KSCE Journal of Civil Engineering (2017) 21(4):1443-1452 Transportation Engineering

Copyright ⓒ2017 Korean Society of Civil Engineers

DOI 10.1007/s12205-016-0915-5 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

An Analysis of the Relationship between Pedestrian Traffic Volumes

and Built Environment Around Metro Stations in Seoul
Taehyun Kim*, Dong-Wook Sohn**, and Sangho Choo***
Received November 3, 2015/Revised March 13, 2016/Accepted April 29, 2016/Published Online July 11, 2016

··································································································································································································································

Abstract

The success of Transit-Oriented Development (TOD) significantly depends on the capacity of pedestrians to navigate and access
the range of land uses in close proximity to transit stations. In this context, creating pedestrian-friendly environments around transit
stations is considered to be the core of TOD strategies. The concept of a “station catchment area” was employed extensively in the
early stage of developing TOD strategies to designate target areas where urban design strategies or regulations for TOD can be
implemented. This study primarily aims to expand our knowledge about factors in the built environment that affect walking trips
around metro station catchment areas. The study focuses on analyzing the associations between pedestrian traffic volume in the
streets and characteristics of built environments, particularly at the street and the neighborhood levels around metro stations based on
the street’s proximity to the closest metro station. The findings indicate that the relationships between walking and built environment
vary according to the proximity to the metro station. In particular, there was considerable differences in the relationship between
pedestrian traffic volume and the built environment factors at the distance of 400m from the station. Pedestrians near metro stations
tend to prefer walking on wider streets, whereas narrower streets were preferred in areas further from the metro station. Finally, street
connectivity and mixed land use were the most consistent predictors of pedestrian traffic volume.
Keywords: Transit-Oriented Development (TOD), walking, station catchment area, land use, public transit
··································································································································································································································

1. Introduction for TOD can be implemented (Kim and Nam, 2013). TOD strategies
developed by local governments generally impose uniform
Transit-Oriented Development (TOD) has gained popularity as urban design strategies based on Jigu-unit plans (Jigu-unit plan is
an alternative urban planning strategy to address several urban the main spatial planning tool used in the urban planning
problems such as traffic congestion, shortage of affordable system of South Korea) to promote walking activity over the
housing, air pollution, and sprawl. The purpose of TOD is to entire target area designated as a station’s catchment area.
increase the use of public transit and human-powered transport The guidelines for Jigu-unit plans developed by the Ministry
within areas of walking distance to transit centers instead of of Land, Infrastructure, and Transportation (MOLIT) use 1km
automobile use (Wey, 2013). An often unspoken but key component as the size of a station catchment area (Korean Ministry of
to TOD theory is pedestrian access between the transit station Land, Infrastructure, and Transportation, 2014). However,
and the immediate surrounding area because the success of TOD transportation literature indicates that pedestrians’ travel
significantly depends on the capacity of pedestrians to navigate behavior may be diversified by the combined effects of the
and access the range of land uses in close proximity to transit built environment and the pedestrians’ proximity to transit
stations (Schlossberg and Brown, 2004; Kalakou and Moura, (Fruin, 1979; O’Sullivan and Morrall, 1996), suggesting the
2014). In this context, creating pedestrian-friendly environments need for applying differentiated design and planning strategies
around transit stations is considered to be the core of TOD based on the area’s proximity to transit.
strategies (Calthrope, 1993). This study aims to expand our knowledge about factors in the
South Korea has made strong efforts to encourage TOD based built environment that affect walking trips around metro station.
on urban metro transit (subway transit) systems. The concept of a The study focuses on analyzing the associations between
“station catchment area” or “station influence area” is extensively pedestrian traffic volume in the streets and characteristics of built
employed in the early stages of developing TOD strategies to environments, particularly at the street and neighborhood levels
designate target areas where urban design strategies or regulations around metro stations based on the street’s proximity to the

*Member, Research Fellow, Korea Environmental Information Center, Korea Environment Institute, Sejong 30147, Korea (E-mail: [email protected])
**Associate Professor, Dept. of Architecture & Architectural Engineering, Yonsei University, Seoul 03722, Korea (Corresponding Author, E-mail:
[email protected])
***Member, Associate Professor, Dept. of Urban Design and Planning, Hongik University, Seoul 04066, Korea (E-mail: [email protected])

