Proceedings of 20th International Congress on Acoustics, ICA 2010

23-27 August 2010, Sydney, Australia

ICA 2010 1

Assessment of railway soundscape in rural areas

Hyo Joo Yoo (1), Pyoung Jik Lee (1) and Jin Yong Jeon (1)

(1) Department of Architectural Engineering, Hanyang University, Seoul, Korea

PACS: 43.50.-X, 43.50.LJ,

ABSTRACT

In the present study, railway soundscape in rural areas was assessed by field measurements. Landscape merics of ru-ral areas were analysed, then a total of 10 sites were chosen covering different composition of landscape metrics. Au-dio-visual recordings were carried out at selected sites; acoustical characteristics of train noises were analyzed in terms of sound quality metrics, ACF (auto correlation function) and IACF (interaural cross correlation function) pa-rameters. It was found that noise levels of high speed trains ranged from around 70 and 90 dBA in terms of A-weighed equivalent noise levels. And it can be seen that IACC values of train noises were dependent on the layout of recorded sites, and perception of train noises were affected by background noise levels.

INTRODUCTION

The high speed train has been developed since the 1960s, particularly in Japan and Europe, and currently the high speed train has been adopted as a new transportation system in various countries. South Korea is the seventh country to develop a high speed train technology called the Korea Train eXpress (KTX). KTX was opened in 2004 and its maximum speed is around 300km/h. As the KTX has been used pro-gressively as a major transportation system, noise problem caused by the high speed train has grown from a local to a national issue. Many complaints from the residents and visi-tors in rural areas have been directed upon the maximation of the speed of the high speed train.Previously many studies have been focused on community response to high speed trains, which have are based on the social survey and mea-surement of sound pressure level [1,2]. However, the percep-tions of rural areas with high speed trains cannot be explained by only sound pressure levels since high speed trains are just one of the components of rural area consists of other va-riables. Therefore, perceptions of rural areas with high speed trains should be investigated as a concept of soundscape. In this study, audio-visual recordings were carried out in rural areas with high speed trains; acoustical characteristics of train noises were analyzed in terms of sound quality metrics, ACF (auto correlation function) and IACF (interaural cross corre-lation function) parameters.

RURAL SOUNDSCAPE

A rural area is an area outside of cities and towns which have a low population density, where noise levels are dependent on landscape metrics. The variation of noise levels in rural areas is larger, from quiet to noisy. However, in this study, only noisy rural areas have been investigated.

Table 1 shows nine landscape metrics which were chosen based on the research output of the landscape study [3,4]. Landscape metrics consists of two parts; natural and artificial features. Natural features contain agricultural crops, moun-tains, woods, and water features such as lakes and rivers. Artificial features include buildings, man-made constructions, traffic roads, elevated railways, tunnels, and noise barriers.

Sites for audio-visual recordings were selected on the basis of analysis of landscape metrics.

FIELD MEASUREMENT

Site Selection

There are two KTX lines in Korea; the Gyunbu and Honam line. In this study a total of 10 sites, along with these two lines, were selected for field measurement. They are located in the Gyungi province near Seoul, where both KTX lines pass through. Table 2 shows the charactersistics of 10 sites chosen in different landscape metrics, covering 9 codes, to obtain data from a wide range of rural soundscapes.

Table 1. Landscape metrics of rural area Code Description

Natural features

1 Agricultural crops (patch or rice field) 2 Undisturbed valley (mountain) 3 Woods 4 Water features (lake, reservoir, river)

Artificial features

5 Buildings and manmade constructions 6 Traffic road (highway / state road) 7 Elevated railway and bridge 8 Tunnel 9 Noise barrier

Table 2. Characteristics of sites

CodeSites

1 2 3 4 5 6 7 8 9 101 ○ ○ ○ ○ ○ ○ ○ ○ ○2 ○ ○ ○ ○3 ○ ○ ○ 4 ○ ○ ○5 ○ ○ ○ ○6 ○ ○ ○ ○ ○ ○ ○ ○7 ○ ○ ○ ○ ○ ○8 9 ○ ○ ○ ○

2

2

A

FcanfomDre

Inca wdcthin

A

ScateaqanP

TanInrustththT

Wtih0

S

TadLtoCso

3-27 August 201

Audio-visual r

From the field onducted at eacnd the binauraor the audio rec

menter and dumDuring the meaecorder (Fostex

n order to inveeption of sound

camcorder (Swere carried oudicularly away onducted at dahe assumption ng this period.

