Landscape and Urban Planning 100 (2011) 223–230

Contents lists available at ScienceDirect

Landscape and Urban Planning
journal homepage: www.elsevier.com/locate/landurbplan

A case study on the relation between city planning and urban growth using
remote sensing and spatial metrics
Hai Minh Pham a,∗ , Yasushi Yamaguchi a , Thanh Quang Bui b
a
b

Department of Earth and Environmental Sciences, Graduate School of Environmental Studies, Nagoya University, Japan
Department of Geography, Hanoi University of Science, Viet Nam National University, Viet Nam

a r t i c l e

i n f o

Article history:
Received 23 February 2010
Received in revised form
23 December 2010
Accepted 28 December 2010

Keywords:
Urbanization
Hanoi
Landsat
Image processing


∗ Corresponding author. Tel.: +81 52 789 3023; fax: +81 52 789 2523.
E-mail addresses: (H.M. Pham),
(Y. Yamaguchi), (T.Q. Bui).
0169-2046/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.landurbplan.2010.12.009

increase in urbanization and the concomitant effect that it has on
land use means that it is becoming increasingly for city planners to
adopt appropriate sustainable land use plans.
Planning and managing urban spaces depends on knowledge
of the underlying driving forces, combined with the chronology
and impacts of urbanization (Klosterman, 1999). City planners,
economists and resource managers therefore need advanced methods and a comprehensive knowledge of the cities under their
jurisdiction to make the informed decisions necessary to guide
sustainable development in rapidly changing urban environments.
Remote sensing provides spatially consistent coverage of large
areas with both high spatial detail and temporal frequency, which
is useful for examining historical time series (Jensen and Cowen,
1999). Moreover, remote sensing data is effective to monitor the
land use change in areas, especially where information on land
use management is inconsistent and insufficient. For example,
recently the economic development in Hanoi impacts the land use
change in the suburb occurring rapidly. With the current land use
map is updated every 5 years, local land use managers have not
enough information to monitor land use change of Hanoi. Therefore, with increased availability and improved multi-spatial and
multi-temporal resolution, remote sensing can now be applied to


224

have a particular relationship in terms of the topography. The rivers
running inside these cities as Red River (Hanoi), Connecticut River
(Hartford), and Hangpu River (Shanghai) separate them to East and
the West parts. The successful land use planning of Hartford and
Shanghais will be valuable for Hanoi to solve the gap of the urban
development process between the West and the East of Hanoi due
to the effect of the Red River. While the primary focus was on urban
development in Hanoi, we expected that the analysis of urbanization patterns in other cities is considered useful not only for Hanoi
but also for Vietnamese policy makers and related officials to have
appropriated local land use plans.
The surface land cover maps of the four cities were generated
from satellite images using the classification methods described in
Pham and Yamaguchi (2007) (Section 3.1). The ‘percentage of like
adjacencies’ (PLADJ) was then used to quantify urban fragmentation and to generate urban change pattern maps (Section 3.2.1).

The statistical program FRAGSTATS (McGarigal, 2002) was used
to perform the PLADJ analysis (Section 3.2.2). Finally, the urban
growth pattern maps and the FRAGSTATS results were used to analyze urban growth within the context of urban planning (Section
4). The results of this study are expected to assist local officials in
their understanding of urban dynamics, and in so doing, promote
future sustainable growth.
2. Study area and input data
2.1. Study area
Hanoi is the capital of the Socialist Republic of Vietnam (Fig. 1a).
It is an ancient city located on the banks of the Red River and
retains the Old Quarter, which has a history that spans 2000
years and represents the eternal soul of the city. Hanoi was
originally planned as a grid, with areas small residential houses
located along narrow streets. In 2005, Hanoi covered approximately 921 km2 (the study area covers 400 km2 ) and the population
numbered approximately 3.3 million (Hanoi Statistical Yearbook,

inner city has resulted in extensive pollution of the city environment ( One of the most
notable achievements of the city’s urban plan has been the construction of the new Pudong New Area, which is located on the east
of the Hangpu River, because it promotes the development away
from the city centre. In order to satisfy the transport demands of an
estimated 70 million visitors to the Shanghai 2010 Expo, a considerable amount of investment is currently being channelled into the
development of a transport infrastructure both within and between
cities in the region.
2.2. Data sources
Sets of multi-spectral and multi-temporal satellite data for
Hanoi, Nagoya, Hartford, and Shanghai were obtained for the
years 1975–2003 from the Tropical Rain Forest Information Centre, Michigan State University, USA (Table 1). Cloud cover was less
than 10% in all images and the visible and near infrared (NIR) bands
used for data processing were rectified geometrically to a common
Universal Transverse Mercator coordinate system.