− 1443 −
 Taehyun Kim, Dong-Wook Sohn, and Sangho Choo

closest metro station. This study thus attempts to find implications review of studies of physical activity and the built environment.
for TOD around metro stations in terms of urban design and They claimed that easy access to destinations is the key factor
planning strategies. in whether the built environment promotes transportation
walking whereas safety, aesthetics, and parks are more
2. The station Catchment Area important than destinations for recreation walking.
At the neighborhood level, density and mixed land use are
Station catchment area refers to the spatial area in which key factors associated with walking. Density is arguably one of
stations typically have the greatest impact on land use and the most critical “D” variables that influence transportation
development and from which high potential exists to generate behavior (Cervero and Kockelman, 1997). Density can be
transit ridership (American Public Transportation Association, linked to walking because increasing density creates urban
2009). The concept of the station catchment area is employed settings in which land use is compact and destinations are
extensively in the practice of urban design, planning, and closer, which, in turn, makes walking more feasible and
transportation to clarify target areas where various TOD advantageous and eventually creates a critical mass of
strategies should be concentrated for maximizing TOD’s positive pedestrians (Forsyth et al., 2007; Sun et al., 2014). Several
effects. Several sizes of station catchment areas have been studies have presented the statistical significance of the
suggested. For example, Guerra et al. (2011) noted that a half relationship between walking and density using various density
mile (800 m) became the accepted distance for gauging a transit measures such as gross population and employment density
station’s catchment area in the U.S. However, based on analysis (Frank and Pivo, 1994), development density (Lee et al., 2013;
of walking trips to transit stations, Cervero (1997) claimed that Sohn and Kim, 2010), population density (Saelens et al., 2003),
the catchment area could be larger in suburban areas due to the and residential density (Thomsen, 2011). Many studies have
low residential densities and vast inventories of parking. also intensively documented the positive effect of mixed land
Harrison (2012) argued that 1km was the most accepted distance use on walking (Frank and Pivo, 1994; Lee and Sohn, 2012;
of a rail station’s walking catchment based on the empirical Cervero and Kockelman, 1997). The advocates of mixed land
evidence from Irish planning practices. Although the size of use insist that highly mixed land uses promote walking by
station’s catchment area seems to vary by the regional context bringing origins and destinations closer together (Cervero and
and the type of transit (O'Sullivan and Morrall, 1996; American Kockelman, 1997).
Public Transportation Association, 2009), a consensus supports At the street level, the effects of smaller-scale measures of the
that a station’s catchment area must be defined in ways that built environment, such as pedestrian facilities, the quality of
effectively capture the potential demand of transit ridership. sidewalks, access to public transportation, and the presence of
Transit riders, in general, include pedestrians, bikers, park-and- shops and businesses were examined in the study of walking
riders and those using other non-motorized travel modes behaviors (Ferrer et al., 2015; Kim et al., 2013; Oakes et al.,
(Thompson et al., 2012). In the regional context of Seoul, transit 2007; Sisiopiku and Akin, 2003). The findings of these studies
riders are mostly pedestrians; therefore, a station’s catchment indicate that higher development intensity, greater mix of land
area is designated based on the consideration of the distance that uses, crossing aids, visual quality, and street conditions (e.g.,
people are willing to walk to take a transit ride (Kim and Nam, slope, width, pavement) are associated with greater pedestrian
2013). activity in the street (Borst et al., 2008; Rodriguez et al., 2009;
Lee and Moudon, 2004; Cauwenberg et al., 2012; Park et al.,
3. Correlations between Built Environment and 2013).
Walking In summary, significant associations between walking and
physical environment have been confirmed by previous studies.
Several meta-analysis studies conducted extensive reviews At the neighborhood level, density and diversity of land use
on the link between the built environment and walking. Handy seem to be the most significant correlates of walking. At the
et al.’s study (Handy et al., 2002) provided an overview of street level, factors such as transit access, street connectivity, and
urban planning studies to explore the relationship between the density and diversity of land use were found to significantly
built environment and physical activity (walking and bicycling) influence walking. These findings provide a broad range of
and suggested that mixed-use development, street connectivity, insights to develop urban design and planning strategies for
and good design were supportive of walking. Saelens et al. TOD.
(2008) also reviewed literature on environment and walking
and reported that density, distance to non-residential destinations, 4. Analysis of Built Environment
and land use mix demonstrated a consistent positive relationship
with walking. More recently, Kerr et al. (2012) analyzed the A large body of literature in transportation research has
relationship between the built environment and three different attempted to analyze the characteristics of built environment to
types of physical activity (transportation walking, recreation examine its association with travel behavior. Among various
walking, and total physical activity) based on a comprehensive built environmental factors, this study focuses on density, land-

− 1444 − KSCE Journal of Civil Engineering

 An Analysis of the Relationship between Pedestrian Traffic Volumes and Built Environment Around Metro Stations in Seoul

use mix, street configuration, and street quality to analyze the 5. Research Design
characteristics of the built environment around metro stations.
Density is one of the core attributes of built environment which 5.1 Study Area and Data
is strongly associated with travel behavior (Cervero and The study area of this research was the city of Seoul, South
Kockelman, 1997). The findings from transportation research Korea, with a population of approximately 10 million people in
suggest that higher density development reduces the trips taken 2009 and a population density of 16,694 people/km2. The
by auto (Krizek, 2003) and increases pedestrian traffics (Saelens development density of Seoul was higher than that of New York
et al., 2003). Although residential density and employment City, and its metro lines stretched over 300 km, connecting most
density have been the most extensively used as density measures of its urban centers and neighborhoods.
in transportation studies (Cervero, 1996; Kockelman, 1997; The analysis combined three datasets: 1) 2009 Seoul field
Anderson and Bogart, 2001), these measures are limited in survey data on pedestrian traffic volume, 2) 2009 metro ridership
reflecting the overall level of density in urban areas because they data obtained from the Korea Transport Database (KTDB), and
only capture one aspect of land use (i.e. ‘residential use’ and 3) Geographic Information System (GIS) data of streets and
‘commercial use’). Development density, on the other hand, parcels for Seoul. All datasets were constructed by the city of
focuses on the physical aspect of development regardless of land Seoul and made available to the public for research purposes.
use types, and thus can be used to measure the overall density The 2009 Seoul field survey on pedestrian traffic volume was
level of urban areas (Sohn, 2006). In highly developed urban conducted by the city of Seoul between August 2009 and
areas where residential and commercial uses are mixed in a November 2009 to obtain information on pedestrian traffic
complex way, development density can be an effective tool for volume at the street-segment level. Using a total number of
measuring the overall intensity of land uses. Various data sources 2,200 trained observers, the built environment attributes of street
related to development capacity (e.g. building sqft per acre, floor segments and pedestrian traffic volume (the average number of
area ratio, and building story) are used to estimate development pedestrians passing through the observation points on the street
density (Chatman, 2008; Sohn and Kim, 2010; Handy et al., from 7 am to 9 pm—14 hours—every weekday) were measured
2002). at 11,114 observation points along the streets (Fig. 1). The
Land-use mix is another core attribute of built environment, locations of data collection were assigned among the following
which is typically analyzed by measuring the level of land-use areas: 1) areas near urban centers (downtown, local centers,
diversity. Entropy index and dissimilarity index are most neighborhood centers), 2) old town areas ( The old town areas in
commonly used in transportation studies to assess the effect of Seoul include Chungaechun, Jong-ro and Ulchi-ro), and 3) areas
land-use mix on travel behavior (Hess et al., 2001). Entropy with significant levels of residential density (City of Seoul,
index generally quantifies homogeneity of land use in a given 2010). The midpoints of the arterials and local streets with a
area, while the dissimilarity index estimates the degree to which significant level of pedestrian traffic volumes were chosen as the
different land uses come into contact with one another (Handy et final observation points within the areas.
al., 2002; Cervero and Kockelman, 1997). The limitation of The metro ridership dataset was obtained from the Korea
these index measures, however, is that they do not distinguish Transport Database (KTDB). This dataset included information
between different types of land-use mix and its implication to on daily and monthly metro ridership statistics (the total counts
travel behavior (Hess et al., 2001). In order to overcome this
shortcoming, researchers proposed to use the areal ratio
measures estimating the proportion of land (or development)
areas used by specific land use type to those used by another
(Maat and Harts, 2001; Woo, 2005). Employing such measures
allows the examination of the effect of different types of land-use
mix on travel behavior.
The characteristics of street configuration are proven to
significantly affect travel behavior. Measures such as street network
connectivity, street density, and intersection density have been
employed in previous research to analyze the characteristics of
street configuration (Borst et al., 2008; Chin et al., 2008; Ozbil et
al., 2011). Along with these measures, attributes related to the
quality of street network such as safety, path quality, and path
context have been found to influence travel behavior (Ewing and
Cervero, 2001; Southworth, 2005). For example, traffic calming
(Pucher and Dijkstra, 2003) and neighborhood design quality
(Wells and Yang, 2008) are operationalzed in the literature to Fig. 1. Map of Seoul with the Survey Sample Points Around the
capture the quality of street network. Metro Networks