ANALYSIS O

Sound quality morrelation funcs well as sounderistics were acoustical chara

quality metrics:nd sharpness w

Pulse Software o

To observe the nd ACF were cn all calculationunning step watudy [5]. The he first maximuhe 1 and large

The IACC is the

When IACC is ion of the inco

hear the train no0.

Sound pressu

Table 3 shows tmbient noise w

dies indicated thLA90 as well aso explain the b

Coensel also poundscape as 4

Type

Train Noise

Ambient Noise

0, Sydney, Austr

recording

measurements,ch site. The du

al microphone (cording. The he

mmy head was aasurements, noix, FX2) and DA

estigate the effedscape, visual iSamsung, HMt at two positiofrom the railw

aytime from 11that outdoor ac

OF HIGH-SP

metrics and parction and interad pressure leveanalysed in ordacteristics of h: loudness, rouwere introducedof B&K.

ACF/IACF pacalculated for tns, the integratas 0.1 s, as a 1 and 1 are thum peak of the

er the 1, indicae maximum pea

1, the subjects oming train nooise but it is dif

ure level

the noise levelswithout high sphat percentile ns A-weighted eqbackground noiproposed the l1dBA in terms

Metrics

LAeq

LA10

LA50

LA90

LAmax

LAeq

LA10

LA50

LA90

LAmax

alia

audio-visual rummy head (Ty(Type 4101; Beight of the eararound 1.6 m frises were reco

AT (Sony, PC20

ect of visual imimages were caX-H106). Theons: 50 m and

way road. The r1:00 am to 3:00ctivities are mo

EED TRAIN

rameters extracaural cross corrl (LAeq) and fre

der to investigahigh speed traiughness, fluctud; these were c

arameters, the sthe duration of tion interval wasame manner

he delay time ane normalized Aate a lower andak of the norma

can clearly peroise, whereas thffused when IA

s of high speedpeed train noisenoise levels sucquivalent noise se level of ruralimit of quiet of LA50. LA50

1

80.7 7

85.2 8

72.5 7

56.8 7

87.4 8

62.2 7

63.2 7

56.1 6

54.4 6

77.2 8

recordings werype 4100; B&K&K) were usedrs of the experirom the groundrded on a field

08Ax).

mage on the peraptured by usinge measurement

100 m, perpenrecordings wer0 pm, based on

ost frequent dur

NOISE

cted from autorelation functionequency characate the psychon noise. Sounduation strengthcalculated using

short-time IACFf the train noiseas 0.5 s, and thof the previound amplitude o

ACF. The longed stronger pitchalized IACF.

rceive the direche subjects can

ACC approache

d train noise ande. Previous stuch as LA50 andlevel are usefu

al areas [6]. Dlimit of rura

0 levels of am-

Table 3. Noise

2 3

70.5 83.3

84.1 88.7

78.5 58.1

71.6 47.6

88.8 91.3

71.6 59.4

72.8 64.2

69.3 54.6

68.0 47.4

83.4 66.7

Pro

e K)

d i-d. d

r-g ts n-re n r-

o-n

c-o-d h, g

F e. e

us of er h.

c-n

es

d u-d

ul e

al

biento70to hhighhigh

Whemosof ruspeeand trainaroumetreffe

Tabof tithensureheignois

It wwereleveincr

Fre

Freqshoweventhe nas w

Peaknoissultsat 1

e leves of rural

4

82.7

87.7

72.4

57.9

89.7

60.6

63.1

59.9

55.3

67.3

oceedings of 20th

nt noise for 10 0 dBA, thus thehis definition. Eher than other hway nearby the

en high speed st sites except fural area can died trains. Eventhe railway wa

n noise showedund 70 to 88 drics such as elct on noise leve

le 3 shows the ime. The LAeq inn decreased as e levels were vghts of the railse barriers and o

was also found te different acc

el increased rapreased.

