical signals al-
lowed for the differentiation of spontaneous and noise induced reactions
during sleep.

Additionally, study participants wore actimeters 24 hours a day, i.e., also
during the daytime.


3.6      Biochemical methods


During the laboratory studies, excretion rates of the stress hormones corti-
sol, adrenalin, and noradrenalin and electrolytes (potassium, sodium, mag-
nesium, calcium) were determined from urine samples collected within two
defined periods daily: between 7 pm and 11 pm (evening urine), and be-
tween 11 pm and 7 am (night urine).


                                     19
During the field studies, however, only night urines were collected between
the time of going to bed in the evening and the waking-up time in the
morning.

The urine samples for adrenalin and noradrenalin were analyzed by high
performance liquid chromatography (HPLC) and electrochemical detector.
Unconjured cortisol concentrations in urine were determined by immuno
assays, samples from 164 subjects by a radio immunoassay (RIA), and from
32 subjects by a solid phase chemo luminescence enzyme immunoassay
(LEIA). Both methods are not immediately comparable. Therefore, here sta-
tistical analyses on relative changes were applied too, i.e. against night 2,
the baseline night. The determination of potassium and sodium concentra-
tions was by ion selective electrodes that of magnesium and calcium by
photometry of their colored complex compounds in standard clinical rou-
tine methods.


3.7        Performance


Computer-assisted performance tests represent a very important method
for the detection of secondary effects of nocturnal aircraft noise on mental
performance.

The tests used for the evaluation of changes in mental performance capaci-
ties were taken from the so called AGARD-STRES-battery [8] and contained:

• Single reaction time – SRT

• Memory search task with 4 letters – MS4

• Memory search task with 6 letters – MS6

• Unstable tracking task – UTT

In order to attain a stable performance, both in the laboratory and in the
field studies subjects had to perform 40 training sessions of all tests prior to



                                      20
the beginning of the study. All performance tests (total process time about
25 min) were conducted in the evening (about two hours before going to
bed) and in the morning (after getting up) to be able to detect effects of
sleep disturbances induced by aircraft noise.


3.8       Psychology


The investigation of psychological effects contained the measurement of
subjective changes of sleep sensation, annoyance, mood, strain and recrea-
tion depending on the nocturnal aircraft noise exposure. Hence, in both
laboratory and field studies the following questionnaires were used:

• Fatigue Checklist FAT [26]

• Aircraft Noise Questionnaire FNL [5]

• Multidimensional Mood Questionnaire (Mehrdimensionaler Befindlich-
   keitsfragebogen MDBF) [29]

• Strain and Recreation Questionnaire (Erholungs- und Belastungsfrage-
   bogen EBF) [13]

All variables were analyzed in connection with physical parameters (maxi-
mum SPL LAS,max, continuous equivalent sound level LAS,eq, number of ANE)
as well as psychological moderators in order to derive dose-response-
relationships.


3.9       Data analysis


Data analysis was performed with the software packages SPSS (SPSS Inc.,
version 11.5.1) und EGRET (Cytel, version 2.0.31). Chapters 4 to 8 inform
about the individual statistical methods used for data analysis of special
topics (sleep, performance, etc.).




                                     21
4         Acoustics – results and discussion of the data of the
          field studies

          Uwe Müller



In the field studies the sound pressure level was recorded continuously dur-
ing the entire night outside and inside the bedroom at the sleeper's ear.
Additionally, all those noises were recorded with 24 kHz sample rate,
whose sound pressure level was at least 4 dB(A) above the background
level (see also chapter 3.4.2).

This permitted an unambiguous identification of each noise event during
the entire night. For every study night each noise event was marked at the
beginning and the end and identified with a comment. For each event, the
analysis program computed the overall length, the t10-time, the maximum
level, the time of occurrence of the maximum level, the level rise, the en-
ergy equivalent continuous noise level Leq_event (e.g. for airplanes, cars, all
traffic noise etc.), and the background level one minute prior to the start of
each event. For the evaluation of noise effects the sound pressure level at
the top of the bed was relevant in order to allow for an event correlated
analysis of acoustical data (measured at the sleeper's ear) and electro-
physiological signals. However, the above-mentioned values were calcu-
lated both for inside and outside the bedroom.

In the 14 months of the field studies, 46 locations containing a total of 64
subjects were examined for nine consecutive nights. The investigations took
place at nine measuring points in the spring, at eleven in the summer, at 19
in the autumn and at seven in the winter. 394 nights, in which acoustical
data of both inside and outside measurements were available, were consid-
ered in the following analyses.




                                      22
                                    2.0


       freq. of occurence / night
                                            outside                                                median:      64.0 dB(A)
                                    1.5


                                    1.0


                                    0.5


                                    0.0
                                            20                  40              60                       80
                                                             maximum sound pressure level [dB(A)]
                                    2.0
       freq. of occurence / night




                                          at the sleeper's ear
                                    1.5                                                          median:      44.0 dB(A)


                                    1.0

                                                                                    average total noise length: 78.0 [s]
                                    0.5


                                    0.0
                                            20                    40               60                    80
                                                                 maximum sound pressure level [dB(A)]




 Figure 4.1: Frequency distribution of the maximum sound pressure levels of aircraft
 noise events 2 meters in front of the windows and inside the bedroom (at the sleeper’s
 ear) in 1 dB(A)-steps, averaged over 394 nights. For this evaluation only aircraft noise
 events not disturbed by other noise sources were considered.


In these 394 nights, altogether 16,102 airplanes were counted (that corre-
sponds to 40.9 per night, averaged over the entire week). 14,247 aircraft
noise events were not disturbed by other noise events at the sleeper's ear
and will be designated "undisturbed" from now on. Furthermore, 12,256
cars (11,653 undisturbed), 239 motorcycles (217 undisturbed) and 127
trucks (120 undisturbed) were identified. This list indicates that the measur-
ing points were predominantly selected in residential areas strongly af-
fected by aircraft noise but calm in terms of other traffic noise. Figure 4.1
shows the frequency distribution of the maximum sound pressure levels of
the aircraft noise events, averaged over all 394 nights and thus not differ-
entiated by the weekday.



                                                                             23
Subjects of the field studies were asked not to change their usual sleep
habits during the investigation, with the restriction that a minimum bed
time was given from midnight to 6.00 am. Hence, subjects were also free
to chose the window position they wanted to sleep with.

The difference of the sound pressure levels measured inside (at the
sleeper's ear) and outside the bedroom is particularly determined by the
size of the window gap and the distance of the microphone inside the bed-
room to the window, additionally also by the window size, its sound reduc-
tion index, the frequency distribution of the aircraft noise event and the
weather conditions. In order to consider these additional, non-constant pa-
rameters for the calculation of an average sound pressure level difference,
the Leq_event of each undisturbed aircraft noise event was calculated for each
night. From these values (36 per night on average) the median was calcu-
lated. Thus, for each of the nine study nights at each measuring point a
median was received, from which the total median for the measuring point
was then determined. At three measuring points the window position was
changed for at least one of the nine study nights. These measuring points
contributed twice to the calculation, however with a smaller number of
nights representing the total median of the according window position at
that measuring point.

Figure 4.2 shows the results for the average differences in sound pressure
level measured outside and inside (at the sleeper's ear) the bedroom. Since
no standardized acoustical measuring conditions were present (since the
goal of the study was the determination of the sound pressure level and
the identification of the noise event at the ear of the subjects, which per-
mitted an event correlated analysis), the sound pressure level differences of
the different window positions scatter relatively strong. Although the dis-
tance of the interior microphone to the window was not measured, this
strong dispersion is due to the different distances of the beds to the win-
dows, since with tilted windows the exact gap opening width and the exact
window area and with closed windows the sound reduction index and the


                                     24
window size contribute only little to the sound pressure level difference
compared to standard values for tilted and closed windows [14]. The me-
dian values of 28.4 dB(A) for closed, 18.4 dB(A) for tilted and 13.5 dB(A)
for open windows (only three measuring points) indeed give a good repre-
sentation of the differences in sound pressure levels for aircraft noise events
in practice.


                                 8

                                             closed window,    Med: 28.4 dB(A), N=12
                                 7
                                             tilted window,    Med: 18.4 dB(A), N=34
                                             open window,      Med: 13.5 dB(A), N=3
                                 6
        frequency of occurence




                                 5


                                 4


                                 3


                                 2


                                 1


                                 0
                                     10   15            20          25           30    35
                                          difference of sound pressure level [dB(A)]




 Figure 4.2: Average difference in sound pressure levels measured inside (at the
 sleeper's ear) and outside the bedroom. The median was taken from nine observation
 nights, whereby the smallest and largest value did not enter the calculation. At three
 measuring points the window position changed during the experiment. For a detailled
 description of the calculations of the differences see text.


In contrast to the determination of the energy equivalent continuous sound
level LAS,eq(3) for the controlled acoustic conditions in the sleep laboratory,
the LAS,eq(3) in the field does not exclusively depend on aircraft noise events,
but also on contributions from other traffic and the respective weather
situation.


                                                              25
Averaged over all measuring points for the time between midnight and
6 am in spring an average LAS,eq(3) of 54.8 dB(A) for outside and 34.3 dB(A)
for inside was calculated, in summer 52.8 dB(A) outside and 37.3 dB(A) in-
side, in autumn 53.5 dB(A) outside and 36.0 dB(A) inside and in winter
55.0 dB(A) outside and 36.8 dB(A) inside. Considering the fact that the av-
erage number of aircraft movements between midnight and 6.00 am in the
annual average amount to 21.1% during spring, to 30.0% during the
summer, to 22.9% during autumn and to 25.9% during winter, then one
recognizes that the LAS,eq(3)-calculations are also strongly affected by the dif-
ferent weather situations (wind and rain lead to higher sound pressure lev-
els) and the window positions varying with the seasons.




                                       26
5        Sleep

         Mathias Basner



Sleep is a condition with a restricted perception of the individual towards
himself and the environment. Humans spend about one third of their lives
sleeping. With the sampling of certain biologic signals, the so called poly-
somnography, an awake human with closed eyes can be differentiated
from a sleeping human. Additionally, making use of polysomnographic sig-
nals, sleep can be divided into six distinct and functionally different states:
Firstly, being awake can be distinguished from being asleep. Then, sleep it-
self can be classified into REM-sleep (see below) and non-REM-sleep, while
non-REM-sleep can be subdivided again into sleep stages S1, S2, S3 and
S4. Sleep stages S1 and S2 are also called light sleep and stages S3 and S4
deep sleep. These terms are based upon the fact that humans are more
easily awakened from light than from deep sleep. For the classification, a
night is split into 30 sec epochs, and each of these intervals is categorized
in one of the five different sleep stages or "Awake". Following an interna-
tional standard [21] one of the sleep stages is assigned to each 30 sec ep-
och. Currently, there is no reliable automated sleep stage analysis, and
hence sleep stages have to be classified by an experienced human analyzer.
This process is time consuming and therefore expensive, which is why poly-
somnography has been applied only in smaller noise effects research stud-
ies with only a few subjects in the past.

Polysomnography consists of the following electrophysiological signals:

Electrodes attached to the scalp are used to derive the electroencephalo-
gram (EEG). Besides EEG-frequency and -amplitude special EEG-patterns
(e.g. K-complexes and spindles) play an important role for sleep stage clas-
sification. Slow waves with high amplitude are characteristic for deep sleep,
e.g., and hence deep sleep is also called slow wave sleep (SWS).


                                      27
Eye movements (electrooculogram – EOG) are derived from two elec-
trodes usually attached close to the eyes. Slow eye movements can be ob-
served while falling asleep (S1), whereas rapid eye movements are charac-
teristic for REM-sleep. These eye movements are the reason for the term
"REM-sleep" (rapid eye movement sleep).

Muscle tension (electromyogram – EMG) is derived from two electrodes
attached to the skin above the chin muscles. Muscle tension decreases with
increasing depth of sleep from stage S1 to S4, but shows the lowest values
during REM-sleep.

