Traffic noise abatement: How different pavements, vehicle speeds and traffic densities affect annoyance levels

Abstract

In this paper, annoyance ratings from traffic noise recorded on cobblestones, dense asphalt, and open asphalt rubber pavements are assessed with regard to car speeds and traffic densities. It was found that cobblestones pavements are the most annoying; also while open asphalt rubber pavement imposes less annoyance than dense asphalt it is not significantly different. Higher car speeds always lead to greater annoyance, as does higher traffic densities. LAeq and LAmax correlate well with annoyance, but loudness is the best predictor. Roughness and sharpness exhibit inconsistent interactions.

Full text

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy
Accident
Analysis
and
Prevention
51 (2013) 11–
17
Contents
lists
available
at
SciVerse
ScienceDirect
Accident
Analysis
and
Prevention
j
ourna
l
h
o
mepage:
www.elsevier.com/locate/aap
Noise
abatement
and
traffic
safety:
The
trade-off
of
quieter
engines
and
pavements
on
vehicle
detection
C.
Mendonc¸
aa,b,∗,
E.
Freitasc,d,
J.P.
Ferreirac,
I.D.
Raimundoc,
J.A.
Santosa,b,e
aSchool
of
Psychology,
Department
of
Basic
Psychology,
University
of
Minho,
Portugal
bCentro
de
Computac¸
ão
Gráfica,
Portugal
cSchool
of
Engineering,
Department
of
Civil
Engineering,
University
of
Minho,
Portugal
dCentre
of
Territory,
Environment
and
Construction,
Portugal
eCentro
Algoritmi,
University
of
Minho,
Portugal
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
19
June
2012
Received
in
revised
form
8
September
2012
Accepted
26
October
2012
Keywords:
Signal
detection
Road
traffic
safety
Noise
Vehicles
Dense
asphalt
Open
asphalt
rubber
Cobble
stones
Psychoacoustic
analysis
a
b
s
t
r
a
c
t
Road
traffic
sounds
are
a
major
source
of
noise
pollution
in
urban
areas.
But
recent
developments
such
as
low
noise
pavements
and
hybrid/electric
engine
vehicles
cast
an
optimistic
outlook
over
such
an
environ-
mental
problem.
However,
it
can
be
argued
that
engine,
tire,
and
road
noise
could
be
relevant
sources
of
information
to
avoid
road
traffic
conflicts
and
accidents.
In
this
paper,
we
analyze
the
potential
trade-offs
of
traffic-noise
abatement
approaches
in
an
experimental
study,
focusing
for
the
first
time
on
the
impact
and
interaction
of
relevant
factors
such
as
pavement
type,
vehicle
type,
listener’s
age,
and
background
noise,
on
vehicle
detection
levels.
Results
reveal
that
vehicle
and
pavement
type
significantly
affect
vehi-
cle
detection.
Age
is
a
significant
factor,
as
both
younger
and
older
people
exhibit
lower
detection
levels
of
incoming
vehicles.
Low
noise
pavements
combined
with
all-electric
and
hybrid
vehicles
might
pose
a
severe
threat
to
the
safety
of
vulnerable
road
users.
All
factors
interact
simultaneously,
and
vehicle
detection
is
best
predicted
by
the
loudness
signal-to-noise
ratio.
© 2012 Elsevier Ltd. All rights reserved.
1.
Introduction
Traffic
related
noise
is
nowadays
the
major
source
of
environ-
mental
noise
in
most
industrialized
nations
and
developing
regions.
Its
negative
impact
has
been
demonstrated
at
several
instances,
from
health
to
school
efficiency
and
overall
emotional
annoyance
(e.g.,
Gorai
and
Pal,
2006;
Passchier-Vermeer
and
Passchier,
2000;
Sanz
et
al.,
1993;
Freitas
et
al.,
2012).
It
is
therefore
a
matter
of
active
concern
for
traffic-related
researchers,
public
authorities
in
health
and
traffic,
as
well
as
transportation
and
road
industries,
to
find
quieter
alternatives
to
the
major
sources
of
transportation
noise.
In
a
near
future,
we
might
expect
a
reduction
of
road
traffic
noise
both
by
pavements
that
are
more
efficient
and
because
of
the
grow-
ing
popularity
of
hybrid
and
all-electric
vehicles.
Therefore,
there
is
an
optimistic
outlook
on
health
improvement
and
annoyance
reduction
due
to
a
quieter
road
traffic
environment,
specifically
for
populations
living
in
urban
areas.
However,
in
urban
areas
traffic
noise
could
also
be
a
key
fac-
tor
for
the
awareness
of
imminent
conflicts
by
vulnerable
road
∗Corresponding
author
at:
Centro
de
Computac¸
ão
Gráfica,
Campus
de
Azurém,
4800-058
Guimarães,
Portugal.
Tel.:
+351
253
510
580;
fax:
+351
253
510
581.
E-mail
address:
catarina.mendonca@ccg.pt (C.
Mendonc¸
a).
users.
In
other
words,
road,
tire
and
engine
noises
might
be
used
as
meaningful
signals
by
pedestrians
and
bicyclists:
they
can
act
as
attentional
triggers,
allowing
for
a
better
perception
of
speed
and
proximity
of
incoming
traffic
and
for
timely
reactions
to
avoid
con-
flicts.