Vol. 21, No. 4 / May 2017 − 1445 −

 Taehyun Kim, Dong-Wook Sohn, and Sangho Choo

Table 1. The GIS Datasets used in the Analysis factors for the segment (the average number of neighborhood's
Data Format Attribute building stories, the ratio of residential building floor area to the
Parcels Polygon shape Parcel ID and parcel area floor areas of all buildings in the neighborhood), and the street-
Predominant land use type, building story, level built environment factors (the width of sidewalks, the slope
Buildings Database
and total floor area
of sidewalks, the width of a roadway, the availability of a bus
Slope, width of sidewalks, and width of
Streets Polyline shape
roadways stop, the availability of a cross walk, the average number of
neighborhood's building stories on the segment, the ratio of
residential building floor area to the floor areas of all buildings
of entries and exists) for each metro station in 2009. For the on the segment).
purpose of identifying the boundary of a station's catchment area, In order to examine the differences in the relationships between
the relationship between pedestrian traffic volume for each pedestrian traffic volume and built environment factors in
observation point and its closest metro station’s average daily association with the street’s proximity to the closest metro
ridership was analyzed. We assumed that the boundary of a station, six sets of OLS regression models were separately fitted
station’s catchment area corresponded to the maximum distance to pedestrian traffic volume data categorized into groups based
between the observation point and its closest metro station, on their proximity to the closest metro station (0-200 m, 200-
demonstrating a statistically significant association between 400 m, 400-600 m, 600-800 m, 800-1,000 m, and 1,000-
pedestrian traffic volume and the average daily metro ridership. 1,200 m). The regression model for areas more than 1200m
The GIS datasets of parcels, buildings and streets came from away from the metro station was unable to be fitted due to a
the city of Seoul in 2010. These datasets included information on limited sample size. All statistical analyses were conducted using
the physical attributes of parcels, buildings, and streets located SPSS 17.0 software, and the significance level was set at P<0.1.
within the administrative boundary of Seoul. Table 1 summarizes
the physical attributes provided in the GIS datasets. The 5.3 Measure of Built Environment
characteristics of the built environment at the street and The definition of “street level” in this research stands for the
neighborhood levels around the randomly selected observation street segment of 50 m on both sides of the data point for
points were measured using these GIS datasets. pedestrian traffic volume combined with the parcels directly
attached to it (Fig. 3). This definition was employed directly
5.2 Regression Model Design from the method used in the field survey on pedestrian traffic
We designed an OLS regression model to analyze the volume in 2009 to utilize its “street level” variables: slope and
relationships between pedestrian traffic volume in the street and width of sidewalks, access to bus transit, and access to crosswalks.
the characteristics of the built environment at the street and the The other spatial unit of analysis used in the research,
neighborhood levels. In the model, built environment variables “neighborhood level”, was a 400 m radius circular buffer around
measured at street and neighborhood levels were regressed on the observation point for pedestrian traffic volume (Fig. 4). This
pedestrian traffic volume, accounting for the ridership of the spatial unit of analysis was used to measure the characteristics of
closest metro station and its proximity to the street (Fig. 2). The the built environment within easy walking distance from the
model can be represented by the following equation: pedestrians’ location. Although little consistency is present
across studies in the operational definition of walking distance,
Logged avg. weekday ped. traffic volume for segmenti =
the distance of 400 m is one of the most frequently used
f(controlsi, neighborhood BE factorsi, street BE factorsi)
measures of easy-walking distance in walking behavior research
where the logged average weekday pedestrian traffic volume in (Colabianchi et al., 2007).
each segment is a function of: control variables (the average
daily ridership of the closest metro station, the distance to the
closest metro station), neighborhood-level built environment

Fig. 3. Illustration for the Measurement of Built Environment at the

Fig. 2. Conceptual Framework of the Analysis Street Level

− 1446 − KSCE Journal of Civil Engineering

 An Analysis of the Relationship between Pedestrian Traffic Volumes and Built Environment Around Metro Stations in Seoul

Nine measures of built environment were selected for the

investigation (Table 2). The neighborhood level measures
estimated the average number of neighborhood's building stories
and the ratio of residential building floor area to the floor areas of
all buildings to capture the neighborhood's development density
and its level of land-use mix. The street level measures estimated
the built environment factors related to the comfort of walking,
access to bus transit, street connectivity, development density,
and land-use mix.