Figu

quency chara

quency charactwn in Figure 2n though noise noise level in d

well as the low f

ks at high freqse and levels ats of engine noiand 2 kHz bec

soundscape

Sites

5 6

88.0 81.1

93.1 85.6

69.7 75.4

59.2 56.3

96.2 88.6

64.6 61.0

59.4 63.5

54.7 58.7

51.9 57.2

87.8 70.9

h International Co

sites in this stuere are no quieEspecially, nois

sites showinge site.

trains pass byfor site 2. Theristinguish and p

n though the disas the same as d large variatiodBA. This is blevated railwayels.

change in the Lncreased until tthe train went

varied around 8lways were difothers did not.

that increase raording to recorpidly while lev

ure 1. Sound pr

acteristics

teristics of hig2. Train noises

levels are slighdBA is mainly gfrequency comp

quency could bt low frequencyise. In general, cause of rolling

7

73.2

77.0

69.9

56.4

78.5

58.1

62.1

53.9

50.8

69.4

ongress on Acous

udy ranged froet rural soundscse level of site g 70 dBA, bec

y, noise levels refore, residentsperceive the noistance between100 m, levels oons in terms obecause differey and noise bar

LAeq of the sites the train passed

by. Maximum85 to 100 dBAfferent and som

ates of sound prrded sites. At

vels of sites 3 a

ressure levels

gh speed trainshowed a sim

htly different. Fgoverned by hiponent.

be assigned to y are expected peaks are obsnoise from wh

8 9

76.6 69.0

80.9 74.5

69.4 57.5

58.4 46.1

84.7 76.5

54.2 43.4

57.9 46.2

49.4 42.2

47.2 38.2

66.5 49.7

stics, ICA 2010

ICA 2010

om around 42 cae according

2 was much cause of the

increased in s and visitors ise from high

n the receiver of high speed of LAeq, from ent landscape rriers had an

as a function d the receiver, m sound pres-A because the

me sites had

ressure levels site 5, noise

and 9 slowly

n noises are ilar tendency For all trains, gh frequency

aerodynamic to be the re-erved around

heel radiation.

10

74.2

79.9

53.0

45.6

83.5

42.9

44.1

42.3

41.5

46.6

23-27 August

ICA 2010

But in this srolling noisshown in thi

Psychoaco

SQ (sound q

Figure 3 repfluctuation sof critical bathe range oftic results. Tcomponents low frequena measure ofand 200 Hzas a function

Figure

t 2010, Sydney, A

study, aerodynae levels, thus is study.

Figure 2. Freq

oustic charac

quality) metrics

presents the castrength of highand. Loudness wf 16 to 20 Bark This indicates tof high speed

cy component if the sensation . At most sitesn of Bark, but s

(a

(b)

(c) Fluce 3. Analysis re

Australia

amic noise levepeaks at 1 an

quency characte

ctersistics

s

alculated loudneh-speed train nwas dominant acontrary to fre

that the effectstrain noise wouin terms of loudfor modulation

s, roughness shites 3, 4, 10 had

a) Loudness

) Roughness

ctuation strength sults of sound q

els were larger nd 2 kHz were

eristics

ess, roughness,noises as a funat high frequenequency characts of high frequuld be larger thdness. Roughne

n between arounhowed less varid some peaks at

quality metrics

than e not

, and nction ncy in teris-

uency han a ess is nd 20 ation t low

A

Proceedings of

Figure 4.

frequency and because rattle modulation in case of fluctuaences between means that diffeffect on slow m

ACF/IACF par

The 1 and strength respec0.2 ms to 50 mtively, and ten time. When trawere large butdecreased. Thudifferent whenby.

The values of proaching and with 1 could ntors perceive hany specific toincreased whenpitch correspon

Figure 4(c) shIACC were lardecreased and

20th Internationa

　

　(

(c) Analysis result

site 5 showed anoises from tthe range of 2

ation strength, ihigh speed trai

ferent types of lmoving modula

rameters

1 are related tctively. As showms correspondin

sites showed siains were apprt when they paus, the perceiven trains are app

f 1 were less leaving. This in

not be perceivedhigh speed trainonal componentn trains are conding to 1 coul

hows the resultrgest when train

slightly increa

al Congress on A

(a) 1

(b) 1

IACC ts of ACF/IACF

a peak at high fthe railway aff0 and 200 Hz. it is hard to indin noises from landscape metriation of train no

to the pitch swn in Figure 4(ang to 5 kHz andimilar tendencyroaching and leassed by, 1 vald pitch of train

proaching and l

than 0.3 whenndicates that pitd. Therefore, ren noise as simpt at that time. B

oming to the red be perceived

ts of the IACCns were approacased when tho

Acoustics, ICA 20

F parameters

frequency. This fected the sounHowever, in th

dentify the diffethe 10 sites. Thics rarely have aoises.

sensation and a), 1 varied frod 100 Hz, respey as a function eaving, 1 valulues dramatical

n noises would beaving or pass

n trains were atch in accordanesidents and visply noise withoBut the 1 valueceivers, thus thby listeners.