Therefore,
due
to
traffic
noise
abatement,
we
might
face
in
the
near
future
an
increasing
trade-off
between
the
improvement
of
population’s
health
and
the
rise
of
accidents
involving
vulnera-
ble
road
users.
Such
trade-off
analysis
has
never
been
approached
from
an
experimental
perspective.
When
compared
to
internal
combustion
vehicles,
elec-
tric/hybrid
engine
vehicles
have
higher
incidence
of
crashes
involving
pedestrians
and
bicyclists
(Garay-Vega
et
al.,
2010;
Hanna,
2009).
On
the
one
hand
road
users
show
substantial
interest
in
driving
quiet
hybrid
or
all-electric
cars;
but
on
the
other
hand
they
are
concerned
with
the
reduced
conspicuity
of
such
vehicles
(Wolgater
et
al.,
2001).
Some
experimental
studies
have
addressed
this
issue.
Ashmead
et
al.
(2012)
analyzed
the
path
identification
of
electric
engine
and
internal
combustion
engine
vehicles
in
quiet
and
noisy
environments.
They
found
that
in
quiet
environments
there
were
timely
path
identifications
of
the
electrical
vehicles,
but
not
in
noisy
ones.
They
also
found
that
these
judgments
were
based
on
sound
level,
the
main
characteristic
that
is
altered
in
elec-
tric/hybrid
cars.
Studies
with
visually
impaired
populations
have
also
revealed
lower
vehicle
detectability
of
hybrid
and
all-electric
0001-4575/$
–
see
front
matter ©
2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.aap.2012.10.018

Author's personal copy
12 C.
Mendonc¸
a
et
al.
/
Accident
Analysis
and
Prevention
51 (2013) 11–
17
vehicles
(Emerson
et
al.,
2010).
All
these
data
have
contributed
to
the
official
recognition
by
the
U.S.
National
Highway
Traffic
Safety
Administration
that
electric
vehicles
in
low-speed
operation
may
induce
a
safety
issue
for
blind
pedestrians
(Garay-Vega
et
al.,
2010).
Other
factors
might
affect
traffic-related
noise
and
hence
vehi-
cle
conspicuity.
In
a
previous
study
(Freitas
et
al.,
2012)
we
have
demonstrated
that
pavement
type
largely
affects
the
levels
of
envi-
ronmental
noise
and
related
subjective
annoyance.
However,
the
way
pavement
type
affects
vehicle
detection
is
still
not
clear.
In
addition,
age
might
be
regarded
as
a
relevant
variable.
Young
pede-
strians
are
more
often
involved
in
accidents
than
older
people
are,
but
while
being
rare,
accidents
with
older
people
are
the
most
severe
(Martin,
2006).
In
experimental
studies
with
children,
the
number
of
correctly
identified
vehicle
sounds
was
significantly
improved
with
age
(Pfeffer
and
Barnecutt,
1996).
Despite
the
strong
evidence
of
the
role
of
several
traffic
noise
factors
on
vehicle
con-
spicuity,
there
has
never
been
a
comprehensive
study
analyzing
the
main
relevant
variables
(Barton
et
al.,
2012).
In
this
paper,
we
present
for
the
first
time
such
an
integrated
approach
to
traffic
noise
variables
and
related
vehicle
detection
lev-
els.
We
address
the
detection
of
approaching
vehicles
as
a
function
of
pavement,
vehicle
type,
background
noise
and
the
age
of
the
lis-
tener.
Binaural
pass-by
noise
samples
were
recorded
using
several
combinations
of
pavement,
vehicle
and
speed.
These
samples
were
then
edited
to
create
scenarios
of
approaching
vehicles
in
noisy
environments.
Under
controlled
laboratory
conditions,
participants
had
to
detect
the
approaching
vehicles.
2.
Materials
and
methods
2.1.
Participants
Eighty-nine
participants
were
recruited
from
educational
and
social
institutions
(7–86
years
old,
M
=
36.68,
SD
=
22.12).
Split
into
age
groups,
26
participants
were
juvenile
(19
years
and
below,
M
=
12.93,
SD
=
2.31),
27
were
early
adults
(20–39
years
old,
M
=
27.98,
SD
=
5.33),
19
were
middle
adults
(40–59
years
old,
M
=
50.51,
SD
=
5.94),
and
17
late
adults
(60
years
and
above,
M
=
71.35,
SD
=
6.96).
To
exclude
prior
major
hearing
deficiency
all
participants
underwent
audiometric
screening
tests
at
250,
1000
and
4000
Hz.
As
major
hearing
deficiency
criterion,
late
adults
all
had
the
1000
Hz
and
4000
Hz
thresholds
under
40
dB
HL.
The
remaining
participants
had
those
thresholds
under
30
dB
HL.
On
average,
children
had
as
thresholds
14,
10,
and
4
dB
HL
at
250,
1000,
and
4000
Hz
respectively.
At
those
frequencies,
adolescents
had
9,
5,
and
−1
dB
HL
thresholds;
juvenile
and
early
adults
had
15,
10,
and
9
dB
HL;
and
late
adults
had
24,
19,
and
21
dB
HL
respectively.
Participants
were
all
volunteers.
They
were
instructed
about
the
general
purpose
of
the
study
and
provided
their
informed
consent
about
the
participation
in
the
tests
and
the
confidential
data
manip-
ulation.
Under-aged
participants
had
the
informed
consent
of
their
caregivers.