Sleep stages are not uniformly distributed over the sleep period. In 128
baseline nights of the laboratory studies the different sleep stages were dis-
tributed as follows: S2 ~ 52%, REM ~ 22%, SWS ~ 16%, Awake ~ 8%
und S1 ~ 2%. The proportion of sleep stage S2 is highest, whereas S1, rep-
resenting a special transitional state from being awake to falling asleep, can
only seldom be observed in undisturbed nights. Furthermore, the fractions
of the different sleep stages in total sleep time are age dependent. A reduc-
tion of SWS-fractions, especially in favor of S2 can be seen in older people.
Often, no SWS at all can be found in subjects very advanced in years.

Alternating episodes of non-REM-sleep and REM-sleep are also called sleep
cycles. Each normal night consists of several cycles lasting between 60 and
120 min each. The composition of the sleep cycles changes in the course of
the night: SWS dominates the first third of the night, whereas REM-sleep
prevails in the last third of the night. A so called hypnogram is obtained by
plotting the different sleep stages against the elapsed sleep time and visual-
izes the structure of sleep. Figure 5.1 shows the hypnograms of two par-
ticipants of the laboratory studies.




                                       28
    Awake
      REM
        S1
        S2
        S3
        S4        B
    Awake
      REM
        S1
        S2
        S3
        S4        A
          23:00       0:00   1:00    2:00        3:00   4:00   5:00    6:00     7:00
                                                 Time




 Figure 5.1: Hypnograms of two participants of the laboratory studies. In both nights
 64 aircraft noise events with a maximum SPL of 65 dB(A) at the sleeper's ear were
 played back.


Both subjects received 64 aircraft noise events with a maximum SPL of
65 dB(A) at the sleeper's ear during the night shown in Figure 5.1. The
group of all participants who were exposed to the combination
64 x 65 dB(A) woke up with a mean probability of 23.3% while being ex-
posed to aircraft noise. Subject A woke up with a probability of only 3.3%,
the lowest percentage observed in the group. Subject B, on the other hand,
woke up with a probability of 88% and therefore expressed the highest
sensitivity to aircraft noise.

Both extremes of the percentage being awakened are shown in Figure 5.1
on purpose, to point out the strong differences in environmental noise sen-
sitivity between subjects. Hence, investigations with small sample sizes, e.g.
16 subjects, are insufficient, as 16 very insensitive or 16 very sensitive sub-
jects may be sampled by chance, which would mean that the results are


                                            29
strongly influenced by chance as well. This bias can be minimized by the
use of larger sample sizes and a suitable selection process of study appli-
cants.

The restorative power of sleep is not only influenced by sleep duration but
also by sleep structure. Present scientific knowledge assumes that the six
sleep stages differ in their recuperative value, although the functions of the
different sleep stages and the mechanisms of these functions are still not
exactly known. Deep or slow wave sleep (SWS) is considered to be particu-
larly important for the restorative power of sleep because of the following
reasons [7]:

• its proximity to sleep onset

• its immediate rebound after sleep deprivation

• its association with high sensory thresholds and the excretion of growth
  hormones

Results of the recent past indicate that SWS is involved in the consolidation
of explicit memory contents, whereas REM-sleep seems to be important for
the consolidation of implicit memory contents [22]. Stages Awake and S1
do not contribute to the recuperative value of sleep or only very little re-
spectively, whereas sleep stage S2 takes an intermediate position.


5.1      Influence of nocturnal aircraft noise on sleep


The regeneration of mental and physical powers primarily takes place dur-
ing sleep. Undisturbed sleep is a prerequisite for this recuperative process.
The human organism recognizes, evaluates and reacts to environmental
sounds even while asleep. These reactions are part of an integral activation
process of the organism and express themselves e.g. as changes in sleep
structure or increases in heart frequency. Environmental noise may decrease
the restorative power of sleep by means of repeatedly occurring activations,



                                     30
where the extent of the reduction depends on the magnitude as well as on
the frequency of these activations. Probability and magnitude of the activa-
tion are predominantly influenced by the type and sound pressure level of
the noise event, beside many other moderating factors.


5.2      Methods


The influence of nocturnal aircraft noise on sleep was systematically investi-
gated in laboratory and field experiments with 192 subjects in 2,240 sub-
ject nights. This chapter deals with the interrelation of nocturnal aircraft
noise and primary sleep disorders under consideration of many other im-
portant control variables.

For a detailed description of methods and results please refer to the re-
search report DLR-FB-2004-09/E: "Effects of Nocturnal Aircraft Noise Vol-
ume 3: Sleep".

In the laboratory studies, data were sampled between 11 pm and 7 am.
Aircraft noise events (ANEs) were played back equidistantly between
11:15 pm and 6:45 am. In contrast to the laboratory studies, during the
field studies subjects were free to choose their bed time with the restriction
of being in bed between 12 pm and 6 am at least.

Sleep stages were classified by experienced personnel according to the
standards of Rechtschaffen und Kales [21]. Body movements accounting for
more than half of the epoch, that would have been scored as Movement
Time, were classified as Awake instead because it was assumed that these
kinds of movements do not occur without respective cortical activation.

Prior to sleep stage classification, sleep files were renamed and the order
was randomly changed. In this way, the person analyzing the file was
blinded as he or she did neither know which file he or she was scoring and
therefore nor whether and how many ANEs were played back (laboratory)
or observed in reality (field) in that night. This process guaranteed the


                                     31
avoidance of a systematic bias. All investigated nights of on subject were
classified by the same scorer in order to avoid the problem of inter-rater
variability in within-subject comparisons. By means of a randomized se-
quence of the nights to be analyzed, a systematic habituation of the scorer
to the specific EEG-patterns of one subject always in the same order
(nights 1 to 13 or 9 respectively) was avoided.

It was possible to analyze 1,050 of 1,072 exposure nights (97.9%). Data
sampling started after 11:15 pm in one night. In 29 cases the quality of one
of the polysomnographic signals faded before 6:30 am in such a way that
sleep stage classification was no longer possible from the time of signal
loss, and hence only a partial analysis of these nights was possible. In total,
1,463,998 epochs were analyzed including adaptation, baseline, control
and recovery nights.

The data of one subject participating in the field study were not included in
the analysis because of an intrinsic sleep disorder (so called REM-sleep be-
havior disorder). Of the remaining 63 subjects, nights two to nine were
analyzed, since night one served as adaptation night. It was possible to
analyze 492 of the remaining 504 nights (97.6%). Because of the loss of
important polysomnographic signals only a partial analysis was possible in
16 nights. In total, 458,196 epochs were visually scored in the field studies.
In both laboratory and field studies, a total of 1,922,194 epochs were clas-
sified by a human scorer.


5.3       Analysis of the influence of noise on sleep stage
          distribution


Nocturnal aircraft noise may lead to a shortening of total sleep time (TST)
on one hand, but it may also cause changes in the structure of sleep, i.e. al-
terations in the fractions of the different sleep stages in total sleep time, on
the other hand. As aforementioned, especially SWS seems to play an im-
portant role for the recuperative value of sleep, whereas stages Awake and


                                      32
S1 do not seem to promote the process of restoration or only very little. To
the contrary, stages Awake and S1 were established as typical indicators of
disturbed and non-recuperative sleep by sleep medicine. Hence, intrinsic or
extrinsic sleep disorders that are associated with decrements of SWS in fa-
vor of light sleep and Awake may be regarded as potentially harmful for
the recuperative value of sleep.

Figure 5.2 shows the changes of sleep stage fractions of all exposure nights
(n = 996) compared to noise-free baseline nights (n = 112) in the course of
the night of the laboratory studies. All nights were synchronized to time of
sleep onset. Figure 5.2 shows mean differences of noise and baseline
nights, i.e. negative values stand for smaller fractions in the exposure
nights.

                                               5

                                                        Awake&S1 (+2.3 min)
                                               4        S2 (+0.6 min)
                                                        SWS (-4.1 min)
                                               3        REM (+1.3 min)
   Cumulated Sleep Stage Fractions (Minutes)




                                               2
         (Difference Noise - Baseline)




                                               1


                                               0
                                                    0     60             120       180         240         300       360   420
                                               -1


                                               -2


                                               -3


                                               -4


                                               -5
                                                                          Elapsed Time Since Sleep Onset (Minutes)



 Figure 5.2: Changes in sleep stage fractions (mean difference Noise – Baseline) in the
 course of the night in the laboratory studies.




                                                                                         33
With a sleep period time of 7 h and 24.5 min, total sleep time was statisti-
cally non-significantly decreased by 1.8 min (p=0.262). Simultaneously,
changes in the structure of sleep were obvious: SWS decreased on average
by 4.1 min especially in favor of stages Awake and S1 (+2.3 min), but the
changes in the fractions of the different sleep stages were not statistically
significant.

Figure 5.2 compared noise and baseline nights irrespective of number or
maximum SPL of ANEs, whereas Figure 5.3 compares baseline nights with
those exposure nights where the maximum SPL of ANEs was 55 dB(A), de-
pending on elapsed sleep time and number of ANEs.

                                     10


                                      8


                                      6
  Cumulated SWS Fraction (Minutes)




                                      4
    (Difference Noise - Baseline)




                                      2


                                      0
                                           0      60          120           180         240          300      360   420
                                      -2


                                      -4


                                      -6       4x55 (N39, B32)
                                               8x55 (N40, B32)
                                               16x55 (N31, B31)
                                      -8       32x55 (N31, B31)
                                               64x55 (N32, B32)
                                               128x55 (N28, B28)
                                     -10
                                                                   Elapsed Time Since Sleep Onset (Minutes)



 Figure 5.3: Changes in SWS-fractions in noise nights with maximum SPLs of 55 dB(A)
 at the sleeper's ear compared to noise-free baseline nights depending on elapsed sleep
 time and number of ANEs (N = number of noise nights and B = number of baseline
 nights used for the analysis).




                                                                                  34
In spite of statistically non-significant changes of SWS-fractions, it is obvi-
ous that with a maximum SPL of 55 dB(A) and low numbers of ANEs per
night (here: 4 to 16 ANEs per night) SWS-fractions rather increased than
decreased compared to baseline nights (see Table 5.1). The compensation
of sleep deficits accumulated in preceding nights might explain the increase
of SWS-fractions in nights with only moderate aircraft noise exposure. If
this was true, it would imply two important consequences:

• Noise exposure in the laboratory studies lead to some, although statisti-
  cally non-significant, decrements in the fractions of SWS, because com-
  pensating mechanisms were seen in nights which were only moderately
  disturbed by aircraft noise.

• The detection of increased fractions of SWS in nights only moderately
  disturbed by aircraft noise shows that the compensation of SWS deficits
  acquired in preceding nights is principally possible in nights with few
  ANEs.

If a compensation of decrements in the fraction of SWS was possible in
nights only moderately exposed by aircraft noise, this compensation process
should also particularly work in noise-free recovery nights (night 12). But
analysis of these recovery nights of laboratory studies 1 to 3 did not reveal
any increases in the fractions of SWS.

The fraction of SWS was statistically non-significantly decreased in strongly
exposed nights (here: 32 to 128 ANEs per night). It is also evident that the
reduction of SWS-fractions of some of the combinations of maximum SPL
and number of ANEs per night may be more pronounced than the mean of
4.1 min. (see also Table 5.1).

Table 5.1 shows the change in SWS-fraction in exposure nights compared
to baseline nights depending on maximum SPL and number of ANEs. For
each combination of maximum SPL and number of ANEs the fractions of
SWS were compared with the Wilcoxon signed rank test for paired data.



                                      35
Because of a multiple test situation, significance values were Bonferroni-
corrected and a significant change assumed only if p-values were smaller
than or equal 0.0017 (=0.05/30).