6. Results
Fig. 4. Measurement of Built Environment at the Street and the
Neighborhood Levels
The descriptive statistics of the variables used in the regression

Table 2. The List of Variables used in the Regression Models

Numeric (N)/
Measure Variable Unit
Dummy (D)
Dependent variable Level of walking trip Average weekday pedestrian traffic volume (logged) N Number of people
Average daily ridership of the closest metro station
Metro ridership N Number of people
Control variable (logged)
Proximity to a metro station Distance to the closest metro station (logged) N Meter
Development density Average number of neighborhood's building stories N Building story
Neighborhood level
built environment Ratio of residential building floor area to the floor
Land-use mix N %
areas of all buildings
Width of sidewalks (logged) N Meter
Slope of sidewalks D Yes (1)/no (0)
Comfort of walking
Number of traffic
Width of the roadway N
lanes
Street level built
Access to bus transit Availability of a bus stop D Yes (1)/no (0)
environment
Street connectivity Availability of a crosswalk D Yes (1)/no (0)
Development density Average number of neighborhood's building stories N Building story
Ratio of residential building floor area to the floor
Land-use mix N %
areas of all buildings

Table 3. Descriptive Statistics of Dependent and Independent Variables

0-200 m 200-400 m 400-600 m 600-800 m 800-1,000 m 1,000-1,200 m
(n=335) (n=367) (n=214) (n=109) (n=49) (n=33)
Mean Std. D Mean Std. D Mean Std. D Mean Std. D Mean Std. D Mean Std. D
Dependent variable
Avg. weekday pedestrian traffic volume 1076.57 923.00 789.18 744.11 453.79 314.98 494.51 343.74 475.60 331.92 412.21 281.23
Control variables
Metro ridership 51,665.05 38,658.67 49664.41 39,258.49 48,591.50 40,682.37 46,761.28 39,144.46 35,795.37 32,027.02 32,382.85 21,167.02
Dist. to the station (m) 122.48 51.20 296.19 55.75 480.74 54.85 694.09 54.76 887.55 62.31 1,104.62 58.87
Neighborhood level measures
Avg. # of building stories 3.55 2.41 3.64 2.40 3.76 2.55 4.02 2.45 3.30 1.90 3.81 2.68
Ratio of residential building floor
areas to floor areas of all buildings 31.23 14.26 31.27 13.82 32.11 13.41 32.30 12.86 32.45 13.28 34.88 13.08
Street level measures
Width of sidewalks (m) 4.42 2.33 4.17 2.15 4.22 2.26 3.81 2.47 3.45 1.54 2.86 1.35
Slope of sidewalks (%) .34 .48 .28 .45 .30 .46 .37 .48 .43 .50 .53 .51
Width of roadway (lane) 3.82 2.85 3.35 2.59 3.22 2.53 3.35 2.39 3.20 2.28 3.59 1.96
Availability of a bus stop* .57 .50 .55 .50 .58 .49 .58 .50 .67 .47 .62 .49
Availability of a Crosswalk* .23 .42 .23 .42 .28 .45 .29 .46 .29 .46 .26 .45
Avg. # of building stories 2.52 2.07 2.62 2.27 2.25 1.33 2.57 2.01 2.59 1.72 2.47 1.70
Ratio of residential building floor
.22 .28 .35 .32 .35 .31 .42 .32 .50 .32 .53 .35
areas to floor areas of all buildings
*Yes (1) / No (0)

Vol. 21, No. 4 / May 2017 − 1447 −

 Taehyun Kim, Dong-Wook Sohn, and Sangho Choo

Table 4. Results of Regression Modeling

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
0-200m 200-400m 400-600 m 600-800 m 800-1,000 m 1,000-1,200 m
(n=335) (n=367) (n=214) (n=109) (n=49) (n=33)
Beta Sig. Beta Sig. Beta Sig. Beta Sig. Beta Sig. Beta Sig.
Control variable
Metro ridership .247 *** .195 *** .138 ** .115 -.030 -.066
Dist. to the station -.116 ** -.147 *** .019 -.072 .148 -.078
Neighborhood characteristics
Avg. # of building stories .003 -.058 -.097 -.076 .144 -.007
Ratio of residential building floor areas to
.034 .020 -.083 .098 -.191 ** .055
floor areas of all buildings
Street characteristics
Width of sidewalks .154 *** .074 .083 .108 .415 *** .713 ***
Sidewalk slope -.117 ** -.088 * -.116 -.122 -.227 ** -.190
Width of roadway .158 ** .110 ** .046 .133 -.224 * -.866 ***
Bus stop .145 ** .097 * .106 .205 .307 .717 ***
Crosswalk .054 .118 ** .156 * .195 * .652 *** .589 **
Avg. # of building stories -.028 -.078 .033 -.078 .096 -.091
Ratio of residential building floor areas to
-.123 ** -.180 *** -.205 *** -.191 ** -.637 *** -.185
floor areas of all buildings
2
Adjusted R 0.27 0.24 0.24 0.31 0.81 0.61
*P<0.1
**P<0.05
***P<0.01

models are presented in Table 3. The average weekday pedestrian street in the range of 0-400 m away from the metro station. These
traffic volume in the nearest area to the station was more than results were similar to the findings of previous studies
two-fold higher (1076.57) than that in the peripheral areas from (Washington Metropolitan Area Transit Authority, 2014; Kim
the station (412.21 in the distance rage of 1,000-1,200 m). The and Nam, 2013; Zhao et al., 2003), reporting that the size of a
statistics for neighborhood level built environment measures station's catchment area for subway transit is in the range of 400-
indicated that variations of the development density and the ratio 800 m when it is determined based on people's acceptable
of residential building floor areas to floor areas of all buildings walking distance to subway transit.
were small over the entire distance range. At the neighborhood level, the associations between pedestrian
As for the street level measures, sidewalks were widest traffic volume and the characteristics of the built environment
(4.42 m) in the nearest areas to the metro station and decreased were not confirmed in most cases. None of the six models
down to 2.86 m in the areas located within the distance range of demonstrated a statistically significant association between
1,000-1,200 m from the metro station. The ratio of residential pedestrian traffic volume and the measure of development
building floor areas to floor areas of all buildings at the street density (the average number of stories in a neighborhood’s
level showed a gradual increase from 22% (0-200 m) to 53% buildings). The effect of land-use mix, measured by the ratio of
(1,000-1,200 m) as the distance to the closest metro station residential building floor areas to the total building floor areas in
increased. The variations for other street level measures according the neighborhood, on pedestrian traffic volume was only
to the proximity to the station were marginal. statistically significant (P<0.05) in the model for the distance
Table 4 presents the association between pedestrian traffic range from 800 m to 1,000 m away from the metro station.
volume and the built environment based on the results of six sets At the street level, the results for the six models indicated quite
of regression models. The adjusted R-squares for the fitted different patterns of relationships between pedestrian traffic
models were in the range of 0.24 to 0.81. The collinearity volume and the characteristics of the built environment. In the
statistics (VIF) indicated that no significant multicollinearity core of a station’s catchment area (the area 400 m or less around
existed among the independent variables. metro station), the street level attributes related to comfort of
The results for control variables indicated a statistically walking, access to bus transit, street connectivity, and land-use
significant association (p < 0.05) between the average daily metro mix were significantly associated with pedestrian traffic volume.
ridership and pedestrian traffic volume in the street located less Conversely, none of these variables except for land-use mix (the
than 600 m away from the metro station. The proximity to metro ratio of residential building floor areas to the total floor areas of
transit, measured by the distance to the closest metro station, was buildings attached to the street) were statistically significant in
also a significant predictor of pedestrian traffic volume in the the range of 400-800 m from the metro station. Finally, more