2.2.
Stimuli
and
equipment
The
pavement
surfaces
selected
for
the
tire-noise
recordings
in
this
study
were:
cobble
stones,
dense
asphalt,
and
open
graded
asphalt
rubber.
The
vehicles
were
a
small
passenger
car
(petrol,
Volkswagen
Polo),
a
hybrid
(Toyota
Prius),
and
a
pickup
truck
(diesel,
Mitsubishi
Strakar).
Both
the
representative
sections
of
the
road
surfaces
and
the
recording
techniques
were
selected
according
to
the
European
ISO
Standard
11819-1:1997.
The
controlled
pass-
by
method
(CPB)
was
used,
with
each
single
vehicle
tire-road
noise
recorded
with
speeds
of
30,
40
and,
50
km/h.
The
tire-road
noise
was
binaurally
recorded
with
a
Brüel
&
Kjaer
Head
and
Torso
Simulator
(HATS)
type
4128-C,
a
Brüel
&
Kjaer
Pulse
Analyzer
type
3560-C
and
the
Pulse
CPB
Analysis
software.
The
noise
samples
were
recorded
with
the
HATS
at
7.5
m
from
the
road
centre
and
at
a
height
of
1.7
m
(for
methodological
details
see
Freitas
et
al.,
2012).
From
each
single
vehicle
recording,
sound
samples
with
the
duration
of
2
s
were
produced.
Sound
samples
were
edited
accord-
ing
to
a
time-to-passage
(TTP)
criterion.
As
such,
sounds
were
not
presented
only
by
vehicle
speed
or
distance
to
the
listener,
but
in
a
combined
form
that
is
relevant
for
the
road
user.
The
TTP
for
all
stimuli
was
fixed
to
3.5
s
i.e.,
at
the
end
of
the
stimulus
presenta-
tion
the
vehicle
would
need
3.5
s
to
cross
the
line
of
sight
of
the
observer.
This
TTP
value
is
considered
the
amount
of
time
in
which
a
pedestrian
of
any
age
is
able
to
perceive
and
make
an
informed
decision
about
crossing
a
road
in
safety.
To
mask
the
vehicle
signal,
five
levels
of
white
noise
were
generated
with
WaveLab
6:
−40,
−35,
−30,
−25,
and
−20
dBV,
pre-
sented
through
the
headphones
at
62,
67,
72,
77,
and
82
dB
(A),
respectively.
A
total
of
135
stimuli
with
signal
plus
noise
were
gen-
erated
with
audio
software
(Ardour):
3
pavements
×
3
vehicles
×
3
speeds
×5
noise
levels.
The
stimuli
were
presented
through
a
computer
with
a
sound
card
Intel
82801BA-ICH2,
a
custom
built
C++
application,
and
AKG
K
271
MKII
closed
headphones.
The
C++
application
allowed
the
reproduction
of
different
audio
scenes:
multiple
audio
files,
stimuli
with
different
number
of
audio
files,
configuration
of
reproduc-
tion
time,
configuration
of
the
visual
stimulus
that
appeared
along
sound
reproduction,
setting
of
the
sound
pressure
for
each
ear
and
collecting
the
participants’
answers.
Using
the
Brüel
&
Kjaer
HATS
and
the
Pulse
Analyzer
referred
before,
this
system
was
cali-
brated
to
achieve
sound
pressure
levels
identical
to
those
recorded
in
the
real
scenarios.
The
values
of
loudness
were
assessed
with
the
Psysound3
application
(Cabrera
et
al.,
2008).
2.3.
Procedure
Within
each
trial,
the
participant
was
presented
with
two
con-
secutive
sound
samples,
with
a
fixed
gap
of
1
s,
one
with
the
signal
plus
noise
and
the
other
with
only
noise.
Both
noise
backgrounds
of
each
trial
had
the
same
level
of
white
noise.
The
135
trials
were
pre-
sented
in
a
pseudo-random
order
(method
of
constant
stimulus).
Participants
were
requested
to
detect
in
which
of
the
intervals,
i.e.,
first
or
second
sample,
was
the
approaching
vehicle
(two-interval
forced
choice,
2IFC).
To
avoid
biased
answers
from
participants
the
left-right
orientation
of
the
approaching
vehicle
and
the
order
of
intervals
were
randomized
across
the
135
trials.
Each
trial
started
only
after
an
answer
was
given
to
the
previous
trial,
and
no
time
limits
were
imposed.
Therefore,
experiments
did
not
have
a
fixed
duration.
3.
Results
3.1.
Pavement,
vehicle
and
noise
levels
A
preliminary
analysis
of
the
data,
after
computing
detection
thresholds
per
participant,
revealed
clear
differences
as
a
func-
tion
of
age.
The
global
mean
detection
was
of
80.51%
and
the
standard
error
(SE)
of
1.09.
The
results
across
age
groups
were:
for
juvenile
a
mean
of
78.27%
(SE
=
2.07);
for
early
adults
87.93%
(SE
=
1.34);
for
middle
adults
79.84%
(SE
=
1.90);
and
late
adults
72.88%
(SE
=
2.33).
These
results
did
differ
significantly
in
a
one
way
ANOVA
(F3=
10.95,
p
<
0.001).