                                 Number of Aircraft Noise Events

                                 4           8            16          32          64         128

                            45                                                    +1,9 min
                                                                                  p=0,739
                            50                            +1,4 min    -5,6 min    +1,7 min   -9,4 min
                                                          p=0,647     p=0,243     p=0,721    p=0,073
                            55   +7,7 min    +2,9 min     +3,2 min    -3,2 min    -7,9 min   -5,5 min
Maximum SPL LAS,max in dB




                                 p=0,139     p=0,507      p=0,241     p=0,300     p=0,166    p=0,232
                            60   -4,7 min    +0,7 min     -5,7 min    -18,0 min   -0,2 min
                                 p=0,347     p=0,975      p=0,349     p=0,002     p=0,888
                            65   +0,1 min    -2,3 min     -8,9 min    -2,3 min    -6,6 min
                                 p=0,961     p=0,652      p=0,079     p=0,769     p=0,051
                            70   -3,0 min    +2,6 min     -17,8 min   -9,0 min
                                 p=0,252     p=0,376      p=0,004     p=0,089
                            75   -5,6 min    -7,8 min     -4,3 min
                                 p=0,170     p=0,018      p=0,400
                            80   -7,7 min    -8,6 min
                                 p=0,092     p=0,248


       Table 5.1: Fractions of SWS: Difference Noise – Baseline (positive numbers correspond
       to increased amounts of SWS in noise nights compared to baseline nights). Statistically
       significant changes were assumed at p<0.0017 after Bonferroni-correction of alpha.


Altogether, 30 combinations of maximum SPL and number of ANEs can be
differentiated. If nocturnal aircraft noise had no impact on SWS-fractions, it
would be expected that SWS-fractions are both increased and decreased in
about 50%, i.e. in 15 of 30 cases each. The results of the analysis show
(see Table 5.1) that SWS-fractions were decreased in 21 of 30 cases in-
stead. Making the assumption that aircraft noise has no impact on SWS-
fractions, the probability to observe a decrease in SWS-fractions in 21 of 30
cases is only 4.3% (exact binomial test, two-sided, test fraction 50%). This
supports the hypothesis of nocturnal aircraft noise having an impact on
SWS-fractions, even though none of the single combinations was signifi-
cantly affected.


                                                               36
Naturally, there were no baseline nights in the field studies. Thus, after a
median-split, the four nights with the higher exposure were compared to
the four nights with the lower exposure to aircraft noise. This is exemplarily
shown for the number of ANEs in sleep period time (SPT) in Figure 5.4.

                                                 5


                                                 4


                                                 3
     Cumulated Sleep Stage Fractions (Minutes)
      (Difference Number HIGH - Number LOW)




                                                 2


                                                 1


                                                 0
                                                      0        60      120        180         240          300     360   420
                                                 -1


                                                 -2

                                                          SWS
                                                 -3
                                                          Awake & S1
                                                          S2
                                                 -4       REM

                                                 -5
                                                                        Elapsed Time Since Sleep Onset (Minutes)




 Figure 5.4: Changes in sleep stage fractions in nights with on average 52.2 ANEs
 compared to nights with on average 21.9 ANEs (median-split) depending on elapsed
 sleep time (n = 472 nights).


The different sleep stage fractions varied +/- 4 min about zero and did not
differ significantly from null (Wilcoxon signed rank test for paired data). The
analysis was repeated with the following acoustical parameters:

• LAS,eq at the sleeper's ear in sleep period time, induced by aircraft noise
  events with maximum SPLs higher than 35 dB(A)

• LAS,eq at the sleeper's ear in sleep period time, induced by traffic noise
  events with maximum SPLs higher than 35 dB(A)

• LAS,eq at the sleeper's ear in sleep period time all sounds


                                                                                    37
The magnitude of the differences was comparable to those described in
Figure 5.4 and neither were statistically significant.


5.4       Event correlated analysis


An event correlated analysis establishes a direct temporal association be-
tween the occurrence of an ANE and the reaction of the investigated sub-
ject to the ANE. An event correlated analysis is only possible because of the
synchronous sampling of electrophysiological and acoustical signals. A trig-
ger signal guaranteed this analysis with a maximum temporal resolution of
125 ms.

Variables like the nocturnal secretion rate of stress hormones or the annoy-
ance of study subjects asked for in the morning by questionnaires are rep-
resented in a single datum, which summarizes the effects of all nocturnal
ANEs. These integrative measures are unsuitable for an event correlated
analysis, because the connection to single noise events cannot be made. In
sleep medicine often variables are used that characterize the whole night,
as the amounts of the different sleep stage fractions. These are also cumu-
lative measures and therefore not suitable for an event correlated analysis.


5.4.1     What is a suitable descriptor for noise induced sleep disturbances?


The reactions of sleeping humans to aircraft noise are non-specific, since
they may also be observed during natural and undisturbed sleep. Reactions
observed during an ANE can not be differentiated from spontaneous reac-
tions according to electrophysiological criteria. Therefore, it is necessary to
synchronize the sampling of acoustical and electrophysiological data, which
allows for the discrimination of spontaneous reactions from reactions ob-
served during an ANE.

As spontaneous reactions occur irregularly, they may also be observed dur-
ing an ANE. Therefore, if there is a reaction during an ANE, it is necessary


                                       38
to investigate, how often this reaction would have taken place spontane-
ously anyhow, i.e. without the influence of the ANE. In epidemiology the
term attributable risk is often used in this context. The probability of a reac-
tion Pinduced induced by aircraft noise is calculated as:

Pinduced = PANE – Pspontaneous (1)

The time period after the beginning of an ANE screened for reactions of the
sleeper is also called noise window. The length of the noise window was
chosen to maximize the probability of reactions induced by aircraft noise
(Pinduced in equation 1). The size of the noise window was 2 epochs (60 sec-
onds) in the laboratory studies. Probably because of longer noise events on
average, a noise window of 90 seconds turned out to be optimal in the
field studies.

Several potential indicators for noise induced sleep disturbances have been
identified and proposed in the past. Brief EEG- and EMG-activations are
called arousals. Because of their short duration they are not classified as
stage Awake according to the rules of Rechtschaffen und Kales. In contrast
to arousals, awakenings are defined as EEG- and EMG-activations that last
at least 15 sec and therefore lead to a change to the sleep stage Awake.
Sleep stage changes are defined as transitions from one sleep stage to a
different sleep stage. In the context of noise effects research, commonly
only those sleep stage changes leading to a flattening of sleep are consid-
ered, e.g. changes from deep sleep stage S4 to the light sleep stage S2.

Polysomnographic studies conducted in the past predominantly used awak-
enings as the primary indicator of sleep disturbances induced by environ-
mental noise. Because of the following reasons awakenings are appropriate
indicators for sleep disturbances induced by environmental noise:

• The awakening is the strongest form of activation of the sleeping or-
   ganism. The consequences for the restorative functions of sleep are ac-
   cordingly severe.



                                         39
• Awakenings are relatively specific, i.e. the frequency of spontaneous
  awakenings is relatively low compared to other indicators. In the 112
  baseline nights of the experimental group in the laboratory studies, on
  average about 24 spontaneous awakenings were observed. Spontane-
  ous sleep stage changes were seen more than twice as often (on average
  about 52 changes per night). Mathur and Douglas [17] investigated the
  spontaneous onset of EEG-arousals according to the ASDA-criteria [3].
  They found on average about 21 arousals per hour of sleep. If the mean
  sleep period time (SPT) of 411.5 min of the noise-free baseline nights is
  taken as a basis, this value corresponds to about 144 spontaneous EEG-
  arousals per night.

• We observed in our own investigations that the amplitude and/or the
  frequency of heart frequency accelerations are relatively low, if there is
  no awakening reaction in the EEG at the same time. But especially the
  regular occurrence of these nocturnal vegetative reactions seems to
  be a possible cause for the development of high blood pressure and the
  associated diseases of the cardiovascular system (myocardial infarction,
  stroke). The degree of vegetative reactions that come along with sleep
  stage changes or short arousals alone is low compared to reactions asso-
  ciated with awakenings.

• The majority of awakenings lasts for exactly one epoch (15 to 45 sec)
  and, therefore, is too short to be remembered on the next day. On the
  other hand, some awakenings may be longer and, therefore, may be as-
  sociated with the occurrence of waking consciousness. As a conse-
  quence, these longer awakenings may be remembered on the next day.
  In this case, they will also dominate the subjective assessment of sleep
  quality and quantity on the next day. Criteria limiting the number of
  short awakenings (15 to 45 sec) will simultaneously restrict the number
  of longer awakenings that may be remembered on the next day.
  Therefore, these criteria will make sure that the assessment of sleep
  quality and quantity will be high. Sleep stage changes and arousals will


                                    40
   not be remembered on the next day, as they do not lead to the occur-
   rence of waking consciousness.

Because of the sensitivity putatively being too low, some authors tena-
ciously criticize the use of awakenings as indicators for noise-induced sleep
disturbances [15]. However, the results of our study definitely show that
the sensitivity of awakenings as indicators for noise-induced sleep distur-
bances is sufficiently high, as in the laboratory studies no thresholds were
found above 45 dB(A). In the field studies, the observed threshold lay only
about 6 dB(A) above background noise levels (see below).

Sleep stage S1 does not or only little contribute to the recuperative value of
sleep. On the contrary, increased fractions of sleep stage S1 were identified
as typical effects of sleep fragmentation in the past [30]. Hence, in this
analysis not only changes to stage Awake were regarded as relevant sleep
stage changes, but also changes to sleep stage S1. This preventive proce-
dure increased the fraction of reactions associated with ANEs without sig-
nificantly lowering the specifity of the proposed indicator. From now on in
this report, the term "awakenings" implicitly means transitions from sleep
stages REM, S4, S3 or S2 to the sleep stages S1 or Awake.


5.4.2     Methods


The awakening probability does not solely depend on the maximum SPL of
a noise event. On the one hand, other acoustic properties of the noise
event like the spectral composition or the duration play an important role.
On the other hand, situative and individual factors moderate the reaction of
the sleeping organism to ANEs. In order to determine the size of the effect
of the maximum SPL of an ANE, the other important moderating variables
thoroughly have to be controlled. This is possible in a regression model and
is also called adjustment in this context.




                                       41
A logistic regression model was used as awakenings occur (1) or they do
not occur (0), i.e. the dependent variable is dichotomous in this case. Stan-
dard statistical procedures assume that the observations are independent.
This assumption is violated in our study as each subject entered more than
one observation in the analysis, i.e. each volunteer was exposed to more
than one ANE. This non-independency of data was accounted for by the
use of a random effects logistic regression.

In this summary the complex methodological approach used for the event
correlated analysis cannot be described in detail. For a detailed description
please refer to the research report that solely addresses the topic "sleep".
In this report, aspects of sample size estimation, selection of statistical
methods and modeling aspects will be described. However, only the most
important results can be briefly described below.


5.4.3    Results of the laboratory studies


During nights 3 to 11 a total of 30,584 ANEs was played back. In 2,857 ca-
ses (9.3%) the subject was awake or in sleep stage S1 prior to the occur-
rence of the noise event. In another 1,483 cases (4.8%) the subject had not
fallen asleep yet, or the classification of sleep stages was impossible be-
cause of signal losses before the end of the night. 836 noise events (2.7%)
could not be analyzed because the data sampling of the whole night failed.
Therefore, a total of 25,408 ANEs (83.1%) were included in the final analy-
sis. The noise window (see above) was set to 2 epochs (60 sec), which
maximized the number of awakenings attributable to ANEs.

Figure 5.5 shows the results of a random effects logistic regression with
maximum SPL and maximum SPL2 as the only independent variables. The
number of ANEs per night is not considered in this model. The red dia-
monds represent the observed awakening probabilities depending on
maximum SPL. The observed data and the prediction of the regression
match very well. The relation of maximum SPL and awakening probability is


                                     42
obviously non-linear. The result of the logistic regression predicts an awa-
kening probability of about 11% at 45 dB(A) and of about 65% at
80 dB(A).

