You are on page 1of 10

Ecological Modelling 289 (2014) 2635

Contents lists available at ScienceDirect

Ecological Modelling
journal homepage:

From design to digital model: A quantitative analysis approach to

Garden Cities theory
Zhiyuan Yuan a , Xinqi Zheng b, , Lina Lv a , Chunlu Xue a

School of Land Science and Technology, China University of Geoscience, Beijing 100083, China
School of Information Engineering, China University of Geoscience, Beijing 100083, China

a r t i c l e

i n f o

Article history:
Received 4 March 2014
Received in revised form 22 June 2014
Accepted 23 June 2014
Available online 11 July 2014
Garden City
Digital city modeling
Geographic information system (GIS)
Comparative analysis
Land use structure
Open green space accessibility

a b s t r a c t
As a complement to the development of new theories, the reevaluation and knowledge mining of classical
theories can be benecial for urban development. In particular, quantitative analyses for cities can now
take advantage of geographic information systems (GIS). Proposed more than one hundred years ago,
Ebenezer Howards Garden City is a generally acknowledged classical urban theory. On the basis of the
original work, we model a digital Garden City in ArcGIS. The model is accurate to within 1% for both areal
and length measures, and enables our further quantitative evaluation of the urban land-use structure
and open green space accessibility. We then compare the classical theory with a modern-built area for
the quantitative evaluation results. Zhujiajiao Town in Shanghai, winner of the International Award for
Livable Communities in 2008, provides a reference. Although the central areas of Garden City and Zhujiajiao Town have different geographical and historical backgrounds, the measured land-use structures,
including indicators of area proportion and area per capita, exhibits similarities on land-use types of residential, transportation, and ecological conservation, which offer a considerable reference for land-use
structure of a livable urban area. Comparison of the accessibility to open green spaces in both cities shows
that the average access time from a residential area to open green space in Garden City is just 186.77 s,
which is much shorter than that in Zhujiajiao. Our research shows that the classical Garden City theory
can be modeled into highly accurate digital forms, allowing richer information in quantitative terms to be
obtained than from the original theory, and enabling comparisons with modern cities. Besides, the proposed digital modeling approach is widely applicable to classical theories and historical planning cases.
2014 Elsevier B.V. All rights reserved.

1. Introduction
In 1898, Ebenezer Howard proposed the Garden City in his book
To-morrow: A Peaceful Path to Real Reform, which has been widely
known through subsequent editions named Garden Cities of Tomorrow. The Garden City is characterized as a living space that
combines the advantages of urban and rural life, with a social city
designed to overcome further increases in population after the limit
of the Garden City is reached (Howard, 1898, 1902). The theory, its
practical applications in Letchworth (Miller, 1989; Purdom, 1963)
and Welwyn (Reiss, 1920), and the resulting worldwide Garden City
movement (Jin, 2007; Ward, 1992) have all been extensively studied. For example, the theory has been recognized as the cornerstone
of modern urban planning (Alexander, 1992; Buder, 1969). Further,

Corresponding author at: No. 29, Xueyuan Road, Haidian District, Beijing 100083,
China. Tel.: +86 13401184568; fax: +86 1082321807.
E-mail address: (X. Zheng).
0304-3800/ 2014 Elsevier B.V. All rights reserved.

the idea of the towncountry magnet has caused general debate

(Clark, 2003; Mumford, 1961), and the changes in residential living
conditions have also been discussed (Edwards, 1913; Reade, 1913).
Bottlenecks in the progress of urbanization, such as trafc congestion, environmental pollution, and urban smog, have
encouraged people to explore new solutions. The development of
further studies on classical theories and historical urban planning
cases for modern practice can be an effective means of nding
new solutions. Therefore, it is sensible to study Garden City theory more in-depth. However, a survey of studies on Garden Cities
over the past 100 years suggests that Howards theory has only
been considered qualitatively. That is, researchers have attempted
to understand and discuss Howards idea by analyzing the text and
accompanying diagrams in the original work, while an accurate
analysis of the theorys physical characteristics (layout, land-use
structure, per capita indicators, etc.) has not been researched.
In recent decades, geographic information system (GIS) technology has been widely used in urban planning. The advantages
of spatial-data organization, management, and especially spatial

Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

analysis, make GIS particularly suitable for complicated calculations (Fotheringham and Rogerson, 1994; Matejcek et al., 2006).
The application of GIS in urban planning generally falls into one
of the two categories. The rst assists with the visual presentation
and quick editing of planning schemes (Beregovskih et al., 2010;
Malczewski, 2004), and the second supports urban modeling, analysis, and prediction for real cities (Jiang and Yao, 2010). Both are
implemented in modern cities in the real world. Even the leading
studies of smart cities focus on the visual representation and simple
analysis of real cities (Al-Hader and Rodzi, 2009). However, modeling classical and historical theories or cases is a challenge, because
we have only incomplete data, inaccurate diagrams, and poorly
organized descriptions. The archeological mining of these classical theories based on GIS, although rarely reported, can provide
considerable benets to modern urban planning.
Ongoing studies of urban spatial analysis are mainly focused
on open-space accessibility (Geurs and van Wee, 2004; Tsou et al.,
2005), landscape pattern measurement (Herold et al., 2002; Kong
and Nakagoshi, 2006; Seto and Fragkias, 2005), complex transportation network analysis (Brockmann and Helbing, 2013; Crucitti
et al., 2006), and energy metabolism and sustainability (Hall, 2011;
Yang et al., 2014; Zhang et al., 2014). Since the 1950s, accessibility
has been an important indicator for the urban green-space distribution (Van Herzele and Wiedemann, 2003). Accessibility can be
measured by various methods, such as the buffer zone, minimum
distance, travel impedance, and gravity index (Luo and Wang, 2003;
Oh and Jeong, 2007; Talen and Anselin, 1998). Modern-built cities
are generally the basis of these studies, and the analysis of classical
and historical theories or cases is rare.
Therefore, our study considers the following three questions.
First, is there a reliable modeling procedure for those classical
urban theories presented in texts and diagrams but only available
on papery material? If so, what are the differences between the
modeling procedure for cities in classical theories and modern society? Second, is the urban land-use structure comparable in cities
or towns with different geographical and historical backgrounds?
Will the land-use structure designed by Howard be similar to that
in modern cities? Third, as the pioneer of urban planning, Howard
tried to solve the environmental problems of urbanization with a
spatial allocation of open green spaces. Thus, what are the quantitative results of judging his planning scheme through an open green
space accessibility analysis? In the remainder of this paper, we rst
reorganize the data and diagrams in Howards original work, and
model a highly accurate digital Garden City using the ArcGIS software. We then quantitatively analyze Howards theory in terms of
a land-use structure calculation and an open green space accessibility measurement on the basis of the digital model. Finally, taking
Zhujiajiao Town as a reference, we apply the urban modeling procedure of modern cities, and analyze the results with the same
land-use structure calculation and open green space accessibility
measurement procedure used for the Garden City. We then contrast the modeling procedure for classical theories with that for
modern cities, and compare the land-use structure and open green
space accessibility of Howards Garden City with that of modern
Zhujiajiao Town.

2. Method
2.1. Modeling classical theory
The modeling procedure provides the foundation for further
quantitative study. We have previously attempted to model a digital Garden City (Yuan et al., 2013a). Further improvements and
repeated experiments have enabled us to identify the main modeling process for a digital Garden City, and this procedure is also


suitable for other classical theories expressed by a literal description and schematic pictures. The modeling process consists of
an analysis of the original theory, the selection of an appropriate mathematical foundation, data extraction and calculation, the
systematic organization and design of an attribute structure, vectorization and digitization, verication of the model accuracy, and
(if necessary) corrections to the model (see Fig. 1).
2.1.1. Original theory analysis
Analyzing the original theory ensures that the natural idea
and characteristics are understood and accurately reected in the
model. The analysis is also important for the data calculations and
framework design, which are strongly inuenced by the information in the original work.
The modeling process of a digital Garden City offers a specic
example. The analysis of Howards original theory helps us understand what should and can be present and calculated in its digital
form, and thus determine the extent and object of the model. The
Garden City is part of a city group named the social city. Each social
city includes several separate urban areas (one central city encircled by several Garden Cities), agricultural areas in the intervals
between cities, and transportation systems of highways and railroads connecting each urban area. The most detailed part of the
theory concerns the urban area of the Garden City, and this is the
most important part of our study on urban planning. Thus, we consider the models extent to be the entire social city, and focus on
the central area of a single Garden City.
2.1.2. Mathematical foundation selection
Being ideal, the cities in classical theories are usually designed
without spatial reference. Therefore, it is necessary to select a
mathematical foundation that includes a suitable spatial coordinate system, a map projection that minimizes deformation, modern
international units, and a scale that is appropriate to the size of the
modeled city.
Because of the ideal and hypothetical characteristics of the social
city, the mathematical foundation of a digital Garden City model
need not consider the citys geographical location. Therefore, we
build a new projection coordinate system called garden city. To
minimize the deformation of the model, we use the Aitoff projection because of the round shape of the social city and the Garden
Cities. The false easting, false northing, and central meridian are all
set to 0 because of the theorys idealistic construction. The model
uses length units of meters. The geographic coordinate system of
the garden city projection coordinates is GCS Beijing 1954 with
Greenwich as the prime meridian, D Beijing 1954 as the datum,
and Krasovsky 1940 as the spheroid.
2.1.3. Data preparation
The data in theories are usually scattered, incomplete, or hidden within text and diagrams. Data preparation aims to form a
systematic, complete, and accurate data list through data extraction and calculation (see Fig. 2). Extracting the core points of a
theory can make the data more systematic. The main references
for this step are the data given in the text, urban layouts shown
in diagrams, and geometric formulae. In some cases, data cannot
be calculated because of deciencies in the original work. Thus,
it is necessary to add one or more assumptions to the authors
design. These assumptions should be reasonable and reliable, and
be related to the geometric features of modeling objects, such as the
length and width of roads, area of parks, and so on. Both exact numbers and numerical relations can be assumed. Some assumptions
may prove to be false, whereas others may be found to be poor when
compared to other calculation results. Such assumptions should be
abandoned or corrected. After comparing and adjusting, the nal
data list should be based on a single best assumption, or the group of


Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

Fig. 1. Modeling procedure for classical urban theories.

points of the central city and six Garden Cities in the concentriccircle layout of Howards design. The calculation is performed in
the following sequence: (1) urban boundaries of the central city
and Garden Cities; (2) inner and outer radii of the rings of the central garden, the whole town, the factories and warehouses, and
the circular railway; (3) inner and outer radii of Grand Avenue;

assumptions that provide optimal results. The comparison of models formed under different assumptions might offer further insight
into the development of cities.
For the data preparation step in the modeling of digital Garden
City, all original data are extracted from the original work. Then,
the core point system is conrmed precisely according to the center


Modeling object

Modeling object A

Modeling object B1

Modeling object B2

Modeling object B3

Core point B1

Core point B2

Core point B3

Core point A

Text descripon & Data

Diagram pictures



Assumpon Xi
Geometric formula

result Ri
Geometric formula


Review & Contrast & Analysis

results Rm
Calculaon results RA



Data discrepancy or
against common sense



Calculaon results RB1 Calculaon results RB2 Calculaon results RB3

Data list (based on assumpon groups Xm)

Fig. 2. Detailed data preparation procedure.

Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

(4) road (referring to radial roads connecting outer border of the

Grand Avenue and the boundary of central urban area) widths; (5)
widths of residential rings and avenues; and (6) inner and outer
radii of the Crystal Palace, central park, and the ring of large public
buildings. The calculation is based on the four following assumptions: (1) the four residential rings are of equal width; (2) the 2nd,
4th, and 5th avenues have the same widths as the road, because
they have the same urban function (can be regarded as roads dividing two adjacent residential patches); (3) the area of the factory
and warehouse rings in the original work does not include area of
these roads between adjacent patches; (4) the ring of large public
buildings and the Crystal Palace are of equal width (referring to the
difference between inner and outer radii of the ring).
2.1.4. Systematic organization and attribute structure design
The model framework design must consider the available data,
the feasibility of vectorization in a GIS software, and further
research intentions. Each item in the model should be represented
in feature les. Generally, point features are used to denote critical
points in the model, which determine the positions of objects, and
attributes of point features should include location, coordinate,
and governmental authority. The central lines of roads, outer contours of buildings, and park walls are signied by the line feature,
and attributes should include the serial number, length, function,
etc. The polygon features represent areas or regions of pavements,
buildings, and parks, and its major attributes include area, land-use
type, and so on.
The data framework and structure of the digital Garden City
model includes two levels. One is the social city, including the
central city, the Garden Cities, agricultural land, and the transportation systems linking the central city and Garden Cities. The other
is the urban area of a Garden City, which is reorganized according to land-use details, i.e., buildings (including boundaries and
patches of building areas for residential, public, and factorial use),
urban transportation system (including central lines, boundaries
and patches for boulevards, avenues, streets and roads, and patches
of their intersections), and green spaces (including cut-off lines and
patches of green belts, and boundaries and patches of open green
2.1.5. Vectorization and digitization
Vectorization is implemented in the GIS software. The process
follows a specic spatial sequence: (1) core points, (2) other control
points, (3) polylines, (4) topology check and correction of polylines,
(5) polygons, and (6) topology check and correction of polygons.
For the modeling process of the Garden City, vectorization is
implemented in ArcGIS 10.0. All geometric data is precise to three
decimal places. The most time-consuming part is the segmentation
and extraction of the road network and green belts in the urban area
of a Garden City. Further, the topology is built and checked after the
vectorization process.
2.1.6. Accuracy verication and correction
The model accuracy is controlled by comparing the data list
extracted from the theory with the corresponding measurement
results from the digital model in the GIS. If the relative error of
certain parts is unsatisfactory, the items causing the error should
be recalculated, and the entire model should be adjusted or rebuilt
according to the location or calculation sequence of these items.
The model can be accepted and used for further analysis when the
relative errors of all parts and the entire model are small enough to
accurately reect the relevant theory.
In modeling the Garden City, the relative errors are calculated
by comparing the datasets obtained from the original work and the
digital model. We calculate the relative error of each item mentioned in the original work, and summarize the average relative