− 1448 − KSCE Journal of Civil Engineering

 An Analysis of the Relationship between Pedestrian Traffic Volumes and Built Environment Around Metro Stations in Seoul

attributes of the built environment at the street level were outskirts. These findings are consistent with Jiang et al.’s study
statistically significant in areas further away from the metro (2012) on walking behavior around BRT stations, which
station (800-1,200 m): width of sidewalks (p < 0.01), the width reported that distinctive paths with sufficient sidewalks at higher
of the roadway and the availability of a bus stop (1,000-1,200 m, levels of the street hierarchy attracted more pedestrians in
p < 0.01), the availability of a crosswalk (p < 0.05), sidewalk stations’ catchment areas. It is also possible that pedestrians in
slop (p < 0.05) and the ratio of residential building floor areas to areas away from the metro station may choose to walk on
the total floor areas of buildings attached to the street (800- narrower streets because they provide better safety for pedestrians,
1,000 m, p < 0.01). less traffic noise, and less air pollution than wider streets with
higher motor traffic volume.
7. Conclusions Second, the positive effect of development density on pedestrian
traffic volume was not confirmed. Neighborhood level measure
This study analyzed relationships between pedestrian traffic of development density may not be a useful indicator for street
volume and various measures of the built environment at the level walking activity (Greenwald and Boarnet, 2007), which
street and the neighborhood levels near metro stations to test explains the insignificant relationship between pedestrian traffic
whether walking behaviors around metro stations are different volume and the neighborhood level measure of development
from those in areas with lower accessibility to metro transit. density. The insignificant relationship between these two factors
Using six sets of OLS regression models, we found that walking at the street level, on the other hand, could be attributed to the
behaviors in terms of the amount of walking trips generated at poor quality of pedestrian environment in Seoul due to its
the street level had different associations with various factors of extremely dense urban settings (Chung and Chung, 2006). Tao
the built environment according to the proximity to the metro and Qing (2013) claimed that overly intensified buildings along
station. the street damage the quality of the streetscape, making people
The comparison of the results of six regression models uncomfortable to walk. In addition, as noted by Ewing and
provides empirical evidence to identify the size of metro station Cervero (2001), the impact of density on walking may be limited
catchment areas in Seoul based on the relationship between to the extent to which people choose walking over driving at
walking and independent variables according the proximity to highly dense areas to avoid traffic congestion and limited
the metro station. Regarding control variables, the coefficient for parking.
metro ridership was significant up to the point of the distance of Third, the importance of street connectivity and land-use mix
600 m from the location of metro station. The relationship at the street level is noteworthy. These two factors of the built
between pedestrian traffic volume and the distance to the closest environment were the most consistent predictors of pedestrian
metro station was significant up to the point of the distance of traffic volume in the street. Placing mixed use developments
400 m from the station. Several street level variables of the built along the streets and making convenient pedestrian connections
environment associated with the comfort of walking, access to from surrounding areas may be effective in promoting walking,
bus transit, street connectivity, and land-use mix were significant regardless of the street’s proximity to the metro station.
predictors of pedestrian traffic volume in areas near (< 400 m) Finally, the explanatory power of the models was substantially
and on the outskirts (> 800 m) of the metro station, while few of differentiated, with R2 levels ranging from .24 and .81. In
them were significantly associated with pedestrian traffic volume particular, the R2 values of the models for areas located within
in areas with the distance range of 400-800 m. These differences the distance of less than 800 m from the metro station were
can be attributed to the different characteristics of walking mostly low (adjusted R2 < 0.4). Studies investigating the correlates
behaviors between transit walking and non-transit walking (Lee of walking behavior claim that individual, social, and physical
and Moudon, 2006). Given that there was considerable differences environment factors are primary predictors of walking (Giles-
in the relationship between pedestrian traffic volume and the Corti and Donovan, 2002). The low levels of R2 values in models
built environment factors at the distance of 400 m from the 1-4 may be attributed to the fact that pedestrian demographic
station, it can be reasonable to infer that transit walking was factors were not taken into account in the statistical models. In
dominant in areas located less than 400 m away from the station, models 5 and 6, on the other hand, the coefficients indicated that
while non-transit walking prevailed over transit walking beyond the association of pedestrian traffic volume with street-level built
the threshold of 400 m. environment factors were relatively higher than other factors.
The variations of the relationship between pedestrian traffic This implies that the impact of built environment factors at the
volume and the built environment according to the proximity to street-level on walking may become stronger than other factors
the metro station reveals some interesting findings. First, the as the distance from the metro station increases.
relationship between pedestrian traffic volume and the roadway In summary, this study represents the variation of the relationship
width was positive in areas near the metro station (0-400 m), but between walking and the built environment according to the
was negative in areas more than 800 m away from the station, proximity to the metro station. Differentiation of TOD strategies
indicating that pedestrians near metro stations tended to walk on considering the proximity to the transit station should take such
wider streets while narrower streets were more used in the variation into account. It is particularly important to focus on