                          70%
                                          Aircraft Noise Events (Regression)
                          60%             Spontaneous (6.3%)
                                          Observed Data

                          50%
    Percentage Awakened




                          40%


                          30%


                          20%


                          10%


                          0%
                                40   45        50         55       60          65   70   75   80   85
                                                           Maximum SPL LAS,max in dB


 Figure 5.5: Laboratory study. Random effects logistic regression. Maximum SPL and
 maximum SPL2 as the only independent variables. Based on 112 subjects and 25,408
 ANEs.


The spontaneous awakening probability was estimated in the noise-free
baseline nights for each individual subject. The noise window of each ANE
was screened for spontaneous awakenings in the noise-free baseline nights
at the same time as in the exposure nights. The awakening probability for
these so-called virtual noise events was calculated as 6.3% (see lower pink
line in Figure 5.5).

Figure 5.5 also shows that in the laboratory studies a threshold for noise
induced awakenings was not found above maximum SPLs of 45 dB(A), i.e.
15 dB(A) above the constant background noise level caused by the air con-
dition system. The regression line rather asymptotically converges with the



                                                                   43
line representing the spontaneous awakening probability. This result is in
contradiction to the findings of Griefahn et al. [11], who found a threshold
of about 60 dB(A) in a meta-analysis. Maschke et al. [16] assume a thresh-
old of about 33 dB(A) after a re-analysis of the same data. In both cases
linear regression models were used, although some care has to be taken in
the application of linear models in the context of binary response data, as
the assumptions of homoscedasticity and of normally distributed error
terms necessarily are violated.

On the other hand, Figure 5.5 shows that even ANEs with maximum SPLs
of 80 dB(A) did not lead to a transition to sleep stage S1 or Awake in about
35% of the study subjects.

The regression model shown in Figure 5.6 contains the variables "sleep
stage prior to the beginning of the ANE" (PriorSleepStage) and "elapsed
time after sleep onset" (ElapsedTime) as independent variables beside maxi-
mum SPL and maximum SPL2.




                                    44
                         100%

                         90%

                         80%

                         70%
   Percentage Awakened




                         60%

                         50%

                         40%

                         30%

                         20%

                         10%

                          0%
                                45   50   55         60        65          70   75   80
                                               Maximum SPL LAS,max in dB



 Figure 5.6: Laboratory study. Random effects logistic regression with 95% confidence
 limits. Model contains maximum SPL, maximum SPL2, sleep stage prior to ANE (stage
 S2) and elapsed sleep time (middle of 2nd half of the night) as independent variables.
 Based on 112 subjects and 25,408 ANEs.


These two additional independent variables were set to identical values in
the models on which Figure 5.6 (laboratory studies) and Figure 5.8 (field
studies) are based, which allows for a comparison of both models. Because
of preventive reasons, PriorSleepStage was chosen as S2, the sleep stage
with the highest sensitivity for noise induced awakenings (see below).
ElapsedTime was set to 601 epochs (about 5 hours after sleep onset), cor-
responding to the middle of the more sensitive second half of the night in
the field studies. Accordingly, the regression line in Figure 5.6 is shifted to
higher awakening probabilities compared to that line in Figure 5.5.

The precision of the prediction can be read off the width of the 95% confi-
dence interval (grey lines in Figure 5.6). Because of the large sample size
the precision of the prediction is very high. It varies between 3.2% at
49.5 dB(A) and 7.5% at 80 dB(A).


                                                      45
Beside the maximum SPL of the ANE the following variables were checked
first in univariable and then in multivariable models for their influence on
noise-induced awakenings:

Acoustical variables: There was a habituation to ANEs within one night, as
the awakening probability decreased with the number of ANEs already
played back. This effect was more pronounced in ANEs with higher maxi-
mum SPLs. Awakening probability also rose with the length of the noise-
free interval between two ANEs. In the univariable model, the ANEs of pla-
nes taking off was associated with higher awakening probabilities than
those of landing planes. This effect disappeared after adjusting for the du-
ration of the ANE. Awakening probabilities decreased with increasing dura-
tion of the ANE, and the lower steepness of the rise of the SPL may be a
reason for this observation. The agreement of observed data and regression
results was considerably lower when the SEL (single event level) was used
instead of maximum SPL LAS,max. These results are supported by a current
Dutch study [19].

Situative variables: The sleep stage prior to the beginning of an ANE was a
strong moderator. Awakening probability decreased in the order
S2    REM      S3     S4. As the probability of spontaneous awakenings was
comparatively high during REM-sleep, the order of noise induced awaken-
ings (PANE - Pspontaneous) decreased in the order S2   S3   REM   S4. There-
fore, light sleep stage S2 was most sensitive for noise induced awakenings.
Awakening probability also decreased with the number of investigation
nights, which may be interpreted in terms of habituation and/or a cumu-
lated sleep deficit. According to the degradation of sleep debt, awakening
probability increased with elapsed sleep time. The degree of fatigue, asked
for immediately prior going to bed by Fatigue questionnaires (FAT), had no
significant influence on awakening probability.

Individual variables: The mean awakening probability of men (20.1%) was
lower than that of women (21.4%). 45 year old subjects were least sensi-



                                        46
tive for noise induced awakenings, both older and younger subjects
showed higher awakening probabilities. In contrast to these findings, the
results of an actigraphic field study by Passchier-Vermeer et al. [19] indicate
that 46 year old subjects were most sensitive for awakenings observed dur-
ing ANEs. There is no plausible explanation for these diametrical results.
Possible reasons are differences in the study type (laboratory versus field) or
in the composition of the study sample. Subjects highly annoyed by aircraft
noise prior to the study (not to be confounded with the actual physical
noise load) showed higher awakening probabilities than subjects not or
only moderately annoyed. With the data available a differentiation between
cause and effect is impossible. On the one hand, subjects may be more
strongly annoyed because of more frequent noise induced awakenings dur-
ing the night, or on the other hand they may wake up more often because
of the elevated noise annoyance. Subjects with high scores of subjectively
assessed noise sensitivity showed higher awakening probabilities than sub-
jects with low scores.


5.4.4    Results of the field studies


One of the 64 subjects participating in the field studies had to be excluded
from the analysis because of an intrinsic sleep disorder (so-called REM-sleep
behavior disorder). For the other subjects the first night served as adapta-
tion. In the remaining 504 study nights, the sampling of acoustical and/or
electrophysiological data failed in 21 cases (4.2%). Therefore, 483 nights
contributed to the final analysis. In these nights, 18,509 ANEs were sam-
pled and identified. The subjects were awake or in sleep stage S1 prior to
the beginning of the ANE in 1,954 cases (10.6%). In 999 cases (5.4%) the
subjects had not fallen asleep yet, or the classification of sleep stages was
impossible due to signal loss before the end of the night. Therefore, a total
of 15,556 ANEs contributed to the final analysis. In contrast to the labora-
tory studies, a noise window of three epochs (90 sec) was used in the field
studies, maximizing the fraction of noise induced awakenings.


                                        47
In contrast to the laboratory studies, there was no controlled environment
in the field studies. The difference of the SPL of an ANE and the back-
ground noise level (also called emergence) was identified as an important
control variable for the occurrence of a noise induced awakening beside
the maximum SPL of the ANE in the past. Because of the constant back-
ground noise level of about 30 dB(A) during the laboratory studies, the
emergence of the ANEs was also constant in each night of these studies. In
the field studies the background noise level (LAS,eq) needed to be calculated.
This was done by computing it during the last minute before the beginning
of an ANE. These values varied between 16.4 and 58.3 dB(A) with a me-
dian of 27.1 dB(A). 258 of 15,556 were not considered in the final analysis
because of an emergence less than or equal to 0 dB(A).

There may be noises originating from other sources than aircrafts between
two ANEs, and these may originate either inside or outside the bedroom.
These noises were identified in the field studies using the WAV-files. An
ANE only contributed to the final analysis if the following conditions were
met:

• Within the minute before the beginning of an ANE, only sounds origi-
  nating from the subject itself (except snoring) and noise of a preceding
  aircraft were allowed. The exclusion of preceding ANEs would have lead
  to an underestimation of awakening probabilities in periods of high air
  traffic and therefore would have introduced a systematic bias.

• If there were extraneous sounds during an ANE that might have lead to
  awakenings independent of the ANE, this ANE was not considered in the
  final analysis. Sounds originating from the subject itself during an ANE
  might have occurred as a consequence of a noise induced awakening
  and therefore were explicitly not excluded from the analysis.

The regression models presented below are based on a total of 10,658
ANEs which met the criteria mentioned above. One subject, that was inves-
tigated together with his or her partner, snored continuously, and hence


                                     48
not a single ANE fulfilled the criteria mentioned above. Therefore, only 61
subjects were included in the final analysis, compared to the originally
planned number of 64 subjects.

So-called virtual noise events were also used in the field studies in order to
calculate spontaneous awakening probabilities: Each of the remaining
seven nights of each subject was screened for spontaneous awakenings or
transitions to sleep stage S1 at the exact same time after sleep onset, after
verifying that there was no ANE in the corresponding noise window. Ulti-
mately, 19,608 events contributed to the final analysis of virtual noise
events.

                         25%

                                         Aircraft Noise Events (Regression)

                         20%             Spontaneous (8.7%)
   Percentage Awakened




                         15%




                         10%




                         5%




                         0%
                               25   30        35       40       45        50       55        60       65   70   75
                                                    Maxim um SPL LAS,max in dB at the Sleeper's Ear



 Figure 5.7: Field studies. Random effects logistic regression. Model contains maximum
 SPL, background noise level LAS,eq 60 sec prior to the ANE and their interaction as
 independent variables. Based on 61 subjects and 10,658 ANEs. For the background
 noise level the observed median of 27.1 dB(A) was chosen.


Figure 5.7 shows the results of a multivariable random effects logistic re-
gression model based on the data of the field studies. The model contains
the maximum SPL of the ANE, the background noise level LAS,eq one minute
prior to the beginning of the ANE and the interaction term of both. The lat-


                                                                     49
ter was set to a constant value of 27.1 dB(A) in Figure 5.7 (median). Be-
cause of the variability of the background noise level it is not feasible to
present both the regression line and the observed data at the same time in
Figure 5.7 like in Figure 5.5.

At the same maximum SPL, the awakening probability was noticeably less
pronounced in the field studies compared to the laboratory studies (for a
comparison see Figure 5.9). In the field studies, the highest SPL measured
inside the bedroom was 73.2 dB(A). At this level, subjects woke up with a
probability of about 19% (laboratory: 47.4%).

The spontaneous awakening probability calculated with virtual noise events
was on average 8.7%. Considering the length of the noise window of
three epochs this equals 2.9% per epoch. With 3.1% per epoch, the prob-
ability of spontaneous awakenings was a little higher in the laboratory stud-
ies than in the field studies (6.2% with a noise window length of two ep-
ochs).

In contrast to the laboratory studies, a threshold of about 35 dB(A) for
awakenings induced by aircraft noise was found in the field studies using
the model shown in Figure 5.7. Therefore, only ANEs with maximum SPLs
above 35 dB(A) lead to awakening probabilities higher than the spontane-
ous awakening probability. This threshold lay about 8 dB(A) above the
background noise level which was set to the median of 27.1 dB(A).

The regression model shown in Figure 5.8 contains the variables "sleep sta-
ge prior to the beginning of the ANE" (PriorSleepStage) and "elapsed time
after sleep onset" (ElapsedTime) as independent variables beside maximum
SPL, background noise level one minute prior to the beginning of the ANE
and their interaction term.




                                     50
                         100%

                         90%

                         80%

                         70%
   Percentage Awakened




                         60%

                         50%

                         40%

                         30%

                         20%

                         10%

                          0%
                                25   30   35        40      45        50     55      60         65   70   75
                                               Maximum SPL LAS,max in dB at the Sleeper's Ear



 Figure 5.8: Field studies. Random effects logistic regression with 95% confidence
 limits. Model contains maximum SPL, background noise level LAS,eq 60 sec prior to ANE
 (= median 27.1 dB), their interaction, sleep stage prior to ANE (= stage S2) and
 elapsed sleep time (= 601 epochs) as independent variables. Based on 61 subjects and
 10,658 ANEs.