In
a
post-hoc
Sheffé
test,
it
was
found
that
the
early
and
middle
adults
did
not
differ
significantly
(F
=
0.05,
n.s.),
and
neither
did
the
juvenile
differ
from
the
late

Author's personal copy
C.
Mendonc¸
a
et
al.
/
Accident
Analysis
and
Prevention
51 (2013) 11–
17 13
Fig.
1.
Mean
detection
percentages
and
SE
for
pavement,
vehicle
and
noise
(n
=
89).
adults
(F
=
1.2,
n.s.),
but
all
other
age
group
comparisons
remained
significantly
different.
Therefore,
average
results
only
represent
an
overall
tendency,
and
they
should
be
interpreted
separately
by
group
age.
As
such,
the
following
analyses
of
detection
thresholds
per
pavement,
vehicle,
and
noise
are
based
first
on
the
overall
data
and
then
on
age
group
aggregated
values.
The
analyses
of
vehicle
detection
as
a
function
of
the
main
vari-
ables
(pavement,
vehicle
type,
and
background
noise)
revealed
clear
trends
as
depicted
in
Fig.
1.
Detection
percentages
increase
from
the
open
asphalt
rubber
pavement
(M
=
71.87
SE
=
1.25)
to
the
dense
asphalt
(M
=
76.79,
SE
=
1.25)
and
cobble
stones
(M
=
92.99,
SE
=
1.16).
Differences
in
vehicle
detection
per
pavement
type
were
significant
across
all
age
groups
in
a
two-way
ANOVA
for
repeated
measures
(F2,3 =
99.84,
p
<
0.001).
Values
as
a
function
of
vehicle
type
show
a
smoother
but
still
consistent
trend
from
the
hybrid
(M
=
78.72,
SE
=
1.18)
to
the
small
passenger
car
(M
=
80.22,
SE
=
1.17)
and
pickup
truck
(M
=
82.56,
SE
=
1.17).
We
found
a
significant
interaction
between
vehicle
type
and
detection
across
age
groups
(F3,2 =
14.74,
p
<
0.001),
and
an
overall
significant
difference
in
detection
level
across
vehi-
cles
(F2=
3.37,
p
<
0.05),
but
post-hoc
mean
testing
revealed
that
only
the
hybrid
and
pickup
truck
reached
a
significant
difference
(F
=
2.32,
p
<
0.05).
For
background
noise,
detection
percentages
were
above
a
threshold
of
75%
for
the
lowest
three
levels
of
62,
67
and
72
LAeq
(dB
A),
with
mean
values
of
91.48
(SE
=
1.24),
88.01
(SE
=
1.38)
and
83.09
(SE
=
1.42),
respectively.
Mean
detection
was
close
to
the
threshold
for
the
noise
level
of
77
(M
=
74.44,
SE
=
1.34)
and
below
it
for
higher
noise
level
of
82
(M
=
65.72,
SE
=
1.04).
Detection
as
a
function
of
age
showed
similar
trends
as
the
global
ones
described
above
but
with
clear
differences
across
groups
(see
Figs.
2–4).
An
inverted
U
pattern
of
results
is
evident,
with
participants
from
20
to
39
years
old
reaching
detection
per-
centages
above
75%
for
all
but
the
noisiest
background
(M
=
72.04,
Fig.
2.
Mean
detection
percentages
and
SE
for
pavements
and
age
groups.

Author's personal copy
14 C.
Mendonc¸
a
et
al.
/
Accident
Analysis
and
Prevention
51 (2013) 11–
17
Fig.
3.
Mean
detection
percentages
and
SE
for
vehicles
and
age
groups.
SE
=
1.54);
while
younger
and
older
participants
perform
clearly
worse.
Participants
of
60
years
old
and
above
did
not
reach
a
mean
75%
threshold
in
several
conditions,
and
in
some
cases,
they
performed
close
to
random.
For
the
pavements
of
open
asphalt
rubber
and
dense
asphalt,
the
mean
values
were
of
62.41
(SE
=
2.65)
and
67.94
(SE
=
2.20),
respectively.
Detection
means
were
of
69.65
(SE
=
2.57)
for
the
hybrid
and
of
73.18
(SE
=
2.68)
for
the
small
car.
Impair-
ment
of
late
adults
was
also
clear
for
the
highest
background
noise
levels
of
72,
77
and
82
dB(A),
with
mean
detection
values
of
74.47
(SE
=
2.95),
67.29
(3.11)
and
57.53
(2.22),
respectively.
Younger
participants
(7–19
years
old)
performed
slightly
bet-
ter
than
the
oldest
listeners,
but
clearly
worse
than
early
adults.
In
addition,
the
detection
ratios
of
juvenile
participants
were
similar
or
lower
than
those
of
middle-aged
adults
(40–59
years
old).
For
the
pavements
of
open
asphalt
rubber
and
dense
asphalt,
the
juve-
nile
mean
values
were
of
69.85
(SE
=
1.98)
and
75.23
(SE
=
2.20),
Fig.
4.
Mean
detection
percentages
and
SE
for
background
noise
and
age
groups.
respectively.
Detection
was
barely
above
the
threshold
for
the
hybrid
vehicle
(M
=
76.38,
SE
=
1.86).
For
the
highest
noise
levels
of
77
and
82
dB(A),
detection
was
below
threshold
with
mean
values
of
72.77
(SE
=
2.14)
and
63.65
(SE
=
1.77),
respectively.