errors of the length measurements, areal measurements, and count

measurements. The nal digital model and accuracy calculation
results are given in Section 3.1.
2.2. Measurement of open green space
On the basis of the digital model, we quantitatively measure
the Garden Citys open green space as the proportion of the whole
central urban area and the open green space area per capita, and
study the accessibility in terms of travel impedance.
The accessibility analysis measures the shortest distance from a
starting point (only residential areas are considered in this study)
to the destination area (i.e., the open green space). If we regard
the travel impedance distribution as homogeneous and undifferentiated in the city, the shortest distance should be the length of
a straight line connecting the starting point and destination area.
However, factors such as the land-use type, population distribution, and trafc network affect travel impedance. Various routes
connecting the starting point and destination area are therefore
considered, and that one with the minimum travel impedance
is selected and used to conrm the corresponding accessibility
between the starting point and destination area.
In this study, we use the access time of the route with the minimum travel impedance to represent the accessibility. The better
the accessibility, the lower the minimum accumulated impedance,
and thus the shorter the access time. We divide the entire area into
cells (i.e., rasterization), and value each of them according to the
impedance per unit distance. For a given walking speed, the time
required to move diagonally through each cell can be calculated.
Then, the total access time for each possible route linking the starting cell to the identied destination areas is easily obtained, and
the accessibility can be conrmed by comparing the access times
of all possible routes. The necessary calculations and comparison
can be performed by the spatial analyst tool in ArcGIS. The specic steps are as follows (Van Herzele and Wiedemann, 2003; Zhou
et al., 2008):
(1) For each factor affecting travel impedance (e.g., land-use type,
population distribution, trafc network), determine the relative travel impedance under every condition and the weight of
each factor. Differences caused by the means of transportation
and road grades should be taken into account. In our accessibility analysis, the land-use type is the only factor affecting travel
impedance in the digital Garden City. This is because the population distribution is missing from the original work, and the
effect of all the roads can be reected by regarding their patches
as a land-use type. The categories and corresponding relative
impedances of land-use types are determined on the basis of the
land-use conditions and a summary of similar research cases in
China (Guo et al., 2012; Hu et al., 2005; Li et al., 2008; Yin and
Kong, 2006; Yu et al., 1999; Zhou et al., 2008; Zhou and Guo,
2003) (Table 1).
(2) Add attribute elds of factors affecting travel impedance in the
digital model vector le and insert corresponding attributes to
each patch. The impedance attribute of the patches in the vector
le can then be obtained by a weighted-sum calculation, i.e., the
sum of three products multiplying the travel impedance value
under the specic condition of each factor and the associated
weight. In the digital Garden City, the impedance of each patch
depends on its land-use type.
(3) Set the transportation means for access and basic travel
speed, and assign travel speeds to patches according to their
impedance. Travel speed is a representation of impedance
that combines all affecting factors and transportation means.
Because of the urban size and centralized planning of the Garden City, we only consider walking, and set a basic walking


Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

Table 1
Relative travel impedance of different land-use types for walking in Garden City.
Land-use types

Physical items

Relative travel impedance

Residential areas
Open green space
Administrative areas
Commercial- and industrial-related areas

Boulevards, avenues, streets, roads, and their green belts

Four residential rings
Central garden, central park, and the grand avenue
Administrative ring, schools, and churches
Crystal palace, industrial ring







speed of 1 m/s according to the general setting ranges in trafc engineering (Oh and Jeong, 2007), regardless of the impact
of different road grades on the travel cost. The relative travel
speed is numerically equal to the travel impedance when the
walking speed on a road is 1 m/s.
Select the physical categories of open green space according to
the original theory, and extract them as a single vector le. This
is the source data for the cost distance. The open green space in
the Garden City is composed of the central garden, the central
park, and the grand avenue.
Convert the polygon data to raster data with the conversion
tools in ArcGIS. Recheck the attribute representing impedance
conditions, namely the travel speed, of each cell. In the digital
Garden City, we use a cell size of 10 m 10 m for rasterization.
Extract the study area (in this study, the residential areas) as
a single polygon layer. This layer should be selected as the
processing extent in the environment settings in the ArcGIS tool
of cost distance. The four residential rings in the digital Garden
City are considered to be study areas.
Calculate the minimum distance (measured in accumulated
access time) of each cell in the study area to the open green
space using the cost distance tool in ArcGIS, and output the
distance raster (the distance is measured by access time). The
whole calculation and comparison process can be performed by
the ArcGIS software.
Classify all cells in the output raster of step (7) according to
travel time. The intervals can be determined autonomously. The
classication map can be obtained in the software by setting
the intervals in the classication label of the les properties.
Statistics on the area and proportion of each accessibility class
can also be computed with the spatial analysis tool of ArcGIS.
This allows the accessibility distribution in the study area to be
plotted. In the digital Garden City, we select 300 s (5 min) as the
interval, considering the results of step (7). The equal interval
classication map and statistics with a class interval of 5 min
walking time are shown in Section 3.2.2.