The values of the additional independent variables and the scale of the y-
axis were chosen in a way that allows a direct comparison of the model
based on data of the field studies shown in Figure 5.8 and the model based
on data of the laboratory studies shown in Figure 5.6. Because of preven-
tive reasons, PriorSleepStage was set to the most sensitive sleep stage S2.
ElapsedTime was set to 601 epochs (about five hours) corresponding to the
middle of the more sensitive second half of the night in the field studies.
Hence, the regression line of Figure 5.8 is shifted to higher awakening
probabilities compared to Figure 5.7.

The precision of the prediction can be read off the width of the 95% confi-
dence interval (grey lines in Figure 5.8). Because of the large sample size
the precision of the prediction is very high. It varies between 3.1% at
39 dB(A) and 10.5% at 73.2 dB(A).



                                                                 51
Beside the maximum SPL of the ANE the following variables were checked
first in univariable and then in multivariable models for their influence on
noise-induced awakenings:

Acoustical variables: The background noise level, defined as the LAS,eq one
minute prior to the beginning of the ANE, had a significant impact on the
awakening probability. This impact was less pronounced in ANEs with high
maximum SPLs compared to ANEs with low maximum SPLs, i.e. a significant
interaction was found. Short ANEs with steeply rising noise levels (dB/sec)
were associated with higher awakening probabilities than longer ANEs. The
habituation to ANEs within a single night, as observed in the laboratory
studies, was less pronounced in the field studies and statistically non-
significant.

Situative variables: The sleep stage prior to the beginning of an ANE was a
strong moderator. In contrast to the laboratory studies, awakening prob-
ability decreased in the order REM        S2        S3      S4. As the probability of
spontaneous awakenings during REM-sleep was almost as high as during
ANEs, the order of noise induced awakenings (PANE - Pspontaneous) was inverted
and decreased in the order S2        S3        S4        REM. Therefore, light sleep
stage S2 was most sensitive for noise induced awakenings. In contrast to
the laboratory studies, the number of the investigated night had no impact
on awakening probability. Temperature and humidity in the bedroom had
no impact on awakening probability either. As observed in the laboratory
studies, the awakening probability increased with elapsed sleep time ac-
cording to the depletion of sleep debt during the course of the night.

Individual variables: In contrast to the laboratory studies, no significant in-
fluence of age or gender on noise induced awakening probabilities was
found in the field studies. The subjective assessment of aircraft noise an-
noyance, global noise sensitivity or fatigue prior to going to bed did not
show any impact either and were statistically not significant.




                                      52
5.4.5                      Comparison of the results of laboratory and field studies


Figure 5.9 compares the dose-response relationships found in the labora-
tory and the field studies. Only probabilities of noise induced awakenings
are shown, i.e. the probabilities of spontaneous awakenings have already
been subtracted. Only in this way a comparison of both dose-response rela-
tionships is possible, as noise windows of different length were used in or-
der to maximize noise induced awakening probabilities. Dose-response rela-
tionships are shown only over the range of observed values, i.e. between
45 and 80 dB(A) in the laboratory and up to 73.2 dB(A) in the field.

                         70%
                                         Field Studies Noise Induced
                         60%
                                         Laboratory Studies Noise Induced
   Percentage Awakened




                         50%

                         40%

                         30%

                         20%

                         10%

                         0%
                               30   35      40       45      50        55   60   65   70    75   80   85
                                               Maximum SPL LAS,max in dB at the Sleeper's Ear



 Figure 5.9: Random effects logistic regression. Comparison of the results of laboratory
 and field studies. Awakenings attributable to aircraft noise are shown (noise induced
 awakenings). For a detailed description of the models see text.


Both models are multivariable and contain maximum SPL, PriorSleepStage
and ElapsedTime. The field studies' model also contains the background
noise level LAS,eq one minute prior to the beginning of the ANE and the in-
teraction term of maximum SPL and background level. As in Figure 5.7 and
Figure 5.8, a constant background noise level of 27.1 dB (median) was as-


                                                                       53
sumed. With the addition of PriorSleepStage and ElapsedTime the threshold
for noise induced awakenings described above is shifted from about
35 dB(A) to about 32.7 dB(A). The model of the laboratory studies addi-
tionally contains maximum SPL2 as an independent variable. Because of pre-
ventive reasons, in both models PriorSleepStage was set to the most sensi-
tive sleep stage S2. ElapsedTime was set to 601 epochs (about five hours)
corresponding to the middle of the more sensitive second half of the night
in the field studies.

The differences between laboratory and field studies can be observed over
the total range of the maximum SPL. The differences in awakening prob-
abilities simultaneously increase with increasing SPL.

The relatively low probability of noise induced awakenings in the field com-
pared to the laboratory was reported by Pearsons [20] as the result of a
meta-analysis of several studies concerned with traffic noise and sleep. The
awakening probabilities were even lower than those observed in our field
studies.

A possible reason for the low awakening probabilities observed in the field
studies is the fact that subjects are observed in the familiar environment
of their own bedroom and their own bed. Hume und Whitehead [12]
showed in a study where ANEs were presented to subjects via loudspeakers
in their own homes that awakening probabilities were lower compared to
those usually observed in laboratory studies. These awakening probabilities
were still higher than those reported by Pearsons [20], which are mostly
based on field studies with real life noise events. Therefore, not only the
familiar environment, but also a habituation to the specific noise scenario at
home seems to play an important role.

These findings are confirmed by a sub-analysis of 20 subjects participating
both in our laboratory and in our field studies. As some of these subjects
had been exposed to aircraft noise in their homes for many years, one may
expect that these subjects woke up less frequently when confronted with


                                      54
aircraft noise than other subjects, who usually had not been exposed to air-
craft noise in their homes. As a matter of fact, the subjects participating in
both laboratory and field studies exhibited an awakening probability that
was even a little higher than that one of the other subjects who were only
investigated in the laboratory. Beside the unfamiliar environment, it is pos-
sible that the ANEs played back in the laboratory studies differed from the
specific acoustical situations experienced at home, which may vary consid-
erably depending on the distance from the actual flight paths. Therefore,
residents living in the vicinity of major airports do not generally seem to ha-
bituate to aircraft noise, but rather to the specific acoustical situation ex-
perienced at home.

There is another possible explanation for the differences in awakening
probabilities observed in the laboratory and in the field: An ANE that causes
an awakening in the laboratory may not cause an awakening in the field
because of habituation processes, but it may nevertheless lead to more sub-
tle changes in sleep structure, like transitions from sleep stage S4 to S2,
and therefore may still have an impact on sleep. This possibility was also
checked with the data of our studies. However, the analysis of sleep stage
changes showed differences between laboratory and field studies only in
the same range as the analysis of changes to sleep stage S1 and Awake
alone.


5.5      Awakening duration


Both the number and the duration of awakenings play an important role
for the critical appraisal of the impact of nocturnal aircraft noise on sleep.
As described in chapter 5.4.1, long awakenings are considered especially
harmful as they may be remembered on the next day and therefore may in-
fluence the subjective assessment of sleep quality and quantity. Hence, it is
very important to investigate whether the duration of awakenings is also
increased by ANEs, as was already shown for the frequency of awakenings.



                                      55
Thus, the duration of spontaneous awakenings was compared to the dura-
tion of awakenings induced by aircraft noise in the laboratory studies. The
result of this analysis is shown in Figure 5.10. ANEs with low exposure lev-
els (45 to 60 dB) were differentiated from ANEs with high exposure levels
(65 to 80 dB).

                                                                                                         2


                                                                                                         0
                                                   (Compared to Subjects with Spontaneous Awakenings)




                                                                                                              0 min   0.5 min   1 min   1.5 min   2 min    2.5 min   3 min   3.5 min   4 min   4.5 min   5 min
   Number of Subjects Having Fallen Asleep Again




                                                                                                                        (53)     (73)     (81)     (86)      (89)     (91)     (92)     (93)     (94)     (95)


                                                                                                         -2


                                                                                                         -4


                                                                                                         -6


                                                                                                         -8


                                                                                                        -10


                                                                                                        -12                                                              Maximum SPL 45 - 60 dB (n=1850)

                                                                                                                                                                        Maximum SPL 65 - 80 dB (n=2286)

                                                                                                        -14
                                                                                                                                       Elapsed Sleep Time after the Awakening
                                                                                                                       (% of Subjects Having Fallen Asleep Again after Spontaneous Awakening)



 Figure 5.10: Duration of awakening depending on maximum SPL (laboratory).


First of all, Figure 5.10 shows that after a spontaneous awakening 53% of
the subjects have already fallen asleep again after 30 sec. After three min-
utes, more than 90%, and after five minutes even more than 95% have al-
ready fallen asleep again. Research in the past has shown that awakenings
have to last for about four minutes before the waking consciousness is acti-
vated [4]. Therefore, only about every tenth spontaneous awakening will be
remembered in the next morning.

The duration of awakenings caused by ANEs with low exposure levels rang-
ing from 45 to 60 dB(A) was only very moderately increased compared to
spontaneous awakenings: After 30 sec, 4% less of the subjects had fallen


                                                                                                                                                      56
asleep again, i.e. 49% compared to 53%. After 1.5 min no difference was
observed. Hence, ANEs with low exposure levels between 45 and 60 dB(A)
are unlikely to induce awakenings that are remembered on the next day.

The differences in awakening duration are more prominent if spontaneous
awakenings and those induced by ANEs with higher exposure levels rang-
ing from 65 to 80 dB(A) are compared. Here, after 30 sec, 12% less of the
subjects had fallen asleep again, i.e. 41% compared to 53%. Even after
four minutes, 2% of the subjects were still awake that had already fallen
asleep again in the spontaneous situation. Because of the duration of this
prolonged awakening it may be remembered on the next day. To put it dif-
ferently, every 50th awakening induced by ANEs with exposure levels
greater than or equal to 65 dB(A) will most probably be remembered on
the next day.

To summarize, the duration of spontaneous awakenings and awakenings
induced by aircraft noise differs, as shown in Figure 5.10, and depends on
the maximum SPL of the ANE.


5.6       Falling asleep again


The analyses described in chapter 5.4 only refer to situations where the
subject is sleeping before the beginning of the ANE, i.e. where PriorSleep-
Stage equals S2, S3, S4 or REM. As mentioned above, subjects were in
sleep stage S1 or awake prior to the beginning of the ANE in 2,857 cases
(laboratory) and 1,954 cases (field) respectively.

In these situations ANEs may negatively affect sleep by disturbing the proc-
ess of falling asleep again. Situations where the subject has regained wak-
ing consciousness before the beginning of the ANE may additionally cause
strong annoyance reactions. Hence, residents living in the vicinity of a major
airport very often complain about ANEs in the early hours of the morning.
The decreased sleep debt at this time of the night not only causes lower



                                      57
awakening thresholds, but it may also lead to problems in falling asleep
again, especially if ANEs prevent them from doing so.

The analysis of the time needed to fall asleep again is shown in Figure 5.11
for the laboratory studies. Obviously, after a spontaneous awakening in the
second half of the night, subjects fall asleep again later than in the first half
of the night: In the first half of the night, after five minutes about 75%
(grey line) have already fallen asleep again, whereas in the second half of
the night only 63% (pink line) have fallen asleep again.

                                                      80%


                                                      70%
  Percentage of Subjects Having Fallen Asleep Again




                                                      60%


                                                      50%


                                                      40%


                                                      30%

                                                                                                        Exposure Nights First Half (n=852)
                                                      20%
                                                                                                        Baseline Nights First Half (n=764)
                                                                                                        Noise Nights Second Half (n=1261)
                                                      10%
                                                                                                        Baseline Nights Second Half (n=1217)


                                                      0%
                                                            0.5   1       1.5      2       2.5      3        3.5        4         4.5        5
                                                                      Elapsed Minutes after the Beginning of the Noise Event



 Figure 5.11: Laboratory Studies. Percentage of subjects having fallen asleep again
 depending on the half of the night in baseline and exposure nights.