As
a
complementary
analysis,
we
selected
some
case
scenarios.
We
compared
a
foreseeable
combination
of
low
noise
pavements
with
hybrid
cars
against
traditional
pavements
with
conventional
passenger
cars.
In
the
first
case,
the
impairment
of
vehicle
detection
is
even
clearer
(Fig.
5).
The
overall
mean
detection
in
the
sample
is
of
69.60
(SE
=
1.64)
for
the
hybrid/open
asphalt
rubber
condition
with
only
the
age
group
of
20–39
years
old
being
able
to
reach
a
suprathreshold
of
79.48
(SE
=
2.36).
Older
participants
reached
the
lowest
mean
detection
percentage
of
58.41
(SE
=
3.17)
in
that
con-
dition
and
they
were
still
below
threshold
for
the
ordinary
vehicles
–
dense
asphalt
scenario
with
a
percentage
of
only
70.76
(SE
=
2.55)
of
correct
detections.
The
combination
of
pavement
and
vehicle
types
seems
to
have
an
interactive
effect
on
detection
performance,
which
is
consistent
across
age
groups.
Considering
the
overall
data
from
the
sample,
the
detection
decreases
4.95%
and
21.15%
from
dense
asphalt
and
cobble
stones
to
the
open
asphalt
rubber
pavement,
respectively.
The
mean
decrease
is
of
2.67%
from
the
internal
combustion
engine
vehicles
to
the
hybrid.
In
the
extreme
scenarios,
the
detection
of
the
approaching
vehicles
decreases
8.55%
from
ordinary
vehicles/dense
asphalt
to
ordi-
nary
vehicles/cobble
stones.
The
highest
difference
is
found
when
comparing
hybrid
vehicles/open
asphalt
rubber
with
ordinary
vehicles/cobble
stones,
with
a
23.5%
of
decrease
in
detec-
tion.
Older
adults
of
60
years
old
and
above
are
the
most
impaired
in
these
extreme
scenarios
with
detection
differences
of
12.35%
and
29.88%.
These
results
point
to
a
somehow
additive
effect
where
noisier
vehicles
add
up
to
noisier
pavements
and
interact
with
the
listeners’
auditory
accuracy.
To
analyze
this
interactive
effect,
we
addressed
the
detection
of
incoming
traffic
in
light
of
a
signal-to-noise
perspective.
We
focused
on
the
ratio
between
the
actual
traffic
signal
and
the
back-
ground
noise,
aiming
to
find
the
relation
between
this
ratio
and
vehicle
detection.

Author's personal copy
C.
Mendonc¸
a
et
al.
/
Accident
Analysis
and
Prevention
51 (2013) 11–
17 15
Fig.
5.
Overall
and
age
groups
mean
detection
percentages
and
SE
for
selected
scenarios.
A
systematic
preliminary
analysis
compared
several
equations
to
compute
the
signal-to-noise
ratio
(SNR),
namely:
SNR
=
S/N,
SNR
=
S
−
N,
SNR
=
fS/N,
and
SNR
=
fS
−
N,
where
S
stands
for
sig-
nal,
fS
stands
for
final
signal
(last
250
ms
of
the
stimulus),
and
N
stands
for
noise.
All
these
analyses
were
performed
both
for
the
median
and
for
the
maximum
values
of
the
sound
samples.
From
all
acoustic
measures,
the
best
detection
prediction
was
obtained
by
SNR
=
fS/N.
Both
median
and
maximum
values
yielded
similar
results.
Comparing
all
acoustic
indicators,
we
found
that
the
best
mea-
sure
was
loudness,
when
compared
to
LAeq
and
LAmax.
There
were
no
interactions
with
Sharpness
or
with
Roughness
indicators.
The
scatter
plot
of
all
loudness
SNRs
against
detection
accuracy
levels
is
presented
in
Fig.
6.
Several
data
fitting
methods
were
tested
to
predict
how
both
the
SNR
and
the
detection
levels
interact,
namely
linear,
polyno-
mial
and
logistic
fits.
The
best
correlation
was
obtained
with
a
logistic
fit
(r2=
0.77,
p
<
0.001),
despite
significant
correlations
with
the
polynomial
and
linear
fits
(r2=
0.34
and
r2=
0.67,
respectively).
Fig.
6.
The
signal-to-noise
ratio
[dB
(A)]
against
detection
levels
and
logistic
fit.
Fig.
7.
The
signal-to-noise
ratio
against
detection
levels,
separating
the
vehicle
vari-
able.
Fig.
8.
The
signal-to-noise
ratio
against
detection
levels,
separating
the
pavement
variable.

Author's personal copy
16 C.
Mendonc¸
a
et
al.
/
Accident
Analysis
and
Prevention
51 (2013) 11–
17
LAeq
SNR
did
not
reach
statistically
significant
fits,
while
LAmax
SNR
only
correlated
marginally
with
detection
(linear
fit:
r2=
0.32;
polynomial
fit:
r2=
0.37;
logistic
fit:
r2=
0.38).
To
analyze
how
pavement
and
vehicle
type
affected
the
SNR
detection
values,
we
plotted
each
variable
separately
(see
Figs.
7
and
8).
As
observed
in
Fig.
7,
vehicle
type
is
well
distributed
across
the
plot.
There
is,
however,
a
clear
distinguished
pavement
pattern,
as
seen
in
Fig.
8.