225

Table 1
Data sources.
Hanoi
Nagoya
Hartford
Shanghai

1975(MSS), 1984(MSS), 1992(TM), 2001(ASTER), 2003(ETM+)
1975(MSS), 1985(TM), 1996(TM), 2002(ETM+)
1979(MSS), 1989(TM), 2002(ETM+)
1979(MSS), 1989(TM), 2001(ETM+)

3. Methodology

soil index has the advantage of being able to distinguish between
soil and water, it is difficult to differentiate between urban and
fallow areas. The soil index used in this study was derived from the
vegetation-soil-water (VSW) index of Yamagata et al. (1997). By
combining the two aforementioned supervision methods we were
able to delineate the urban areas and reduce the relative disadvantages associated with each of the methods. Based on the results,
the NIR band was then used to mask the contribution of water
bodies by diminishing the contribution of water to urban area
detection. Then, manually defined visual interpretation thresholds
were employed to extract the urban areas and to reduce the extent
of the areas that were affected by the mixel problem (Pham and
Yamaguchi, 2007). The results of this integration method were
validated against the reference data sources such as land use maps
published in 2002 for Hanoi and Nagoya. Comparison between
the results of classification, and reference data for certain sample
sites was conducted visually and interpreted quantitatively. The
mixel problem was thus considerably reduced and the integration
method provided a reliable approach for the detection of urban
areas. Finally, the results were classified into three categories:
urban, non-urban and water (Fig. 2).
3.2. Spatial metric calculations

Fig. 3. Example of counting PLADJ.

a maximum disaggregated pattern occurs in the current class or
when there are no like adjacencies, and equals 100 when the computed areas cover a single class or all adjacencies are in the same
class (maximally contagious). Low PLADJ percentages imply that
the extent of fragmentation is high or that there are many individual
urban units on the map. In order to discriminate between developed (urban) and non-developed (non-urban) pixels, a positive
PLADJ value was assigned to the centre pixel if it was originally nondeveloped and conversely, a negative PLADJ value was assigned to

i=1

m
g
i=1 ii
m
g
k=1 ik

(1)

where gii is the number of like adjacencies between pixels of patch
type i, gik is the number of adjacencies between pixels of patch
types i and k, and m is the number of pixels in the satellite image.
Firstly, a 5 × 5-pixel-moving window was used to compute the
percentage of urban fragmentation for the centre cell of the window
(Fig. 3). PLADJ moves randomly conditional probabilities through
the pixels in the moving window, with each calculation involving like adjacencies between four pixels; orthogonal cells were
counted, but diagonal cells were ignored. PLADJ equals zero when

Fig. 4. Urban heterogeneity in southern Hanoi in 2003 calculated by PLADJ.


H.M. Pham et al. / Landscape and Urban Planning 100 (2011) 223–230

227

scientific purposes, information derived solely from urban change
maps does not adequately explain the forces driving urbanization
and additional information is required to link the spatial structure


In order to visualise change patterns based on the PLADJ metric
results, this study utilised the landscape transformation scheme
presented by Forman (1995). Using this method, three urban
growth patterns were adopted to describe and map urban sprawl:
infill, expansion, and outlying. The infill pattern is mostly encountered inside the existing developed areas, while the expansion
pattern dominates the urban fringe. The outlying pattern tends to
occur some distance from the existing developed areas.
The result is a series of maps illustrating the changes in the urban
structure of four cities from 1975 to 2003, which are discussed further in Section 4 (Fig. 5). These maps provide valuable reference
information for city planners because they can be used to illustrate
the historical evolution of a particular urban area. However, for

The urban land use plan for Shanghai was designed to transform the city from being mono-centric to a multi-centric metropolis
in order to decentralize the population and economic activities
(Haixiao, 2000). As reported by Haixiao, satellite towns have been
planned so that the suburbs will absorb the development potential of the central city of Shanghai. As can be seen in Fig. 2, the
satellite towns of Shanghai have had a significant impact on the
progress of urban growth and urbanization in the city. Based on
the slight increase in the NP (Fig. 6), the development of Shanghai
over the period 1979–1989 was characterized by moderate growth
of the urban patches. While the central urban area changed slowly,
there was a rapid increase in size of the satellite towns (Fig. 2b4).
This observation suggests that, by restricting the development of
existing urban areas, the government promoted the development
of metropolitan areas on the city fringe. Furthermore, the mass
transport lines that were constructed to link the satellite towns
to the urban core from 1989 to 2001 may have been a key factor contributing to the rapid expansion of the urban areas in the
region. The spatial characteristics of the urban areas of Shanghai
had become increasingly complex by 2001, which correlated with