ANEs in the first half of the night were associated with a decrease of the
fraction of subjects that had already fallen asleep again: After 30 sec only
19% had fallen asleep again compared to 29% of noise-free baseline
nights. After 2.5 min these fractions did not differ anymore between expo-
sure and baseline nights. After three minutes, the fraction of subjects hav-



                                                                                            58
ing already fallen asleep again was even a little higher in the exposure
compared to the baseline nights, which may be due to a sleep debt accu-
mulated in preceding nights or in the preceding hours of the same night.

These differences between baseline and exposure nights can be seen in the
second half of the night as well. Here, in contrast to the first half of the
night, after 3.5 min the same fraction of subjects had fallen asleep again
under both conditions.

The analysis shows that the probability of a conscious perception of an ANE
is higher in the second than in the first half of the night, as the duration of
spontaneous awakenings is longer in the second compared to the first half.
ANEs encountering an organism being awake or in sleep stage S1 in the
laboratory studies lead to a delay in the process of falling asleep again.
However, with average durations of 2.5 min (first half of the night) and
3.5 min (second half of the night) only minor delays were observed.


5.7      Importance of the results for the discussion of aircraft noise
         protection criteria


Number above threshold (NAT)- and Leq-criteria dominate the suggestions
for the amendment of the aircraft noise protection law in Germany. An ex-
haustive discussion of advantages and disadvantages of the different crite-
ria is not intended here, but major problems will briefly be discussed.

NAT-criteria assume that there are only little if any effects of aircraft noise
below the NAT-value. In the field studies it was shown, that a threshold of
awakenings induced by nocturnal aircraft noise exists at a maximum SPL of
about 33 dB(A), i.e. only below maximum SPLs of 33 dB(A), noise induced
awakenings are no longer expected. The NAT-values currently discussed are
situated far above this threshold, and hence awakenings induced by ANEs
below the NAT-value but above the threshold of 33 dB(A) will only be ac-
counted for by the Leq. But the so-called absorbing criteria based on the Leq



                                      59
are not able to fully compensate for this reactions below the NAT-value, as
the Leq itself is dominated by ANEs with high maximum SPLs.

A reduction of the number of ANEs with the same maximum SPL by 50% is
usually associated with a simultaneous reduction of the Leq(3) by 3 dB. The-
refore, criteria solely based on the Leq(3) implicitly assume that there is not
only equivalence of energy but also equivalence of effects, i.e. that a reduc-
tion of the Leq(3) by 3 dB also leads to a reduction of the effects associated
with the ANE, e.g. a reduction of the number of awakenings induced by
aircraft noise by 50%.

Instead, Figure 5.12 shows that there is no equivalence of effects. Analo-
gous to the term numbers needed to harm, which is often used in Epide-
miology, the number of ANEs needed to induce one additional awakening
depending on the maximum SPL is shown in Figure 5.12.

                               72       10.6

                               69       11.8

                               66        13.2
   Maximum SPL LAS,max in dB




                               63         14.9

                               60          17.0

                               57              19.6

                               54                22.9

                               51                  27.4

                               48                       33.7

                               45                              43.1

                               42                                     58.5

                               39                                              89.0

                               36                                                                           175.7

                                    0                           50                100                 150
                                                          Number of ANEs Causing One Additional Awakening



 Figure 5.12: Number of ANEs causing one additional awakening depending on
 maximum SPL. Results are based on the dose-response relationship found in the field
 studies.




                                                                             60
For instance, if the maximum SPL is reduced from 72 to 69 dB(A), the num-
ber of ANEs may increase by only 11% from on average 10.6 to 11.8, not
by 100%, in order to receive the same effects. The percentage increase of
the number of ANEs going along with a reduction of the maximum SPL by
3 dB(A) continuously rises from 11% (reduction of maximum SPL from 72
to 69 dB) to about 97% (reduction of maximum SPL from 39 to 36 dB), i.e.
only at maximum SPL values very close to the threshold found in this study
equivalence of effects regarding awakenings induced by ANEs can be ob-
served.


5.8       Transfer of the results to a German airport


According to the aircraft noise protection law of 1971, which is still in ef-
fect in 2004, noise immission values are calculated for every location in the
vicinity of major airports. These calculations are based on the number of
ANEs, the mix of different types of aircrafts and the six months of a year
with the highest amount of traffic. Until today, noise protected areas are
solely based on acoustical parameters, e.g. regions are declared where the
Leq(4) exceeds 75 dB(A).

The dose-response relationship established in this study can be utilized to
additionally define noise effects criteria beside acoustical criteria. Based on
the dose-response relationship, fractions of the population, that addition-
ally wake up exactly one, two or three times due to nocturnal aircraft noise,
can be calculated for every location in the vicinity of airports, if the neces-
sary acoustical information is available. This prognosis based on noise ef-
fects may serve as an important aid for enhancing the protection of resi-
dents living in the vicinity of airports. It is based on the data of the field
studies as only these data represent the realistic experience of people living
near airports.

To illustrate the potential of this noise effects prognosis, Figure 5.13 exem-
plarily compares two areas of Frankfurt Airport based on Leq(3)-criteria with


                                      61
regions outside of which less than one, two or three additional awakenings
caused by nocturnal aircraft noise are expected on average. The chosen
values of one, two and three additional awakenings were exemplarily cho-
sen and are not to be confused with recommendations for legislative bod-
ies.
 5556


                                                                         Frankfurt am Main

  52
                                                                                                     Offenbach
                                                                                                                    Obertshausen
  48
         Erbenheim                           Kelsterbach
                                                                                 Neu-Isenburg         Heusenstamm

                               Eddersheim                      Zeppelinheim
  44
                       Floersheim
                                 Raunheim                                            Dreieich
                                                           Walldorf
  40
                      Ruesselsheim                                            Langen
                                                    Moerfelden
                                                                           Egelsbach
  36
                               Nauheim

                                                 Worfelden
  32
                                 Gross-Gerau
                                                           Weiterstadt
                                             Buettelborn

  28

                                                                      Darmstadt

  24



  20



 5516
         8
       344       52   56       60           64        68              72        76              80      84          88       3492




 Figure 5.13: Prognosis of noise effects for Frankfurt Airport: on average one, two or
 three additional awakenings caused by aircraft noise (red lines), Leq(3) 55 dB (blue),
 Leq(3) 50 dB (green).


Apparent qualitative differences are obvious: The areas delimited by the red
lines are, on the one hand, noticeably larger in regions with comparatively
low maximum SPLs (landing planes), but, on the other hand, noticeably
smaller in regions with comparatively high maximum SPLs (starting planes).




                                                             62
6        Psychological effects

         Julia Quehl



No significant dose-response relationships have been found with respect to
fatigue (Fatigue Checklist: FAT), mood (Multidimensional Mood Question-
naire: MDBF) nor recreation and load processes (Strain and Recreation
Questionnaire: EBF).

Regarding noise annoyance, representing statistically the most salient psy-
chological noise effect, significant dose-response relationships were derived
by means of logistic regression analyses with random effects for both labo-
ratory and field studies. Unlike the common consideration of evaluations on
the upper 25-30% of the rating scale constituting the group of highly an-
noyed persons according to Schultz [28] categories 3 to 5 of the original
scale were combined into a single category indicating the presence of an-
noyance. Categories 1 and 2 denoted no annoyance. This procedure was
reasonable since merely 20% of all laboratory annoyance ratings and only
4% of all field assessments distributed on the grades 4 and 5 (Figure 6.1).
Moreover, a limitation to the upper range of the response scale would have
had the drawback that subjects only moderately bothered were ignored.




                                     63
               100

               90

               80

               70

               60         65
     Percent




               50

               40

               30
                     29        30
                                                                     Setting
               20
                                    22   20
               10                                                        Laboratory
                                                    13
                                              9
                0                                            7           Field
                      not       little   moderate   rather   very


                               Aircraft noise annoyance

 Figure 6.1: Percentage distribution of annoyance ratings (question: „How much have
 you been annoyed by the aircraft noise of the previous night?“) of the experimental
 laboratory (N = 112) and field samples (N = 64).


On the one hand, LAS,eq [2] and LAS,max combined with number of aircraft
noise events on the other, [6,27] are considered as valid physical indicators
of aircraft noise annoyance. Thus, two laboratory specific logistic regression
models were developed for these physical parameters. Furthermore, two
field related models were derived for number of nocturnal flight events as
well as LAS,eq. Aircraft noise specific (judged health risk, avoidability and ne-
cessity of air traffic, general attitude towards air traffic, health risk of air-
craft noise) and personal (pre-annoyance due to aircraft noise, noise sensi-
tivity, habituation to aircraft noise, age, gender) moderator variables were
integrated into all models and tested for significance.

According to the laboratory regression models L1 (number of aircraft noise
events and LAS,max) and L2 (LAS,eq) gender (significantly increased female pro-


                                              64
portion of aircraft noise annoyed persons), pre-annoyance (the more pre-
annoyed, the more bothered persons) and rated necessity of air traffic (the
smaller necessity of air traffic, the more annoyed persons) represent signifi-
cant psychological variables. Moreover, L1 indicates a significant influence
of LAS,max and number of flight events on aircraft noise annoyance (Figure
6.2). The degree of annoyance rises linearly with increasing LAS,max as well as
with the number of nocturnal flights. This finding is in accordance with
previous field studies on aircraft noise annoyance [10,23-25]. 128 times
55 dB(A) flights induce the greatest percentage of bothered individuals
(about 80%). Unlike recent field studies the part of annoyed subjects also
rises significantly with increasing LAS,max at few air traffic activities (<16
flights per night). L2 predicts a significant increment of bothered persons
with growing LAS,eq (from 15% at 30.6 dB(A) up to 70% at 46.6 dB(A))
(without illustration).

                                                                 100%
     Group of aircraft noise annoyed persons (categories >= 3)




                                                                 90%

                                                                 80%

                                                                 70%

                                                                 60%

                                                                 50%

                                                                 40%
                                                                                                                number of events 4
                                                                                                                number of events 8
                                                                 30%
                                                                                                                number of events 16
                                                                 20%                                            number of events 32
                                                                                                                number of events 64
                                                                 10%                                            number of events 128


                                                                  0%
                                                                        45   50   55   60             65   70            75            80
                                                                                            LAS,max



 Figure 6.2: Percentage of aircraft noise annoyed persons (categories >=3) predicted by
 regression model L1 depending on LAS,max and number of aircraft noise events.


According to the field specific regression models F1 (number of aircraft
noise events) and F2 (LAS,eq) the habituation to aircraft noise (the smaller the


                                                                                       65
habituation, the more annoyed persons) is a significant psychological mod-
erator. Additionally, F1 takes the age into account (the older the more
bothered persons). F1 indicates a significant effect of number of air traffic
activities on aircraft noise annoyance (without illustration). The proportion
of annoyed individuals grows with the increment of number of nocturnal
flights. 128 aircraft noise events per night induce the highest percentage of
bothered subjects (about 55%). F2 predicts a significant increase of an-
noyed persons with growing LAS,eq, achieving the maximum of 30% at
47.3 dB(A) (Figure 6.3).


                                                  100%
                                                              Laboratory
                                                  90%
                                                              Field
   Percentage of aircraft noise annoyed persons
     (categories >=3) +/- 95% confidence limits




                                                  80%

                                                  70%

                                                  60%

                                                  50%

                                                  40%

                                                  30%

                                                  20%

                                                  10%

                                                   0%
                                                         30             35                       40       45
                                                                      LAS,eq in dB at the sleeper's ear


 Figure 6.3: Percentage of aircraft noise annoyed persons (categories >=3) predicted by
 laboratory and field specific regression models L2 and F2 depending on LAS,eq.