Curve
fitting
for
each
separate
pavement
variable
was
per-
formed.
We
found
a
significant
linear
correlation
between
the
open
asphalt
rubber
pavement
(r2=
0.79,
p
<
0.001)
and
the
dense
asphalt
(r2=
0.75,
p
<
0.001),
but
not
for
the
cobble
stones
(r2=
0.22,
n.s.).
It
was
the
logistic
fit
that
provided
the
best
result
prediction
for
each
variable,
as
seen
in
Fig.
8,
open
asphalt
rubber
obtaining
a
correlation
of
r2=
0.86
(p
<
0.001)
and
dense
asphalt
of
r2=
0.83
(p
<
0.001).
Again,
there
was
no
significant
interaction
between
the
cobble
stones
pavement
and
the
detection
levels
with
the
logistic
fitting
procedure.
This
result
highlights
the
specific
properties
each
pavement
type
and
suggests
that
each
variable
should
be
analyzed
separately.
4.
Discussion
Our
results
clearly
show
a
negative
impact
of
traffic
noise
abatement
on
the
detection
of
approaching
vehicles.
Detection
is
significantly
lowered
by
low
noise
pavements
and
quieter
vehicles.
Interestingly,
pavement
type
had
a
stronger
effect
than
vehicle
type
on
the
detection
levels.
This
might
reveal
that
tire-road
noise
is
a
more
relevant
cue
for
vehicle
detection
than
engine
noise,
namely
at
lower
traffic
velocities
such
as
those
used
in
this
study.
The
anal-
ysis
of
this
finding
may
become
very
complex
since
tire-road
noise
at
low
speeds
is
influenced
not
only
by
the
type
of
road
surface
but
also
by
tire
characteristics
such
as
pattern
and
wear.
In
this
study,
the
tires
of
the
hybrid
vehicle
were
nearly
new,
therefore
less
noisy,
while
the
tires
of
the
other
vehicles
were
worn.
In
this
way
the
effect
of
the
vehicle
type
was
clearly
differentiated.
This
finding
should
be
taken
into
consideration
in
future
stud-
ies
on
traffic
noise
abatement.
Indeed,
there
is
barely
any
research
related
to
pavement
type
and
specifically
low
noise
pavements,
when
compared
to
the
high
data
volume
on
vehicle
engines.
Our
data
strongly
suggest
that
different
asphalt
mixtures
will
con-
tribute
differently
to
traffic
conspicuity
and
vulnerable
road
users’
safety.
Furthermore,
a
novel
approach
to
tire
effect
on
detection
should
be
carried
out.
Also,
age
is
a
critical
factor.
Younger
and
particularly
older
par-
ticipants
are
the
most
impaired.
The
worse
detectability
levels
in
older
listeners
most
likely
reflect
the
typical
hearing
loss
associated
with
age.
On
the
other
hand,
the
decreased
detectability
in
younger
groups
is
congruent
with
data
pointing
out
that
as
they
grow
older,
children
increase
their
accuracy
in
vehicle
identification
(Pfeffer
and
Barnecutt,
1996).
Not
only
the
variables
revealed
direct
and
separate
effects
on
the
vehicle
detectability,
but
they
also
showed
interactive
effects.
This
fact
points
out
to
the
need
for
comprehensive
approaches
that
account
for
subject’s
age
(or
listening
abilities),
vehicle
and
pavement
type,
as
well
as
background
noise.
These
interactions
might
be
regarded
as
a
result
of
loudness
additivity.
Noisier
cars
and
pavements
should
be
more
accurately
identified
by
better
lis-
tening
groups.
Loudness
signal-to-noise
ratio
did
indeed
reveal
some
predictability,
but
it
did
not
account
for
all
variables.
Cob-
ble
stones
traffic
sounds
remain
highly
detected
despite
varying
loudness
levels,
probably
due
to
their
spectral
or
rhythmic
patterns.
Nevertheless,
the
finding
that
loudness
is
the
best
acoustic
measure
to
predict
vehicle
detection,
against
LAeq
and
LAmax,
is
consistent
with
our
previous
results,
pointing
to
loudness
as
the
best
predictor
of
traffic-noise
annoyance
(Freitas
et
al.,
2012),
and
brings
further
support
to
the
claim
that
environmental
noise
assessment
should
have
this
measure
as
a
standard.
One
major
concern
standing
out
from
this
study
relies
on
some
age
groups
(younger
and
older
participants)
performing
below
a
threshold
of
75%,
or
even
close
to
random,
in
several
traffic
scenarios.
In
the
real
world,
the
detection
performance
is
likely
to
be
even
worse.
On
the
one
hand,
we
used
a
standard
white
noise
background,
while
in
everyday
situations
road
traffic
contributes
heavily
to
the
noise
environment,
thus
reducing
the
conspicuity
of
the
sound
envelope
of
each
vehicle.
On
the
other
hand,
in
our
exper-
iments,
participants
only
had
to
detect
one
approaching
vehicle
at
a
time
instead
of
simultaneously
facing
several
targets,
which
would
be
the
case
in
common
urban
scenarios.
Moreover,
transition
periods
between
vehicle
or
pavement
type
are
potentially
very
difficult
and
risky.
Vulnerable
road
users
will
inevitably
have
to
cope
with
a
growing
mix
of
vehicles
and
pave-
ments,
with
varying
degrees
of
conspicuity.