Vol. 21, No. 4 / May 2017 − 1449 −

 Taehyun Kim, Dong-Wook Sohn, and Sangho Choo

improving micro-level attributes of the built environment when DIO: 10.1080/01944368908976014.

developing urban design strategies to promote walking activities Cervero, R. (1996). “Job-housing balance revisited: Trends and impacts
at the street level. In addition, this study suggests that adopting in the San Francisco Bay area.” Journal of the American Planning
Association, Vol. 62, Issue 4, pp. 492-511, DIO: 10.1080/
TOD strategies to increase the level of land-use mix, develop
01944369608975714.
comfortable sidewalks, and improve street connectivity and Cervero, R. (1997). “Rail access modes and catchment areas for the
transit access could be more effective to promote walking than to BART System.” The University of California Transportation Center,
increase development density in overly intensified urban areas. UCTC, No. 307, pp. 1-55.
Methodological limitations are present in this study. Because Cervero, R. and Kockelman, K. (1997). “Travel demand and the 3Ds:
this study relied on quantitative research methods, the aspects of Density, diversity and design.” Transportation Research, Part D 2,
built environment considered in the analyses were limited to pp. 199-219, DOI: 10.1016/s1361-9209(97)00009-6.
those that could be objectively measured. The demographic Chatman, D. (2008). “Deconstructing development density: Quality, quantity
and price effects on household non-work travel.” Transportation
factors of population, which may be associated with walking
Research Part A, Vol. 42, Issue 7, pp. 1008-1030, DIO: 10.1016/
behaviors, were not considered in the model due to the lack of j.tra.2008.02.003.
micro-level data on these factors. In addition, the gap in the time Chin, G., Van Niel, K., Giles-Corti, B., and Knuiman, M. (2008).
of data collection between the built environment data (2010) and “Accessibility and connectivity in physical activity studies: The
the pedestrian observation data (2009) might have had influenced impact of missing pedestrian data.” Preventive Medicine, Vol. 46,
the reliability of the results since any changes made in the built Issue 1, pp. 41-45, DOI: 10.1016/j.ypmed.2007.08.004.
environment around metro stations during the one-year gap Chisholm, G. (2002). Transit-Oriented development and joint development in
could have had a significant effect on the pedestrian traffic the United States: A literature review, Washington D.C: National
Academy Press.
volume. Future research employing analytical methods to examine
Chung, W. and Chung, J. (2006). “A study about the problem of
the qualitative aspects of built environment and/or socio- pedestrian environment in the Urban Street - Focused on the system
demographic factors of pedestrians in the neighborhoods around and the policy.” Annual Conference of Architectural Institute of
transit station is needed to expand our knowledge on the link Korea, Vol. 26, Issue 1, pp. 189-192.
between walking and the built environment around transit City of Seattle (2013). Implementing transit oriented development in
stations. Seattle: Assessment and recommendations for action, Seattle, WA.
City of Seoul (2010). White paper on Seoul pedestrian traffic volume
survey 2009, Seoul, South Korea.
References Colabianchi, N., Dowda, M., Pfeiffer, K., Porter, D., Joao, M., Almeida,
C., and Pate, R. (2007). “Towards an understanding of salient
American Public Transportation Association (2009). Defining transit neighborhood boundaries: Adolescent reports of an easy walking
areas of influence, Washington D.C. distance and convenient driving distance.” International Journal of
Anderson, N. and Bogart, W. (2001). “The structure of sprawl: Identifying Behavioral Nutrition and Physical Activity, Vol. 4, Issue 66, pp. 1-4,
and characterizing employment centers in Polycentric metropolitan DOI: 10.1186/1479-5868-4-66.
areas.” American Journal of Economics and Sociology, Vol. 60, Dittmar, H. and Ohland, G. (2009). The new transit town: Best practices
Issue 1, pp. 147-169, DOI: 10.1111/1536-7150.00058. in transit-oriented development, Washington D.C: Island Press.
Belzer, D. and Autler, G. (2002). Transit oriented development: Moving Ewing, R. and Cervero, R. (2001). “Travel and the built environment: A
from rhetoric to reality, A discussion paper for the Brookings synthesis.” Transportation Research Record, No. 1780, pp. 87-113,
Institution Center on urban and metropolitan policy and the great DOI: 10.3141/1780-10.
American station foundation. Ferrer, S., Ruiz, T., and Mars, L. (2015). “A qualitative study on the role
Borst, H., Miedema, H., Vries, S., Graham, J., and Dongen, J. (2008). of the built environment for short walking trips.” Transportation
“Relationships between street characteristics and perceived attractiveness Research Part F, No. 33, pp. 141-160, DOI: 10.1016/j.trf.2015.07.014.
for walking reported by elderly people.” Journal of Environmental Forsyth, A., Oakes, M., Schmitz, K., and Hearst, M. (2007). “Does
Psychology, Vol. 28, Issue 4, pp. 353-361, DOI: 10.1016/j.jenvp. residential density increase walking and other physical activity?.”
2008.02.010. Urban Studies, Vol. 44, Issue 4, pp. 679-697, DOI: 10.1080/
Calthrope, P. (1993). The next American metropolis, New York: Princeton 00420980601184729.
Architectural Press. Frank, L. and Engelke, P. (2005). “Multiple impacts of the built environment
Calthrope, P., Yang, B., and Zhang, Q. (2014). Transit oriented development on public health: Walkable places and the exposure to air pollution.”
in China - A manual of land-use and transportation for low carbon International Regional Science Review, Vol. 28, Issue 2, pp. 193-
cities, Beijing, China: China Architecture and Building Press. 216, DOI: 10.1177/0160017604273853.
Cauwenberg, J., Holle, V., Simons, D., Deridder, R., Goubert, L., Nasar, J., Frank, L. and Pivo, G. (1994). “Impacts of mixed use and density on
Salmon, J., Bourdeaudhuij, I., and Deforche, B. (2012). “Environmental utilization of three modes of travel: Single-occupant vehicle, transit,
factors influencing older adults' walking for transportation: A study and walking.” Transportation Research Record, No. 1466, pp. 44-
using walk-along interviews.” International Journal of Behavioral 52.
Nutrition and Physical Activity, Vol. 9, Issue 85, pp. 1-11, DOI: Fruin, J. (1979). Pedestrian transportation, Public Transportation Planning,
10.1186/1479-5868-9-85. Operations and Management eds, G. E. Gray & L. A. Hoel, 188-203.
Cervero, R. (1989). “Jobs-housing balance and regional mobility.” Journal Prentice-Hall, Inc.
of American Planning Association, Vol. 55, Issue 2, pp. 136-150, Giles-Corti, B. and Donovan, R. (2002). “The relative influence of