3. Results
3.1. Digital Garden City
3.1.1. Interface of the digital model
Through the modeling process described in Section 2.1, we successfully transformed Howards Garden City theory into a digital
model (see Fig. 3). This digital form is equipped with systematic
organization and structured attributes, and enables quantitative
measurements and calculations.
3.1.2. Accuracy verication
We calculate the relative errors by comparing two datasets, one
extracted from the original work and the other measured from
the digital model (Table 2). The results conrm that the digital
model is t for further analysis, with an average relative error of
0.90% for the areal measures, 0.57% for the length measures, and
0 for the count measures. The maximum errors occur for the total
area of the gardens and the total road length (measured along the

centreline), and are 5.57% and 3.65%, respectively. Considering the

high accuracy of the individual parts and the entire model, we conclude that this model is accurate enough for further analysis.
3.2. Measurement of open green space in the digital Garden City
3.2.1. Summary statistics for open green space
The results from the veried model show that the total area of
open green space is 1422763.01 m2 , which is 35.16% of the total
urban area. The area per capita, according to the population of
30,000 set by Howard, is 47.43 m2 .
3.2.2. Accessibility analysis of open green space
The statistics show that the shortest walking time from a residential area to the open green space is zero (for residents living next
door to an open green space). Residents living in the farthest areas
have a travel time of 520 s (82 /3 min), and the average access time
is 186.77 s (about 3 min). The standard deviation of access time for
the entire area is 86.24 s.
Using a classication interval of 300 s (5 min), Fig. 4 shows a
thematic map of the spatial variation in accessibility. Most people
in the study area can access open space within 300 s, whereas parts
of the innermost and outermost rings have access times in the range
300520 s. According to statistics of spatial analysis tool in ArcGIS,
some 14725 cells are within 0300 s, accounting for 89.10% of the
four residential rings. Only 1747 (10.90%) cells have access times
greater than 300 s. Therefore, residents of the Garden City enjoy
fairly good spatial accessibility.
4. Discussion
Zhujiajiao Town in Shanghai won the International Award
for Livable Communities in 2008 (silver award for category B,
which is aimed at populations of 20,00075,000) (United Nations
Environment Programme, 2008). Therefore, we consider it a realistic reference for the digital Garden City model and its calculated
accessibility given its population and urban characteristics. We
modeled Zhujiajiaos township applying regular urban modeling
procedure of modern cities in GIS, and analyzed its accessibility
statistics in accordance with the process for the digital Garden City.
The base map for the digitization is a regulatory land-use planning map. Both the text of Zhujiajiaos Regulatory Detailed Planning
(20012020) and remote sensing image data obtained from Google
Earth were used as references.
4.1. Comparison of the modeling of classical theories and real
By comparing the modeling process of the Garden City theory and Zhujiajiaos township, we can identify several differences
between urban modeling in classical theories and the modern
world, especially in terms of the modeling foundation and emphasis, factors affecting model accuracy, and the requirements of
modelers (Table 3).
In addition, models based on theories may be limited because
the data needed has not been explicitly presented in the original

Z. Yuan et al. / Ecological Modelling 289 (2014) 2635


Fig. 3. From sketches (in Howards original work) to digital Garden City (in ArcGIS 10).

work. For example, Howard did not set the population distribution
or house heights for his Garden City. There are also some incompatibilities between theoretical and real cities in terms of the different
historical backgrounds. Thus, some land-use types do not exist in
Garden City, such as multiple public utilities, and some land-use
designs are no longer suitable for modern cities, such as ringshaped trafc systems which were designed for gharries but are
not suitable for cars. However, these non-existent land-use types
and unsuitable designs provide wide scope for city-related studies.
4.2. Similarity of land-use structures between Garden City and
There are differences between the geographical and historical
backgrounds of Garden Citys urban area and Zhujiajiaos township.
However, there are also similarities, as can be demonstrated by
land-use conditions for several urban functions. We divided landuse types of Garden Citys urban area and Zhujiajiaos township
based on land-use type division in Code for Classication of Urban
Land Use and the Planning Standards of the Development Land

(Ministry of Housing and Urban-Rural Development of the Peoples

Republic of China, 2011) and corresponding urban functions. Area
proportion and per capita indicators for land-use structure are compared (Table 4).
The residential, transportation, and ecology land-use types are
quite similar in both proportion and area per capita. The ecological land uses are mainly green space and water. Without water
design, the proportion of green space in Garden City is equal to
the sum of the water and green space in Zhujiajiao. The comparison of land-use types for other functions shows large differences.
For example, Garden City lacks public utilities, whereas Zhujiajiao
has no industrial and warehousing areas, which cause differences
in the land-use regions of multiple utilities, industrial areas, and
logistics and warehouses. As Zhujiajiaos central industry, leisure
tourism accounts for a larger proportion of commercial use than
the Crystal Palace in Garden City. The difference in administration and public services is partly caused by the number of sites
of cultural relics in Zhujiajiao, with the enhancement and development of government during the past 100 years also playing a


Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

Table 2
Accuracy verication of digital Garden City model.
Physical item


Data from original


Converted data
from original work

Measured data in
digital model

error (%)

Built area

Area (round)
Length (radius)

1000 acres
1240 yards
(approx. 3/4 miles)

4046856.42 m2
(1207.01) m

4046375.77 m2
1134.87 m



Length (width)

120 feet

36.58 m

36.58 m


Central garden

Area (round)

5.5 acres

22257.71 m2

22223.04 m2


Central park

Area (ring)