For common ranges of LAS,eq the laboratory-specific dose-response curve lies
clearly over the field curve (up to 40% difference) (Figure 6.3). At
30.6 dB(A) the predicted part of aircraft noise bothered individuals is still
approximately identical in both settings (12% to 15%). At 47.3 dB(A) it is



                                                                               66
30% in the field versus 70% in the laboratory, i.e., in the laboratory sub-
jects have been significantly more annoyed by nocturnal aircraft noise than
at home. However, one has to bear in mind that laboratory and field re-
lated models integrate different psychological moderators (laboratory: gen-
der, pre-annoyance due to aircraft noise, rated necessity of air traffic; field:
habituation to aircraft noise).




                                      67
7         Stress hormones

          Hartmut Maaß



It is impossible to present all results of the stress hormones analyses in de-
tail here. We restrict ourselves to the most relevant aspects and refer to the
final report. The generally accepted stress model accepted in general as-
sumes a reaction chain. The stressor noise is perceived and processed cere-
brally, followed eventually by a secretion of hormones like catecholamines
(adrenalin, noradrenalin) or cortisol. These hormones may lead to electro-
lyte shifts on the sub cellular level especially changing magnesium and cal-
cium concentrations. Recently urine samples from all night collections were
taken for the analyses of stress hormones, and this method was adopted
for the current study.

Electrolytes: The nocturnal excretion rates of potassium, sodium, magne-
sium, and calcium were determined. A balanced food control, however,
was not ensured in the evenings. No connection with nocturnal aircraft
noise is detectable. There is a difference between excretion rates obtained
in the lab and in the field. All mean excretion rates of electrolytes under lab
conditions are increased.

Adrenalin: Adrenalin excretion rates in all night urines are unchanged with
night aircraft noise. They remain on extremely low levels. Under laboratory
conditions the adrenalin concentrations in more than 2/3 of all collected
urines are below detection limit (1 ng/ml), under field conditions in roughly
1/2 of all samples. There are no statistically relevant secretion rates that dif-
fer from those without aircraft noise.

Noradrenalin: Noradrenalin excretion in all night urine samples is statistically
constant and not influenced by nocturnal aircraft noise. There are no
changes depending on the equivalent noise level LAS,eq, nor on the maxi-



                                       68
mum sound pressure LAS,max, nor on the number of events. No difference is
observed between results taken from the lab and the field (see Figure 7.1).
Also, the number of investigated nights is irrelevant for a potential influ-
ence of noise on the excretion of noradrenalin.


                 Comparison LAB vs. FIELD Mean noradrenalin excretion in all night
                                         urine samples
            25
                  112    223   207    128    111         119   lab nights     94   69    32     16


                   140   135   111    76     55         23     field nights
            20


            15
   ng/min




                                                                                          LAB

                                                               FIELD
            10



             5



             0
                 <= 30 30-33 33-36 36-39 39-42          >42                 45-48 48-51 51-54   > 54
                                     LAS,eq (night, total, indoors) in dB




 Figure 7.1: Mean noradrenalin excretion +/- SD in urine collected all night under
 laboratory conditions (n = 112, no control groups, experimental groups only, bold line)
 and under field conditions (n = 64, bars) with classes of equivalent noise levels LAS,eq.


Cortisol: Unfortunately, during the laboratory studies a change of determi-
nation methods occurred. Thus, absolute results from this single study
phase are not immediately comparable. Under laboratory conditions the ex-
cretion rates of cortisol are influenced by noise. There is a significant trend
(Jonckheere test) depending on maximum sound pressure LAS,max and the
number of noise events. Also with increasing equivalent noise levels LAS,eq a
significant trend is shown. At the same time, however, a trend of increasing
cortisol excretion is detectable with the time of investigation in the lab
without any noise. These trends are not observed in the field. Since cortisol


                                                   69
excretion shows a distinct circadian rhythm, this property has to be taken
into account especially during the field studies. During the weekends with
very low air traffic and noise, cortisol excretions are considerably higher.
This is not due to increased stress, but rather much longer mean sleeping
times (mean waking-up time nights Saturday/Sunday is 07:47:46h, in com-
parison all nights considered with the weekends mean wake-up time
06:10:52h). In Figure 7.2 which compares the excretion of cortisol under
laboratory and field conditions only nights with latest wake-up times of
07:00h are included, in correspondence with our lab study design. How-
ever, the mean wake-up time in the field is approximately 50 minutes ear-
lier than in the lab. From this fact alone, due to the circadian rhythm of the
cortisol excretion, a difference between lab and field results is predictable.
The figure shows the mean excretion rates of cortisol from lab and field
studies obtained by identical determination method and exclusion of
nights, when subjects slept longer than 07:00h. Results show a significant
difference between laboratory and field, where results from the lab are in-
creased.




                                     70
                   Comparison LAB vs. FIELD Mean absolute cortisol excretion rates in
                                 night urine samples (0:00h-7:00h)

             120

                     87    205    155    95       79        94   lab nights      71    47         16   16
             100
                     57    82      78    45       37        15    field nights


              80
    ng/min




              60
                                                                                            LAB


              40
                                                                     FIELD


              20


               0
                     <30   30-33 33-36 36-39 39-42          >42                  45-48 48-51 51-54     >54
                                              LAS,eq (total, indoors) in dB



 Figure 7.2: Mean cortisol excretion +/- SD in urine collected all night under laboratory
 conditions (n = 88, only experimental groups with identical determination method, no
 control groups, bold line) und under field conditions (n = 64, wake-up time before
 07:00h, bars) with classes of equivalent noise levels LAS,eq.


Regarding the stress hormones and electrolytes the determination of cate-
cholamines from all night urine collections seems futile. There are neither
changes with various noise conditions, nor are there any differences be-
tween laboratory and field. Possibly a single measure for the entire night is
too insensitive to shed light on minute and short time excretions of cate-
cholamines during the complete nocturnal phase. From the literature, con-
tradictory results are reported.

Electrolyte determinations from all night urine samples are futile, unless a
controlled and balanced intake of food and beverages is ensured. Any pos-
sible mobilization of electrolytes by stress hormones under noise are cov-
ered by renal regulation of the electrolytes after food and beverage intake
during dinner and the evening.


                                                       71
Cortisol is a parameter that correlates by trend with noise, especially under
laboratory conditions. The influence of the endogenous circadian rhythm of
cortisol aggravates a concrete statement. If subjects wake-up much earlier
on weekdays, as they do in the field, lower excretion levels of cortisol and
no correlation with nocturnal noise are seen. The significant difference be-
tween laboratory and field results is partly founded on this fact. Addition-
ally, that high noise strain applied in the lab was not recorded under field
conditions. The trend of increasing cortisol levels with time of investigation
in absence of any noise has to be taken into account as well. An explana-
tion might be a prolonged sleeping time in the lab, in contrary to the sub-
jects' home, and the consecutively earlier onset of endogenous secretion in
the morning. To obtain more reliable data several samples taken during the
night are indicated. For the current study, however, special attention was
focused on sleep without additional interference of investigators, and ex-
clusively disturbed by noise events, if at all.




                                        72
8        Performance

         Ernst-Wilhelm Müller und Jürgen Wenzel



All investigated performance parameters show a prominent difference be-
tween morning and evening sessions due to circadian rhythmicity; gener-
ally, performance was better in the evening, corresponding to the individual
ability of the different subjects. Therefore, morning and evening sessions
have to be considered separately. The following analysis concentrates on
the morning results which had a higher variation through the series of tri-
als, and which also should reflect the impact of last night's sleep most di-
rectly. However, for the field data the weekend sessions had to be excluded
because longer bed times and later awakenings led to pronounced cir-
cadian differences to regular work day sessions.

A basal estimate of mental performance disposition is delivered by Single
Reaction Task. In this test a stimulus appears on the screen in random inter-
vals, and the subject has to respond via keyboard as fast as possible; reac-
tion time is measured and logged. In the special design (Dinges [8]) applied
in this study, the stimulus is a running digital stop watch freezing to the re-
action time achieved after the subject's response, displayed for 3 seconds.
After another random interval with dark screen, the next stop watch run
appears. Thus, the subject gets a feedback to the efficiency of each single
response, conserving motivation for the next run, necessary for the high
number of test repetitions in the study design. Primary mental components
for a good test result are alertness and vigilance, information processing at
higher centers is not necessary.

Figure 8.1 shows the median reaction time in milliseconds of all subject
morning sessions arranged with rising noise exposure, separated into labo-
ratory and field data. The grand total of approximately 240 ± 5 ms shows
considerable inter-individual variances due to the rather inhomogeneous


                                      73
subject population with respect to gender and age, leading to the rather
large confidence interval. However, there is neither any obvious effect of
the amount of nocturnal noise exposure nor any statistical significance, all
median values are placed within the collective confidence range.



                                  270
                                           Field                                                    89
                                                                                                              85
                                  260      Laboratory                                    137
                                                 8                            111
   Reaction time in ms (median)




                                  250
                                                                                                      143       111      150        96
                                                                    65                      191                                            117
                                                                                112
                                                          49          191
                                  240


                                  230


                                  220


                                  210


                                  200
                                        < 21   21 < 24   24 < 27   27 < 30   30 < 33     33 < 36   36 < 39   39 < 42   42 < 45   45 < 48   >48

                                                                   LAS,eq inside at the sleeper's ear in dB




 Figure 8.1: Reaction time in the Single Reaction Task (median and 95% confidence
 interval) as a function of noise exposure (LAS,eq at 3 dB steps). 112 laboratory subjects
 (9 nights with noise exposure), 64 field subjects (8 nights with noise).


Another test of some elementary components of mental performance can
be seen in Unstable Tracking. A vertical cursor bar appears in the center of
screen, shifted to the edges of screen under program control, with increas-
ing speed as the distance from center position gets larger. The subject has
to keep the cursor as close as possible to the center position by use of a
joystick. The distance to center is recorded with appropriate time resolu-
tion, additionally loss of control is stored each time the cursor reaches edge
of screen. This task investigates hand-eye-coordination as well as dexterity
and stands for a typical operator's task in a technical environment.


                                                                                    74
The UTT results are shown in Figure 8.2 in a format similar to those of SRT
(Figure 8.1); however, instead of reaction time the cursor's median root
mean square displacements from center position is used as performance
marker (normalization of left-right displacements). Like in SRT, no noise ef-
fect is noticed, all medians of about 8 mm displacement are positioned in-
side the confidence range of about ± 0.5 mm.



                                                10
                                                            Field
    Deviation from the central position in mm




                                                            Laboratory                                                  112
                                                                                                  137 192          85
                                                 9                                                                                  96   117
                                                                                       110                   143              151
                                                                              65 192
                                                               8                        112                 89
                                                                       49
                                                 8
                     (median)




                                                 7



                                                 6



                                                 5
                                                     < 21     21 < 24 24 < 27 27 < 30 30 < 33 33 < 36 36 < 39 39 < 42 42 < 45 45 < 48    >48

                                                                            LAS,eq inside at the sleeper's ear in dB




 Figure 8.2: Displacement (RMS-value) in the Unstable Tracking Task (median and 95%
 confidence interval) as a function of noise exposure (LAS,eq in 3 dB steps). 112
 laboratory subjects (9 nights with noise exposure), 64 field subjects (8 nights with
 noise).


By a comparison of symbols Memory Search Task evaluates a higher mental
function, making use of the choice response design. Contrary to SRT, a
stimulus response with one of two alternate keys is preceded by a mental
processing of the given item. Corresponding to the Sternberg paradigm a
group of letters (variable number) is displayed prior to the test run, the sub-
ject has to memorize them. In this study, group sizes of 4 (MS4) and 6


                                                                                             75
(MS6) letters were used. After starting the test run, one letter is displayed
on the screen, the subject has to respond by one of the two alternate keys
whether the letter belongs to the memorized group or not. After response,
the next letter is displayed. Like in SRT, the task comprises the notice of
stimulus plus motoric response, additionally a memory operation plus right-
wrong decision has to be performed. The response time increases with
growing memory group size; according to Sternberg, this increase is pro-
portional to the number of letters in the group. The task evaluates func-
tions of working and short term memory, resulting into a reaction time
which is considerably longer than in SRT, additionally the error rate can be
used, possibly differentiating in false-negative or false-positive reactions.