ED equals the sum of length (m) of all edge segment involving the
urban patch type, divided by the total landscape area (m2 ),
multiplied by 10,000 (to convert to hectares).
LPI equals the area (m2 ) of the largest patch of the corresponding
patch type divided by total area covered by urban land type (m2 ),
multiplied by 100 (to convert to percentage).
MNN equals the distance (m) mean value over all urban patches to
the nearest neighbouring urban patch.
Area weight mean value of the fractal dimension values of all
urban patches, the fractal dimension of a patch equals two times
the logarithm of patch perimeter (m) divided by the logarithm of
patch area (m2 ).

and farmland (non-developed areas) was converted to urban use;
however, the improved infrastructure and transportation improved
urban living standards which contrasted with that in the central
part of the city.
The urban structure of Hartford follows the Concentric Zone
Model (Robson, 1969). The urban areas were classified to zones,
such as the Central Business District (in the central part of the city),
Transitional Zone, Working Class Zone, Residential Zone, and Commuter Zone. The Transitional Zone includes factories, Working Class
Zone includes single family tenements, Residential Zone includes
single family homes, yards and garages, and the Commuter Zone
consists of the suburbs. As opposed to focussing on the expansion
of satellite towns (as in Shanghai), urbanization in Hartford was
characterized by the outlying pattern from 1975 to 1989 (Fig. 5),
with pronounced urban development occurring on both sides of
the Connecticut River. Most of this new development activity arose
through the conversion of vacant land along the periphery of the
city near the major transportation routes and far away from the

Range

Hectare

CA > 0, no limit

None
Metres per Hectare

NP ≥ 1, no limit
ED ≥ 0, no limit

Percent

0 < LPI ≤ 100

Metres

MNN > 0, no limit

None

1 ≤ AWMPFD ≤ 2

agricultural land in the urbanization control zone. The expansion
and occurrence of outlying development in the western areas of
the city resulted in the size of the urban area increasing, which was
indicated by a peak in the CA in conjunction with an increase in
the ED. The urban growth of Nagoya started declining from 1985;
instead, from 1985 to 1996, urbanization shifted to the eastern

informal settlements adjacent to industrial zones, transport hubs,
and major markets along the city fringes. Within the context of official housing policy, these areas were considered to be illegal urban
areas (Do, 2007). Fig. 2b1 and c1 shows urban development in the


H.M. Pham et al. / Landscape and Urban Planning 100 (2011) 223–230

10000
Hanoi
Hanoi
1000

Metric value

100

10
1
1975
CA

1984
NP

LPI

1992

2001



2002

AWMPFD

MNN

Hartford

Metric value

1000
100
10
1
1989

1979
CA

NP

LPI

ED

2002
AWMPFD

MNN

urban centres over time, the 1984 and 1992 maps show that urban
expansion has occurred in two directions, one to the west and a linear branch-type to the south. While expansion to the west was the
predominant trend, the construction of the first national highway
to the south of the city resulted in the linear expansion of the city to
the south. In both cases, there was an increase in the development
of urbanized patches occurred some distance from the urban core; a
rise in the ED and a corresponding decrease in MNN confirmed this
trend. At this time, the urbanization of Hanoi was also characterized
by the development of urban areas along newly constructed roads
and highways. The area of the inner city transportation infrastructure increased from approximately 3000 ha in 1975 to 5000 ha in
1992 (Pham and Yamaguchi, 2007), implying that new road construction was a powerful catalyst for the urbanization to the west