In
such
a
transition
scenario,
hybrid
and
all-electric
vehicles,
circulating
on
low
noise
pavements,
might
prove
quite
difficult
to
detect.
Therefore,
a
trade-
off
between
a
more
pleasant
and
healthy
urban
road
environment
and
an
increase
of
traffic
conflicts
and
accidents
involving
pedestri-
ans
and
bicyclists
should
be
a
matter
of
concern.
In
the
next
section
we
approach
this
matter
from
the
decision-maker
point
of
view.
5.
Trade-off
analysis
From
the
data
here
presented
a
new
debate
should
start
on
noise,
urban
design
and
traffic
policy
to
account
for
the
benefits
and
dan-
gers
of
traffic
sounds.
Here
we
present
an
analysis
of
this
trade-off
accounting
for
the
vehicle
detection
levels
above
discussed
and
overall
traffic
annoyance
levels
found
in
another
paper
(Freitas
et
al.,
2012).
The
critical
factors
to
consider
in
the
decision
making
process
are:
pedestrians’
age,
vehicle
type,
pavement
type,
environmental
noise
and
traffic
speed.
i.
Pedestrians’
age:
Age
is
a
critical
factor
in
vehicle
auditory
detection,
but
not
in
traffic
sound
annoyance
ratings.
There-
fore,
vulnerable
road
users,
such
as
children
and
older
adults
detect
less
efficiently
vehicles
in
noisy
environments,
but
they
are
still
as
affected
by
noise
as
other
age
groups.
On
the
other
hand,
using
age
as
a
criterion
for
traffic
noise
management
would
be
difficult
in
practice.
Planning
urban
design
by
finding
the
most
frequent
target
users
would
require
surveys
or
population
studies,
which
would
not
be
feasible
in
most
cases.
The
simple
solution
of
introducing
elements
that
create
loud
sounds
(such
as
cobble
stone
pavements)
at
obvious
sites
such
as
around
schools
is
also
to
be
avoided,
as
traffic
noise
significantly
affects
chil-
dren
leaning
(Sanz
et
al.,
1993).
From
this
data,
it
stands
out
that
specifically
addressing
these
road
users
by
introducing
loud
ele-
ments
is
not
the
best
approach,
as
the
annoyance
trade-off
would
emerge.
ii.
Vehicle
type:
Vehicle
type
affects
both
detection
levels
and
annoyance
ratings
in
a
symmetrical
way.
Louder
engines
are
more
detected
and
more
annoying,
while
quieter
engines
are
less
detected
and
less
annoying.
Therefore,
from
the
trade-off
point
of
view,
there
is
no
optimal
solution
regarding
vehicle
type.
Some
solutions
have
been
pointed
out,
such
as
introducing
sound
in
quieter
vehicles,
but
such
solutions
would
reintroduce
some
environmental
noise,
with
unpredictable
psychological
reactions.
While
these
solutions
are
still
under
debate,
and
are
therefore
of
difficult
application
and
analysis,
other
approaches
should
be
sought.
iii. Pavement
type:
Interestingly,
this
factor
is
more
crucial
than
vehicle
type
in
lower
speed
traffic
environments,
such
as
those

Author's personal copy
C.
Mendonc¸
a
et
al.
/
Accident
Analysis
and
Prevention
51 (2013) 11–
17 17
where
pedestrians
might
be
involved.
Coble
stone
pavements
are
significantly
more
annoying
than
other
pavement
types.
They
also
provide
the
best
detectability
levels,
being
always
above
85%
on
average.
Dense
asphalt
yields
greater
detectability
than
asphalt
rubber,
while
not
being
significantly
more
annoying.
As
such,
cobble
stone
pavements
are
an
efficient
measure
to
ensure
that
even
in
worst
traffic/environment
combinations
road
users
will
detect
incoming
vehicles.
On
the
other
hand,
when
no
specific
traffic
safety
concern
exists,
dense
asphalt
pavements
should
be
preferred
due
to
their
higher
detectability
with
no
annoyance
trade-off.
iv.
Environmental
noise:
This
critical
factor
influences
both
annoy-
ance
and
vehicle
detection
and
it
is
affected
by
pavement,
speed,
and
vehicle
type.
Average
detectability
results
reveal
that
with
environmental
noise
at
77
dB
(A)
detections
are
close
to
75%
threshold.
Louder
environments
will
be
below
threshold,
and
therefore
more
dangerous.
More
silent
environments
will
be
safer
in
general.
As
such,
environmental
noise
should
be
taken
as
a
decision
criterion.
Louder
environments
should
have
cob-
ble
stone
pavements;
and
silent
environments
(under
77
dB
A)
should
have
asphalt
pavements.
As
such,
in
silent
environments,
annoyance
would
be
prevented
while
preserving
safety.
In
loud
environments,
safety
would
come
first.
v. Traffic
speed:
This
factor
was
not
considered
in
this
study,
as
it
interacts
with
the
other
elements,
such
as
vehicle
dis-
tance.
Faster
vehicles
will
generate
more
noise,
and
therefore
better
identified,
but
they
will
become
more
dangerous.
The
opposite
is
observed
with
slower
vehicles.
But
velocity
strongly
affects
annoyance
ratings.
Both
dense
asphalt
and
asphalt
rub-
ber
become
more
annoying
with
higher
traffic
speed,
but
even
at
70
km/h
they
will
not
be
as
annoying
as
a
cobble
stone
pave-
ment
at
only
30
km/h.