Like in SRT and UTT there is no prolongation of reaction times due to noc-
turnal noise exposure, mean reaction times are about 500 ms from MS4,
about 60 ms longer from MS6 because the sequential memory scan for 6
letters needs more time. The confidence range is about ± 20 - 30 ms.

Thus, results of performance tests do not show any relevant decrements af-
ter nocturnal noise exposure. Results of sleep investigations (see chapter 5)
show a decrease of total sleep time of less than 2 min. Under noise, this is
not sufficient to expect any significant effect on next day's performance.
However, the test battery used here did show relevant performance decre-
ments in a comparative study correlating partial sleep deprivation with
other stressors: Sleep deprivation had to be 3 hours per night over several
consecutive days to produce significant effects.

The average influence of nocturnal noise exposure on performance as pre-
sented here shows little or no effects because the average sleep distur-
bance is rather low. Therefore, it could be useful to concentrate on those
subjects with a significant reduction of sleep quality and time and possibly a
higher impact on day-time performance; likewise, it remains to be analyzed
if the number of noise-induced wake-up reactions has a stronger influence




                                       76
on performance parameters than the mere arithmetic summation of sleep
reduction.




                                 77
9         Summary

          Alexander Samel und Mathias Basner



In the frame of the DLR/HGF-project "Leiser Flugverkehr" ("Quiet Air Traf-
fic") 192 subjects with healthy sleep behavior, aged between 18 and 65
years, were investigated in laboratory and field studies during 2240 nights
in total. The effect of aircraft noise was examined on sleep (by polysomno-
graphy), on the excretion rates of stress hormones (by the assessment of
adrenalin, noradrenalin, and cortisol) and electrolytes from urine collected
all night, on performance (computer tests in the evenings and mornings),
and on annoyance, fatigue, stress, mood, and recuperation (by subjective
ratings in the evenings and mornings). Awakenings caused by aircraft noise
events were monitored by the simultaneous recordings of electrophysio-
logical and acoustical signals.

In the field studies (in the vicinity of Cologne-Bonn airport) on average 41
aircraft noise events per investigated night were detected. These events
showed out-door maximum levels (LAS,max) between 35 dB(A) and 87 dB(A),
corresponding to a median maximum value of 64 dB(A). Values inside the
bedroom varied between 20 dB(A) and 73 dB(A), leading to a median
maximum sound pressure level of 44 dB(A). The median equivalent noise
levels LAS,eq(3), calculated from aircraft noise events between midnight and
6:00 am, resulted in 53.9 dB(A) outside and 36.2 dB(A) inside (at the
sleeper's ear). Depending on the window position, the difference between
outside and inside levels was measured as 28 dB(A) with closed windows,
18 dB(A) with tilted windows, and 13.5 dB(A) with fully open windows.

With respect to the performance tests conducted in the mornings, it was
not yet possible by statistical analyses to establish dose-effect relationships
between acoustical data and performance test results.




                                      78
With regard to subjective ratings of fatigue, mood, stress and recuperation,
again, no dose-effect relationships were detected. However, a significant
dose-effect relationship was found between the equivalent noise level
LAS,eq(3) and the portion of subjects who felt moderately to highly annoyed
by aircraft noise. The amount of annoyance in the field investigations was
significantly and substantially lower than that observed in the laboratory,
when related to similar equivalent noise levels.

The excretion rates of the stress hormones adrenalin and noradrenalin did
not exhibit any significant differences between no-noise and noise condi-
tions. With respect to cortisol excretions, only under laboratory conditions a
significant trend (with exposure to elevated noise levels) was demonstrated;
however, this was not the case under field conditions.

Total sleep time in the laboratory studies was non-significantly shortened by
about two minutes, when comparing control nights (without any aircraft
noise) to noise nights. Simultaneously, the structure of sleep was modified
by aircraft noise: the fraction of slow-wave-sleep (deep sleep) was reduced
in favor of light sleep stages, but again without statistical significance.

In the field, a threshold for awakenings was found: subjects started to have
additional, polysomnographically detected awakenings caused by aircraft
noise at a threshold of about 33 dB(A), monitored at the sleeper's ear. In
the laboratory, beyond 45 dB(A) LAS,max no threshold was found. As for an-
noyance, the observed effects were significantly and substantially lower
under field conditions than under laboratory conditions.

For the first time, highly precise dose-effect curves for awakenings induced
by nocturnal aircraft noise depending on the number and the maximum
sound pressure level LAS,max of aircraft noise events were developed on the
basis of polysomnography. These dose-effect curves are based on more
than 30,000 aircraft noise events in the laboratory and more than 15,000
aircraft noise events in the field. In both situations, acoustical and electro-
physiological data were sampled synchronously.


                                       79
Furthermore, making use of this huge database, a prognostic model for the
prediction of aircraft noise induced awakening probabilities was developed,
that can be applied to any airport with known acoustical conditions in its
vicinity. In combination with acoustical prognoses, the dose-effect curves
for electrophysiologically detectable awakenings developed in this study
permit predictions at each location in the vicinity of an airport with a very
high degree of accuracy, how often (none, once, twice, …) and what per-
centage of the population affected will be additionally awakened by aircraft
noise.

Until now, acoustical criteria (e.g. equivalent noise level, number of aircraft
noise events above an acoustical threshold) are exclusively used to assess
the load of nocturnal air traffic. Using the results and findings of this DLR-
investigation, further criteria, directed to the effects on sleep, can be calcu-
lated (e.g. probability for one, two, three … additional awakenings) in rela-
tion     to   the     actual     amount      of     nocturnal     air    traffic.




                                      80
10       Literature

[1]   DIN 45643: Messung und Beurteilung von Flugzeuggeräuschen
      (Teil 3: Ermittlung des Beurteilungspegels für Fluglärmimmissionen).
      (1984): 1-3
[2]   Aasvang G.M., Engdahl B. (1999): Aircraft noise in recreational areas:
      A quasi-experimental field study on individual annoyance responses
      and dose-response relationships. Noise Control Eng.J. 47(4): 158-162
[3]   American Sleep Disorders Asociation (ASDA) (1992): EEG arousals:
      scoring rules and examples: a preliminary report from the Sleep Dis-
      orders Atlas Task Force of the American Sleep Disorders Association.
      Sleep 15(2): 173-184
[4]   Baekeland, F., Hoy, P. (1971): Reported vs recorded sleep characteris-
      tics. Archives of General Psychiatry 24(6): 548-551
[5]   Basner, M., Buess, H., Luks, N., Maaß, H. et al. (2001): Nachtflug-
      lärmwirkungen - eine Teilauswertung von 64 Versuchspersonen in
      832 Schlaflabornächten. DLR FB 2001-26, Köln, Germany
[6]   Bjorkman, M., Ahrlin, U., Rylander, R. (1992): Aircraft noise annoy-
      ance and average versus maximum noise levels. Archives of Environ-
      mental Health 47(5): 326-329
[7]   Bonnet, M.H. (1986): Performance and sleepiness following moderate
      sleep disruption and slow wave sleep deprivation. Physiol Behav.
      37(6): 915-918
[8]   Dinges, D.F., Powell, J.W. (1985): Microcomputer analysis of per-
      formance on a portable, simple visual RT task during sustained opera-
      tions. Behavior Research Methods, Instruments & Computers 17: 652-
      655
[9]   Fahrenberg, J., Hampel, R., Selg, H. (1994): Das Freiburger Persön-
      lichkeitsinventar FPI. Handanweisung. 6. Auflage
[10] Fields, J.M. (1984): The effect of numbers of noise events on people's
     reactions to noise: an analysis of existing survey data. Journal of the
     Acoustical Society of America 75(2): 447-467
[11] Griefahn, B. (1990): Präventivmedizinische Vorschläge für den nächt-
     lichen Schallschutz. Zeitschrift für Lärmbekämpfung 37: 7-14




                                     81
[12] Hume, K., Whitehead, C. (2003): Sleep disturbance due to introduced
     aircraft noise. In: de Jong, R. G., Houtgast, T., Franssen, E. A. M., and
     Hofman, W. (Editors): Proceedings of the 8th International Congress
     on Noise as a Public Health Problem. 199-200
[13] Kallus, K.W. (1995): Erholungs-Belastungs-Fragebogen. Handanwei-
     sung.
[14] Kötz, W.-D. (2004): Zur Frage der effektiven Schalldämmung von ge-
     öffneten Fenstern. Zeitschrift für Lärmbekämpfung 51(1): 21-26
[15] Maschke, C., Hecht, K. (2001): Stresshormone und Schlafstörungen -
     elektrophysiologische und hormonelle Aspekte., in: Bartels, K.H. und
     Ising, H. (Herausgeber): Nachtfluglärmproblematik, Eigenverlag Verein
     WaBoLu, Berlin, 91-97
[16] Maschke, C., Hecht, K., Wolf, U. (2001): Nächtliches Erwachen durch
     Fluglärm. Bundesgesundheitsblatt 44(10): 1001-1010
[17] Mathur, R., Douglas, N.J. (1995): Frequency of EEG arousals from
     nocturnal sleep in normal subjects. Sleep 18(5): 330-333
[18] Ortscheid, J., Wende, H. (2002): Lärmbelästigung in Deutschland -
     Ergebnisse einer repräsentativen Umfrage. Zeitschrift für Lärmbe-
     kämpfung 49(2): 41-45
[19] Passchier-Vermeer, W., Vos, H., Steenbekkers, J.H.M., Van der Ploeg,
     F.D., Groothuis-Oudshoorn, K. (2002): Sleep disturbance and aircraft
     noise exposure - exposure effect relationships. TNO Report Number
     2002.027, Netherlands
[20] Pearsons, K., Barber, D., Tabachnick, B.G., Fidell, S. (1995): Predicting
     noise-induced sleep disturbance. Journal of the Acoustical Society of
     America 97(1): 331-338
[21] Rechtschaffen, A., Kales, A., Berger, R.J., Dement, W.C. et al. (1968):
     A Manual of Standardized Terminology, Techniques and Scoring Sys-
     tem for Sleep Stages of Human Subjects. Washington, D.C.
[22] Riemann, D. (1997): Schlaf und Gedächtnis, in: Schulz, H. (Herausge-
     ber): Kompendium Schlafmedizin für Ausbildung, Klinik und Praxis,
     ecomed, Landsberg/Lech, 1-2
[23] Rylander, R., Bjorkman, M., Ahrlin, U., Arntzen, E., Solberg, S. (1986):
     Dose-response relationships for traffic noise and annoyance. Archives
     of Environmental Health 41(1): 7-10




                                      82
[24] Rylander, R., Björkman, M., Ahrlin, U., Sörensen, S., Kajland, A.
     (1980): Aircraft noise contours: importance of overflight frequency
     and noise level. Journal of Sound and Vibration 69: 583-595
[25] Rylander, R., Sörensen, S., Kajland, A. (1972): Annoyance reactions
     from aircraft noise exposure. Journal of Sound and Vibration 24: 419-
     444
[26] Samel, A., Diedrich, A., Drescher, J., Lorenz, B. et al. (1997): [Long-
     term monitoring of psychophysiologic values in flight physiology]. In-
     ternist (Berl) 38(8): 755-769
[27] Sato, T., Yano, T., Björkman, M., Rylander, R. (1999): Road traffic noi-
     se annoyance in relation to average noise level, number of events and
     maximum noise level. Journal of Sound and Vibration 223: 775-784
[28] Schultz, T.J. (1978): Synthesis of social surveys on noise annoyance.
     Journal of the Acoustical Society of America 64(2): 377-405
[29] Steyer, R., Schwenkmezger, P., Notz, P., Eid, M. (1997): Der Mehrdi-
     mensionale Befindlichkeitsfragebogen (MDBF). Handanweisung.
[30] Wesensten, N.J., Balkin, T.J., Belenky, G. (1999): Does sleep fragmen-
     tation impact recuperation? A review and reanalysis. Journal of Sleep
     Research 8(4): 237-245




                                     83

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:9/27/2012
language:Unknown
pages:90