− 1450 − KSCE Journal of Civil Engineering

 An Analysis of the Relationship between Pedestrian Traffic Volumes and Built Environment Around Metro Stations in Seoul

10.1016/0191-2607(79)90036-0 individual, social and physical and the land-use pattern of the station areas.” Cities, No. 35, pp. 69-
environment determinants of physical activity.” Social Science & 77, DOI: 10.1016/j.cities.2013.06.010.
Medicine, No. 54, pp. 1793-1812, DOI: 10.1016/S0277-9536(01) Lee, Y. and Sohn, D. (2012). “A relationship analysis between Subway
00150-2. Transit Demand and urban spatial characteristics in the subway
Greenwald, M. and Boarnet, M. G. (2007). “Built environment as station area.” Journal of The Urban Design Institute of Korea,
determinant of walking behavior: Analyzing nonwork pedestrian Vol. 13, Issue 4, pp. 23-32.
travel in Portland, Oregon.” Transportation Research Record: Lund, H., Willson, R., and Cervero, R. (2006). “A Re-evaluation of
Journal of the Transportation Research Board, No. 1780, pp. 33-41, travel behavior in California TODs.” Journal of Architectural and
DOI: 10.3141/1780-05. Planning Research, Vol. 23, Issue 3, pp. 247-263.
Guerra, E., Cervero, R., and Tischler, D. (2012). “Half-mile circle.” Maat, K. and Harts, J. (2001). “Implications of urban development for
Transportation Research Record, No. 2276, pp. 101-109, DOI: travel demand in the Netherlands.” Transportation Research Record,
10.3141/2276-12. No. 1780, pp. 9-16, DOI: 10.3141/1780-02.
Handy, S., Boarnet, M., Reid, E., and Kilingsworth, R. (2002). “How O'Sullivan, S. and Morrall, J. (1996). “Walking distances to and from
the built environment affects physical activity.” American Journal of light-rail transit stations.” Transportation Research Record, Vol. 1538,
Preventive Medicine, Vol. 23, Issue 2S, pp. 64-73, DOI: 10.1016/ Issue 1, pp. 19-26, DOI: 10.3141/1538-03.
s0749-3797(02)00475-0. Oakes, J., Forsyth, A., and Schmitz, K. (2007). “The effects of
Harrison, O. (2012). “Measuring rail station catchment areas in the neighborhood density and street connectivity on walking behavior:
greater Dublin area.” Proceedings of the ITRN2012. The Twin Cities walking study.” Epidemiologic Perspectives &
Hess, P., Moudon, A., and Logsdon, M. (2001). “Measuring land use Innovations, No. 4, pp. 1-9, DOI: 10.1186/1742-5573-4-16.
patterns for transportation research.” Transportation Research Oreskovic, N., Roth, P., Charles, S., Tsigaridi, D., Shepherd, K., Nelson,
Record, No. 1780, pp. 17-24, DOI: 10.3141/1780-03. K., and Bar, M. (2014). “Attributes of form in the built environment
Jiang, Y., Zegras, P., and Mehndiratta, S. (2012). “Walk the line: Station that influence perceived walkability.” Journal of Architectural and
context, corridor type and bus rapid transit walk access in Jinan, Planning Research, Vol. 31, Issue 3, pp. 218-232.
China.” Journal of Transport Geography, Vol. 20, Issue 1, pp. 1-14, Ozbil, A., Peponis, J., and Stone, B. (2011). “Understanding the link
DOI: 10.1016/j.jtrangeo.2011.09.007. between street connectivity, land use and pedestrian flow.” Urban
Kalakou, S. and Moura, A. (2014). “Bridging the gap in planning indoor Design International, Vol. 16, Issue 2, pp. 125-141, DOI: 10.1057/
pedestrian facilities.” Transport Review, Vol. 34, Issue 4, pp. 474- udi.2011.2.
500, DOI: 10.1080/01441647.201 4.915441. Park, S., Kim, J., Choi, Y., and Seo, H. (2013). “Design elements to
Kerr, J., Rosenberg, D., and Frank, L. (2012). “The role of the built improve pleasantness, vitality, safety, and complexity of the pedestrian
environment in healthy aging: Community design, physical activity, environment: Evidence from a Korean neighborhood walkability
and health among older adults.” Journal of Planning Literature, Vol. case study.” International Journal of Urban Sciences, Vol. 17, Issue 1,
27, Issue 1, pp. 43-60, DOI: 10.1177/0885412211415283. pp. 142-160, DOI: 10.1080/12265934.2013.776283.
Kim, H. and Nam, J. (2013). “The size of the station influence area in Pucher, J. and Dijkstra, L. (2003). “Promoting safe walking and cycling
Seoul, Korea: Based on the survey of users of seven stations.” to improve public health: Lessons from the Netherlands and Germany.”
International Journal of Urban Sciences, Vol. 17, Issue 3, pp. 331- American Journal of Public Health, Vol. 93, Issue 9, pp. 1509-1516,
349, DOI: 10.1080/12265934.2013.810463. DOI: 10.2105/ajph.93.9.1509.
Kim, S., Choi, J., and Kim, S. (2013). “Roadside walking environments Rodriguez, D., Brisson, E., and Estupinan, N. (2009). “The relationship
and major factors affecting pedestrian level of service.” International between segment-level built environment attributes and pedestrian
Journal of Urban Sciences, Vol. 17, Issue 3, pp. 304-315, DOI: activity around Bogota’s BRT stations.” Transportation Research
10.1080/12265934.2013.825422. Part D, No. 14, pp. 470-478, DOI: 10.1016/j.trd. 2009.06.001.
Kockelman, K. (1997). “Travel behavior as function of accessibility, Saelens, B., Sallis, J., and Frank, L. (2003). “Environmental correlates
land use mixing, and land use balance: Evidence from San Francisco of walking and cycling: Findings from the transportation, urban
Bay area.” Transportation Research Record, No. 1607, pp. 116-125, design, and planning literatures.” Annals of Behavioral Medicine,
DOI: 10.3141/1607-16. Vol. 25, Issue 2, pp. 80-91, DOI: 10.1207/s15324796abm2502_03.
Korean Ministry of Land, Infrastructure, and Transportation (2014). The Schlossberg, M. and Brown, N. (2004). “Comparing transit-oriented
guidelines for Jigu-unit plans, Korean Ministry of Land, Infrastructure, development sites by walkability Indicators.” Transportation Research
and Transportation, Sejong, South Korea. Record, No. 1887, pp. 34-42, DOI: 10.3141/1887-05.
Krizek, K. (2003). “Operationalizing neighborhood accessibility for land Sisiopiku, V. and Akin, D. (2003). “Pedestrian behaviors at and perceptions
use-Travel behavior research and regional modeling.” Journal of toward various pedestrian facilities: An examination based on
Planning Education and Research, Vol. 22, Issue. 3, pp. 270-287, observation and survey data.” Transportation Research Part F, Vol. 6,
DOI: 10.1177/0739456x02250315. Issue 4, pp. 249-274, DOI: 10.1016/j.trf.2003.06.001.
Lee, C. and Moudon, A. (2004). “Physical activity and environment Sohn, D. (2006). An assessment of market preference for smart growth:
research in the health field: Implications for urban and transportation The effects of neighborhood land use and urban design principles on
planning practice and research.” Journal of planning literature, property values, PhD Dissertation, University of Washington, Seattle,
Vol. 19, Issue 2, pp. 147-181, DOI: 10.1177/0885412204267680. WA.
Lee, C. and Moudon, A. (2006). “Correlates of walking for transportation Sohn, D. and Kim, J. (2010). “Analysis of the relationships between
or recreation purposes.” Journal of Physical Activity and Health, land use characteristics of urban transit centers and the level of
Vol. 3, Suppl 1, pp. S77-S98. transit usage: Case studies of Seoul metropolitan area.” Journal of
Lee, S., Yi, C., and Hong, S. (2013). “Urban structural hierarchy and the The Urban Design Institute of Korea, Vol. 11, Issue 1, pp. 33-44.
relationship between the ridership of the Seoul metropolitan subway Song, Y. and Knaap, G. (2004). “Measuring urban form: Is Portland winning