229

and south of the city. The rate of urbanization decreased after 1992
due to the economic recession in Vietnam and the 1997 economic
crisis in Southeast Asia (Berg et al., 2003). Even so, despite the economic recession, approximately 700,000 immigrants moved to the
city at this time. Indeed, the peak observed in the value of NP in
2001 indicates a steady expansion in the size of the urban areas
in Hanoi. In order to promote decentralization of the city centre,
numerous apartment buildings were constructed in the suburbs.
Taken together, 20,000 ha of agricultural land and water bodies
areas were converted to urban use (Pham and Yamaguchi, 2007)
and the urban areas continued to expand along highways to the
south and south-west of the city. The decline in the MNN value
at this time reflected an increase in both the size and the extent
of fragmentation of the urban area, and the observed peak in the
AWMPFD and ED indexes in 2001 support this trend (Fig. 6). Despite
the observed decrease in the rate of urbanization since 2001, the
urban core coalesced with the various fragmented urban patches

consider the problem of mixed pixels, resulting in the loss of spatial information. However, in this study, the accuracy of urban area
classification was improved by applying the classification method
developed by Pham and Yamaguchi (2007). Using these results, we
were able to examine the changes in the urban land use of four cities
over time. In 2003, the urban areas of Hanoi and Shanghai underwent considerable expansion into the suburban areas, whereas the
direction of urbanization in Hartford seemed to occur from the
periphery of the city toward the city centre. Interestingly, the spatial characteristics of the urban areas around Nagoya varied only
slightly over time.


230

H.M. Pham et al. / Landscape and Urban Planning 100 (2011) 223–230

The results of this study show the relationship between certain
changes of spatial metric parameters and a particular type of city
planning. Fig. 6 clearly highlights this conclusion. The establishment of the urbanization control zones of Nagoya’s land use plan
was reflected by slight change of spatial metrics over time. On the
other hand, the establishment of satellite towns around the existing city centre of Hanoi and Shanghai resulted in the border of the
cities getting larger, which was demonstrated by an increase in ED
and LPI. The rapid development of Residential Zone in the suburb
of Hartford contributed to a sudden increase in CA.
The land use master plans of each city are important for guiding
their future urban expansion. We demonstrated that the legislative
instruments related to land use in urban areas have a significant
affect on the patterns and nature of urbanization; this was particularly apparent in contrasting urban development scenarios in Hanoi
and Nagoya. In Nagoya, the existence of a well-defined master plan
with its provisions for urban control and promotion zones resulted
in only minor changes in the city fringe. In contrast, urban development in Hanoi was less orderly, occurring mainly on the western
side of the Red River and along major transportation routes. As a

of Hanoi using multi-temporal and multi-sensor remote sensing data. In: Paper
Presented at Workshop on GIS-IDEAS 2002 Symposium , Hanoi, Vietnam.
Forman, R.T.T., 1995. Land Mosaic: The Ecology of Landscapes and Regions. Cambridge University Press, Cambridge.
Haixiao, P., 2000. Shanghai from Dense Mono-center to Organic Poly-center
Urban Expansion (accessed 15.11.09) />envirolib/.
2005. Hanoi Statistical Yearbook. Hanoi Statistical Office Press, Data received on
December 2007.
Henebry, G.M., Goodin, D.G., 2002. Landscape trajectory analysis: toward spatiotemporal models of biogeochemical fields for ecological forecasting. In: Paper
Presented at Workshop on Spatio-temporal Data Models for Biogeophysical Fields , La Jolla, CA, USA (accessed 02.11.09) />BDEI/papers/henebry goodin position.pdf.
Herold, M., Goldstein, C., Clarke, N.C.K., 2002. The spatial form of urban growth:
measurement, analysis and modelling. Remote Sens. Environ. 86, 286–302.
Hiep, 2009. The Most Expensive Road in Hanoi (accessed 05.06.09) http://www.
laodong.com.vn.
Jensen, J.R., Cowen, D.C., 1999. Remote sensing of urban/suburban infrastructure and
socio-economic attributes. Photogramm. Eng. Rem. Sens. 65, 611–622.
Klosterman, R.E., 1999. The What if? Collaborative planning support system. Environ.
Plann. B 26, 393–408.
McGarigal, K., 2002. FRAGSTATS: Spatial Pattern Analysis Program for Categorical
Maps
(accessed
10.02.08)
/>research/fragstats/fragstats.html.
Noda, A., Yamaguchi, Y., 2008. Charactering urban sprawl using remote sensing, GIS
and a spatial metric for a medium sized city in Japan. Int. J. Geoinform. 4, 43–50.
O’Neill, R.V., Krummel, J.R., Gardner, R.H., Sugihara, G., Jackson, B., Deangelis, D.L.,
Milne, B.T., Turner, M.G., Zygment, B., Christensen, S.W., Dale, V.H., Graham, R.L.,
1998. Indices of landscape pattern. Landscape Ecol. 1, 153–162.
Pham, M.H., Yamaguchi, Y., 2007. Monitoring land cover change of the Hanoi city
centre under impacts of urbanization by using remote sensing. Int. J. Geoinform.
3, 55–61.