145 acres

586794.18 m2

586537.98 m2


Crystal Palace

Length (farthest distance to


600 yards

548.64 m

548.81 m


Residential area

Area (average area)

Area (the smallest area)
Count (residential population)

20 feet 130 feet

20 feet 100 feet

6.10 m 39.62 m
6.10 m 30.48 m

Not expressed
Not expressed
Not expressed
Not expressed

Grand avenue

Length (width)
Length (green belts length)
Area (ring)
Length (farthest distance to

420 feet
3 miles
115 acres
240 yards

128.02 m
4828.03 m
465388.49 m2
219.46 m

128.05 m
4827.73 m
465330.97 m2
201.46 m



Area (total)

4 acres

16187.43 m2

16187.44 m2


Warehouses, factories, and markets
Ring railway

Area (total)
Area (total)
Length (perimeter)
Length (total)

250 acres
82 acres
4.5 miles
25 miles

1011714.11 m2
331842.23 m2
7242.05 m
40233.60 m

1068058.05 m2
333530.21 m2
7186.66 m
38763.32 m


Areaaverage relative error

Lengthaverage relative error
Countaverage relative error

As a well-recognized modern garden city, the land-use structure

of Zhujiajiao reveals the demand of citizens for a livable and green
urban environment to some extent. The similarity in the land use of
Garden Citys urban area and Zhujiajiaos township demonstrates
that Howards Garden City is livable and green, under a suitable
land-use structure and area per capita, even though it was designed
more than 100 years ago.


4.3. Comparison of open green space accessibility in Garden City

and Zhujiajiaos township
4.3.1. Accessibility analysis of Zhujiajiaos open green space
The open green space in Zhujiajiaos township mainly consists of parks, squares, and structured green areas. We applied
the relative travel impedance of Garden City, adding a relative

Fig. 4. Classication map from the residential area to the open green space (interval: 300 s) of Garden Citys urban area.

Z. Yuan et al. / Ecological Modelling 289 (2014) 2635


Table 3
Comparison of modeling processes of classical theories and real cities.

Modeling cities in classical theories

Modeling real modern cities

Modeling foundation

Data list extracted from the text and diagrams

Modeling emphasis

Systematic understanding and presentation of the ideal

scattered original work (predictive studies are rare
because of a lack of reality and spatiotemporal variation)
Mostly controlled by data extraction and calculation,
accumulated errors mainly caused by contradictions
hidden in the original work or proposed assumptions,
careless geometric calculations, and data loss during
Capable of understanding the entire theory systematically
and recognizing the calculation sequence effectively
because of the implicit spatial logic

Planning scheme or interpretation of the remote sensing

image data
Prediction and spatiotemporal studies

Model accuracy

Requirements of modelers

Depends on the methods used in data transformation and

the interpretation of image data, errors caused by aerial
photo techniques and interpretation

Do not need to deal with the implicit spatial logic

Table 4
Comparison of land-use structures in Garden Citys urban area and Zhujiajiaos township.

Land-use types

Total area

Administration and public services
Commercial and business facilities
Logistics and warehouse
Street and transportation
Municipal utilities
Green space


Area per capita/m2


Garden City


Garden City


Garden City








Table 5
Relative travel impedance of different land-use types for walking in Zhujiajiao.
Land-use types

Physical items

Residential areas

Streets and roads

First class residential, second class residential, traditional residential, and parking
Green area for environmental protection, structured green area, parks, and squares
Administrative areas, public utilities, education areas, and cultural-relic sites
Leisure facilities, commercial areas, commercial-residential complex, tourism
featured business, external trafc areas, and municipal public utilities

Open green space

Administrative areas
Commercial- and industrial-related areas

impedance of 999 for water to represent the difculty of walking via water (Guo et al., 2012; Hu et al., 2005; Li et al., 2008)
(Table 5). Commercialresidential complexes were grouped into
commercial- and industrial-related areas, which have higher travel
impedances than residential areas.
According to our calculations, the shortest travel time from a
residential area to an open green space is again zero, whereas
the longest walking time is 219337.55 s (approximately 365 min).
This is much longer than the maximum travel time in Garden
City (520 s). The average access time in Zhujiajiaos township is
1038.04 s (approximately 17 min), which is not long but still large
compared with Garden City. The standard deviation of the access
time over Zhujiajiaos entire residential area is 2522.37 s. Classication of the residential area based on a 300-s time interval
shows that most residents can easily access open green space, but
some areas (such as the southwest corner) have poor accessibility
(Fig. 5).
4.3.2. Comparison of the accessibility of open green space
We compared the statistical results of a spatial analysis of the
two cities (Table 6). The results indicate that residents of Zhujiajiao
have good accessibility to open green space, with nearly half of the

Relative travel

residential area located within 300 s. However, the accessibility is

still lower than in Garden City. Only 77.98% of the residential area in
Zhujiajiaos township has access to open green space within 600 s,
and 8.91% of the area is more than 1800 s (30 min) away from an
open green space. Despite the worldwide recognition of Zhujiajiaos
green spaces, the planning and distribution of Garden Citys open
green space is superior in terms of its accessibility.