Cobble
stone
pavements
also
become
more
annoying
with
higher
speed,
achieving
80%
of
annoyance
at
50
km/h.
Taking
such
data
into
account,
lower
traffic
speeds
should
always
be
promoted,
namely
in
residential
and
work
areas.
Also,
lower
traffic
speeds
should
be
enforced
in
all
cobble
stone
pavement
areas.
In
conclusion,
from
our
trade-off
analysis,
we
propose
a
simple
decision
model
for
urban
designers.
Environmental
noise
should
always
be
assessed
at
the
planning
phase.
If
under
77
dB
A,
asphalt
pavements
are
preferred.
If
higher,
cobble
stones
pavements
will
better
promote
traffic
safety,
but
additional
measures
to
control
traffic
speed
should
be
envisaged.
Naturally,
most
environmen-
tal
noise
is
indeed
influenced
by
traffic
noise
namely
in
urban
areas.
Therefore,
this
decision
model
should
be
carefully
consid-
ered,
accounting
for
the
predicted
environmental
noise
after
urban
and
traffic
changes.
Additional
studies
should
test
and
validate
our
assumptions,
and
other
potentially
relevant
parameters
should
be
addressed.
Acknowledgements
This
study
was
financed
by
the
Portuguese
Foundation
for
Science
and
Technology
and
FEDER,
projects
FCOMP-
01-0124-FEDER-007560,
FCOMP-01-0124-FEDER-022674,
PEst-
OE/ECI/UI4047/2011,
and
PTDC/TRA/67859/2006.
We
thank
Prof.
Pedro
Arezes
for
his
technical
support.
References
Ashmead,
D.H.,
Grantham,
D.W.,
Maloff,
E.S.,
Hornsby,
B.,
Nakamura,
T.,
Davis,
T.,
et
al.,
2012.
Auditory
perception
of
motor
vehicle
travel
paths.
Human
Factors,
http://dx.doi.org/10.1177/0018720811436083.
Barton,
B.K.,
Ulrich,
T.A.,
Lew,
R.,
2012.
Auditory
detection
and
local-
ization
of
approaching
vehicles.
Accident
Analysis
and
Prevention,
http://dx.doi.org/10.1016/j.aap.2011.11.024
(Available
at:
01.01.12).
Cabrera,
D.,
Ferguson,
S.,
Rizwi,
F.,
Schubert,
2008.
Psysound
3
a
program
for
the
analysis
of
sound
recordings.
In:
Proceedings
of
Acoustics,
IEE,
Paris.
Emerson,
R.W.,
Naghishinen,
K.,
Haperman,
J.,
Wiener,
W.,
2010.
A
pilot
study
of
pedestrians
with
visual
impairments
detecting
traffic
gaps
and
surges
con-
taining
hybrid
vehicles.
Transportation
Research
Part
F:
Traffic
Psychology
and
Behaviour
14
(2),
117–127.
Freitas,
E.,
Mendonc¸
a,
C.,
Santos,
J.A.,
Murteira,
C.,
Ferreira,
J.P.,
2012.
Traf-
fic
noise
abatement:
how
different
pavements,
vehicle
speeds
and
traffic
densities
affect
annoyance
levels.
Transportation
Research
Part
D
17,
321–326.
Garay-Vega,
L.,
Hastings,
A.,
Pollard,
J.K.,
Zuschlag,
M.,
Stearns,
M.D.,
2010.
Quieter
Cars
and
The
Safety
of
Blind
Pedestrians:
Phase
I.
Report
by
National
Highway
Traffic
Safety
Administration,
DOT
HS
811304.
Gorai,
A.K.,
Pal,
A.K.,
2006.
Noise
and
its
effect
on
human
being
–a
review.
Journal
of
Environmental
Science
and
Engineering
48
(4),
253–260.
Hanna,
R.,
2009.
Incidence
of
pedestrian
and
bicyclist
crashes
by
hybrid
electric
passenger
vehicles.
Technical
Report
DOT
HS
811204.
US
Department
of
Trans-
portation,
National
Highway
Traffic
Safety
Administration.
Martin,
A.,
2006.
Factors
influencing
pedestrian
safety:
a
literature
review.
Published
Project
Report
PPR241.
TRLL:
London
Road
Safety
Unit.
Passchier-Vermeer,
W.,
Passchier,
W.F.,
2000.
Noise
exposure
and
public
health.
Environmental
Health
Perspectives
108
(1),
123–131.
Pfeffer,
K.,
Barnecutt,
P.,
1996.
Children’s
auditory
perception
of
movement
of
traffic
sounds.
Child:
Care,
Health
and
Development
22
(2),
129–137.
Sanz,
S.A.,
Gracía,
A.M.,
García,
A.,
1993.
Road
traffic
noise
around
schools:
a
risk
for
pupils
performance?
International
Archives
of
Occupational
and
Environmental
Health
65,
205–207.
Wolgater,
M.S.,
Ornan,
R.N.,
Lim,
R.W.,
Chipley,
M.R.,
2001.
On
the
risk
of
quiet
vehicles
to
pedestrians
and
drivers.
In:
Proceedings
of
the
Human
Factors
and
Ergonomics
Society
Annual
Meeting,
vol.
45,
pp.
1685–1688.