C. The values ching then, muse were leavin

10

3

is nd he er-his an

its om ec-of

ues lly be ed

ap-nce si-

out ues he

of ch

ng.

2

4

Ttrivb

Tlaansinvredn

D

PbwFthtrdn

Treofoinoapth

S

InsomsoEtitrravin

3-27 August 201

Thus, direction rains were apprve the highest s

by in front of lis

The values of IAandscape metricnd 9, the changites are open s

noises from appvalues of sites 2epresents that t

dent on surrounnoise itself.

DISCUSSION

Perception of hibackground noiswere chosen wFigure 5 represhan backgroundrain noise. How

differences betwnoise was maske

Figure

Time history ofecorded sites.

opened, listenerore, noise leveln the figure 6. B

obstacles such appear suddenlyhis case, train n

Fi

SUMMARY A

n the present oundscape with

measurements. oundscape is an

Especially in Koion of opennesrain noises werange of around

values of train nng sites, and th

0, Sydney, Austr

of the noise soroaching and leasubjective diffusteners due to th

ACC showed lacs of recorded

ges of IACC wespace, thus list

proaching and l2, 3 and 7 were the IACC of hinding environm

NS

igh speed trainse levels. For b

with low and hents that train d noise level thuwever, in the cween train noised by road traff

e 5. Effect of ba

f train noises wIf the right an

rs can hear trainl increased slowBut if the right

as mountains any and noise levnoise can produc

igure 6. Effect

AND FUTHE

study, acoush high speed trIt was found

n aggregate of norea, soundscapss. Acoustical re analyzed. Tr70 and 90 dBA

noises were depe perception of

alia

urce is perceivaving. And liste

useness when trhe low IACC va

arge variations asites. In the ca

ere large. This iteners can easieaving trains. Hnot much chan

igh-speed train ments of rural

n noise may habrief comparisohigh backgroun

noise level waus listeners can

case of site 2 wse and backgroufic noise nearby

ackground noise

were also affectnd left sides on noise from fa

wly, for instancet side of the sitend valleys, thenvels would incrce startling effe

of layout of site

ER STUDIES

stical characterrain were inves

that the landnatural and man

pe of rural area characteristics

rain noise leveA. And it was fendent on the laf train noises af

ed clearly wheneners can percerains are passingalues.

according to thase of sites 6, 8is because thosily perceive thHowever, IACCnged. This resul

noise is depenarea as well a

ave an effect onon, sites 9 and 2nd noise levelsas much highe

n easily perceivwith small leveund noise, train

y.

e level

ted by layout oof the site werar away. Thereesite 8 as showne was closed by

n the train wouldrease rapidly. Inects.

es

ristics of rurastigated by fielddscape of ruran-made featureshowed a variaof high speed

els varied in thfound that IACCayout of recordffected by back

Pro

n e-g

e 8, e e C lt

n-as

n 2 s. er e

el n

of e

e-n y d n

al d al es. a-d e C d-k-

grousounlabo

RE1

2

3

4

5

6

oceedings of 20th

und noise levendscape with hioratory experim

FERENCESJ. Lambert, P. from high-speeSound and VibrX. Chen, F. Tamaglev noise Shanghai” Tra(2007) W. Dramsta, Mtionships betwebased indicatorurban planningE. Roggea, F.rural landscapeas landscape m159-174 (2007)R. Shimokuraderground railwINTER-NOISE,B. D. Coenselsoundscape and

h International Co

ls. In the futurigh speed trains

ments.

S Champelovier

ed train noise: Aration 193, 21-ang, Z. Huang impacts on r

nsportation Re

M. Tveit, W. Feen visual landrs of landscapeg 78, 465-474 (2. Nevens and Hes in Flanders:

metrics” Landsc) and Y. Soeta, way in differen

E, Ottawa, Canal and D. Bottd How to chara

ongress on Acous

re, preference s will be carried

and I. Vernet, A social survey28 (1996) and G. Wang, esidents: A ca

esearch Part D

Fjellstad and Gdscape preference structure” La2006) H. Gulinck, “PLooking beyonape and urban

“Train interior nt cross-sectionada (2009) teldooren, “Theacterize it” Acus

stics, ICA 2010

ICA 2010

test on rural d out through

“Annoyance y” Journal of

“High-speed ase study in 12, 437-448

G. Fry, “Rela-ces and map-

andscape and

Perception of nd aestheticse

planning 82,

noise of un-ns of tunnel”

e quiet rural stica 92, 887-