Table 6
Comparison of access-time interval proportions on open green space accessibility
in Garden Citys urban area and Zhujiajiaos township.
Time intervals/s


Area proportion
Garden City (%)

Zhujiajiao (%)





Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

Fig. 5. Classication map from the residential area to the open green space (interval: 300 s) of Zhujiajiaos township.

5. Conclusion


In this study, we built an accurate digital Garden City model

in ArcGIS. The modeling procedure consists of an analysis of the
original theory, the selection of an appropriate mathematical foundation, data extraction and calculation, the systematic organization
and design of an attribute structure, vectorization and digitization,
verication of the model accuracy, and corrections to the model.
Further quantitative studies on land-use structures and open green
space accessibility were conducted on the basis of the digital model.
Using the ofcially recognized garden city of Zhujiajiaos township
as a reference, we showed that Howards Garden City has a similar
land-use structure for residential, transportation, and ecology, and
superior open green space planning. Our research expresses a new
perspective for modeling and quantifying classical theory using GIS,
and reveals the great value of the digitization of classical theories
into models. The processes of data presentation and system design
help us understand the theories in a clearer and more systematic
way. In addition, characteristics that have been missed in previous
qualitative analyses can be elicited by quantitative calculations and
spatial analysis. Classical theories such as Garden City and other
historical planning cases presented in text and diagrams can be
restudied with the methods and creative perspectives promoted in
this research.
Our study rst systematically discussed the distinctions and difculties of modeling processes in classical theories and real cities,
and then proposed a complete modeling procedure for classical
theories. According to the urban functions of land-use types, we
discovered similarities between the residential, transportation, and
ecology land-use types, as well as distinctions between administrative and public services, in Garden City and Zhujiajiao. The study
also measured accessibility to open spaces, and demonstrated that
all residents in Garden City can walk to open green space within
9 min, which is much less than the maximum access time in Zhujiajiao. In future work, we will conduct further calculations with the
digital model (Yuan et al., 2013b), compare more cities from across
the world, and improve the model with regard to standardization,
openness, and non-exclusivity.

This study was supported by the National Natural Science Foundation of China under Grant (No. 40571119). We would like to
express our gratitude to the editors at Editage for polishing the
text and providing useful comments on this manuscript.
Alexander, E.R., 1992. Approaches to Planning: Introducing Current Planning Theories, Concepts and Issues, 2nd ed. Gordon and Breach, Yverdon, Switzerland.
Al-Hader, M., Rodzi, A., 2009. The smart city infrastructure development & monitoring. Theor. Empir. Res. Urban Manag. 4, 8794.
Beregovskih, A., Grishechkina, I., Shalakhina, D., Miller, S., Petrovich, M., 2010. Russian urban planning: modern ideas (legislative base, spatial planning, transport
and ecology, GIS technology). In: Schrenk, M., Zeile, P. (Eds.), 15th International
Conference on Urban Planning, Regional Development and Information Society.
Brockmann, D., Helbing, D., 2013. The hidden geometry of complex, network-driven
contagion phenomena. Science 342, 13371342.
Buder, S., 1969. Ebenezer Howard: the genesis of a town planning movement. J. Am.
Inst. Plan. 35, 390398.
Clark, B., 2003. Ebenezer Howard and the marriage of town and country: an introduction to Howards garden cities of to-morrow (selections). Organ. Environ. 16,
Crucitti, P., Latora, V., Porta, S., 2006. Centrality measures in spatial networks of
urban streets. Phys. Rev. E73, 036125.
Edwards, A.T., 1913. A criticism of the garden city movement. Town Plan Rev. 4,
Fotheringham, S., Rogerson, P., 1994. Spatial Analysis and GIS. Taylor and Francis,
Geurs, K.T., van Wee, B., 2004. Accessibility evaluation of land-use and transport
strategies: review and research directions. J. Transp. Geogr. 12, 127140.
Guo, W., Chen, P., Tang, J., Yu, L.S., 2012. Estimation of carrying capacity of urban
green space system based on landscape accessibility. Chin. J. Ecol. 31, 165171
(in Chinese).
Hall, M.H.P., 2011. A preliminary assessment of socio-ecological metabolism for
three neighborhoods within a rust belt urban ecosystem. Ecol. Model. 223,
Herold, M., Scepan, J., Clarke, K.C., 2002. The use of remote sensing and landscape
metrics to describe structures and changes in urban land uses. Environ. Plan. A
34, 14431458.
Howard, E., 1898. To-morrow: A Peaceful Path to Real Reform. Swan Sonnenschein,
Howard, E., 1902. Garden Cities of To-morrow. Swan Sonnenschein & Co., London.
Hu, Z.B., He, X.Y., Lu, Q.X., Chen, W., Li, Y.H., Liu, C.F., 2005. Green space accessibility
research based on GIS: taking Shenyang as an example. J. Shenyang Jianzhu Univ.
(Nat. Sci.) 21, 671675 (in Chinese).