Vol. 21, No. 4 / May 2017 − 1451 −

 Taehyun Kim, Dong-Wook Sohn, and Sangho Choo

the war on sprawl?.” Journal of American Planning Association, Vol. 70, in Norway.” Urban Design International, Vol. 16, Issue 3, pp. 162-
Issue 2, pp. 210-225, DOI: 10.1080/01944360408976371. 170, DOI: 10.1057/udi.2011.8.
Southworth, M. (2005). “Designing the walkable city.” Journal of Thorne-Lyman, A., Wood, J., and Zimbabwe, S. (2011). Transit-oriented
Urban Planning and Development, Vol. 131, Issue 4, pp. 246-257, development strategic plan, Metro TOD Program. Portland, OR.
DOI: 10.1061/(asce)0733-9488(2005)131:4(246). Washington Metropolitan Area Transit Authority (2014). “What's a
Sun, G., Oreskovic, N., and Lin, H. (2014). “How do change to the built transit “Walk Shed”? In PlanItMetro.” Washington D.C.: Washington
environment influence walking behaviors? a longitudinal study Metropolitan Area Transit Authority.
within a university campus in Hong Kong.” International Journal of Wells, N. and Yang, Y. (2008). “Neighborhood design and walking: A
Health Geographics, Vol. 13, Issue 28, pp. 1-10, DOI: 10.1186/ quasi-experimental longitudinal study.” American Journal of Preventive
1476-072x-13-28. Medicine, Vol. 34, Issue 4, pp. 313-319, DOI: 10.1016/j.amepre.
Tao, Z. and Qing, S. (2013). “Study on the D/H ratio of city road's 2008.01.019.
environmental effect based on BIM.” Journal of Theoretical and Wey, W. and Chiu, Y. (2013). “Assessing the walkability of pedestrian
Applied Information Technology, Vol. 49, Issue 1, pp. 449-455. environment under the transit-oriented development.” Habitat
The City of Seoul (2009). The 2009 Seoul field survey on pedestrian International, No. 38, pp. 106-118, DOI: 10.1016/j.habitatint.2012.05.
traffic volume, Seoul, Korea. 004.
Thompson, G., Brown, J., Bhattacharya, T., and Jaroszynski, M. (2012). Woo, Y. (2005). “An analysis of commuting distance and mode changes
Understanding transit ridership demand for a multi-destination, of recent migration to housing renewal areas.” International Review
multimodal transit network in an American metropolitan area: of Public Administration, Vol. 10, Issue 1, pp. 95-108, DOI:
Lessons for increasing choice ridership while maintaining transit 10.1080/12294659.2005.10805064.
dependent ridership, MINETA Transportation Institute, San Jose, Zhao, F., Chow, L., Li, M., Ubaka, I., and Gan, A. (2003). “Forecasting
CA. transit walk accessibility: Regression model alternative to buffer
Thomsen, J. (2011). “Reflections on the opportunities of urban planning method.” Transportation Research Record, No. 1835, pp. 34-41,
to promote non-vehicular transportation in a sustainable settlement DOI: 10.3141/1835-05.

− 1452 − KSCE Journal of Civil Engineering