Z. Yuan et al. / Ecological Modelling 289 (2014) 2635

Jiang, B., Yao, X., 2010. Geospatial analysis and modeling of urban structure and
dynamics: an overview. In: Jiang, B., Yao, X. (Eds.), Geospatial Analysis and
Modelling of Urban Structure and Dynamics, vol. 99. Springer, Netherlands, pp.
Jin, J., 2007. How to understand the Garden City. J. Beijing City Univ. 4, 112 (in
Kong, F., Nakagoshi, N., 2006. Spatial-temporal gradient analysis of urban green
spaces in Jinan, China. Landsc. Urban Plan. 78, 147164.
Li, B., Song, Y., Yu, K.J., 2008. Evaluation method for measurement of accessibility in
urban public green space planning. Acta Scientiarum Naturalium Universitatis
Pekinensis 44, 618624 (in Chinese).
Luo, W., Wang, F., 2003. Measures of spatial accessibility to health care in a GIS
environment: synthesis and a case study in the Chicago region. Environ. Plan. B:
Plan. Des. 30, 865884.
Malczewski, J., 2004. GIS-based land-use suitability analysis: a critical overview.
Prog. Plan. 62, 365.

Matejcek, L., Engst, P., Janour,

Z., 2006. A GIS-based approach to spatio-temporal
analysis of environmental pollution in urban areas: a case study of Pragues
environment extended by LIDAR data. Ecol. Model. 199, 261277.
Miller, M., 1989. Letchworth: The First Garden City. Phillimore, Chichester.
Ministry of Housing and Urban-Rural Development of the Peoples Republic of China,
2011. Code for Classication of Urban Land Use and Planning Standards of Development Land (GB 50137-2011).
Mumford, L., 1961. The City in History: Its Origins, its Transformations, and its
Prospects. Harcourt Brace, New York.
Oh, K., Jeong, S., 2007. Assessing the spatial distribution of urban parks using GIS.
Landsc. Urban Plan. 82, 2532.
Purdom, C.B., 1963. The Letchworth Achievement. JM Dent, London.
Reade, C.C., 1913. A defence of the Garden City movement. Town Plan. Rev. 4,
Reiss, R.L., 1920. Welwyn Garden City: illustrated. Town Plan. Rev. 8, 179.
Seto, K.C., Fragkias, M., 2005. Quantifying spatiotemporal patterns of urban land-use
change in four cities of China with time series landscape metrics. Landsc. Ecol.
20, 871888.


Talen, E., Anselin, L., 1998. Assessing spatial equity: an evaluation of measures of
accessibility to public playgrounds. Environ. Plan. A 30, 595613.
Tsou, K.-W., Hung, Y.-T., Chang, Y.-L., 2005. An accessibility-based integrated measure of relative spatial equity in urban public facilities. Cities 22, 424435.
United Nations Environment Programme (UNEP), 2008. The International Awards
for Liveable Communities-The Results 2008.
Van Herzele, A., Wiedemann, T., 2003. A monitoring tool for the provision of accessible and attractive urban green spaces. Landsc. Urban Plan. 63, 109126.
Ward, S.V., 1992. The Garden City: Past, Present and Future. E & FN Spon, London.
Yang, D., Kao, W.T.M., Zhang, G., Zhang, N., 2014. Evaluating spatiotemporal differences and sustainability of Xiamen urban metabolism using emergy synthesis.
Ecol. Model. 272, 4048.
Yin, H.W., Kong, F.H., 2006. Accessibility analysis of urban green space in Jinan (in
Chinese). J. Plant Ecol. 30, 1724.
Yu, K.J., Duan, T.W., Li, D.H., Peng, J.F., 1999. Landscape accessibility as a measurement of the function of urban green system. City Plan. Rev. 23 (8), 1143 (in
Yuan, Z.Y., Zheng, X.Q., Xue, C.L., 2013a. Modeling and application of digital garden city based on GIS. In: 2013 IET/IEEE International Conference on Smart and
Sustainable City (ICSSC 2013), Shanghai, China.
Yuan, Z.Y., Zheng, X.Q., Zhang, J.K.,2013b. GIS-based analysis on transportation network features of garden city model. In: 2013 the International Conference on
Remote Sensing, Environment and Transportation Engineering (RSETE 2013).
Atlantis Press, Nanjing, China.
Zhang, Y., Liu, H., Fath, B.D., 2014. Synergism analysis of an urban metabolic system: model development and a case study for Beijing, China. Ecol. Model. 272,
Zhou, T.G., Guo, D.Z., 2003. GIS-based study on spatial structure of urban greenbelt
landscapes: taking Ningbo City as an example. Acta Ecol. Sin. 23, 901907 (in
Zhou, L., Wang, T., Ma, N., Wang, X.S., 2008. Urban public green space accessibility
research based on GIStake Hankou of Wuhan as an example. Yunnan Geogr.
Environ. Res. 20 (11), 1520 (in Chinese).