You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254442295

Analysing Evolution of Urban Spatial Structure: A Case Study of Ahmedabad,


India

Article in Environment and Planning B Planning and Design · September 2011


DOI: 10.1068/b36088

CITATIONS READS

26 11,473

1 author:

Bhargav Adhvaryu
Ahmedabad University (Amrut Mody School of Management)
42 PUBLICATIONS 247 CITATIONS

SEE PROFILE

All content following this page was uploaded by Bhargav Adhvaryu on 19 May 2015.

The user has requested enhancement of the downloaded file.


Environment and Planning B: Planning and Design 2011, volume 38, pages 850 ^ 863

doi:10.1068/b36088

Analysing evolution of urban spatial structure: a case study


of Ahmedabad, India

Bhargav Adhvaryuô
The Martin Centre for Architectural and Urban Studies, Department of Architecture,
University of Cambridge, 1 ^ 5 Scroope Terrace, Cambridge CB2 1PX, England and Churchill
College; e-mail: bhargav@cantab.net, ba247@yahoo.com
Received 21 July 2009; in revised form 21 December 2010

Abstract. This paper analyses trends in evolution of the urban spatial structure (or urban form) of the
city of Ahmedabad, with two key objectives in mind: to generate a quantitative understanding of
the evolution of the spatial structure and use such a quantitative understanding to inform the
formulation of alternative planning policies for the future. Time-series population data from
the Census of India over three decades have been used, which are usually available in most developing
countries. In the case of Ahmedabad, the city exhibits a gradual tendency of dispersal, although
compared with some other mid-sized metropolitan areas of the world, it is relatively compact.
A discussion on the application of the spatial structure analysis to the formulation of alternative
planning policies is included. Such analyses for similar cities in India and the developing world in
general could be carried out to produce a useful catalogue of cities. However, this is not the objective
of the paper, but in due course such literature could be built up.

1 Introduction
There are many measures available for analysing the urban spatial structure. However,
three key measures that fit the available data have been included in this analysis:
density gradient, dispersion index, and the concentration ^ deconcentration measure.
Their application to Ahmedabad, preceded by brief theoretical underpinnings of each
measure, is explained in sections 3 ^ 5. A discussion on the use of such spatial indica-
tors to inform the formulation of alternative planning policies is included followed by
conclusions in the last section. Section 2 provides a brief introduction to Ahmedabad.
This paper is based on my doctoral work (Adhvaryu, 2009).

2 A brief introduction to Ahmedabad


Ahmedabad is the biggest city in the state of Gujarat in western India (see figure 1).
Based on the 2001 Census, Ahmedabad is the seventh largest city in India (3.5 million)
considering the population within the municipal boundary of 2001 (191 km2 ), with an
average household size of 5.08. Its population in 2001 for the urban complex area
(around 510 km2 ) was about 4.5 million.
Until the 1980s Ahmedabad was known as the Manchester of the east, but textile
industries declined in the late 1980s. Based on data from the 2001 Census (AMC, 2007),
the economy is dominated by the service (tertiary) sector comprising about two
thirds of the share. Information-technology-enabled services (call centres, medical
transcription, etc), and banking and finance sectors have been rapidly growing in the
last decade or so. The manufacturing (secondary) sector has nearly a third of the share,
dominated by chemical and pharmaceutical industries. The share of the primary sector
is under 1%.

ô Current address: Faculty of Technology, CEPT University, Kasurbhai Lalbhai Campus, University
Road, Ahmedabad 380009, Gujarat, India.
Analysing evolution of urban spatial structure 851

Figure 1. [In colour online.] Location of Ahmedabad. (Source: the diagram was created by the
author based on images obtained from Wikipedia. The images have permission to be copied,
distributed, and/or modified under the terms of the GNU Free Documentation License.)

The population data for census wards for the years 1971, 1981, 1991, and 2001 have
been used in the analysis (AMC, 1979; 1990; 2007). The main city area, known as the
Ahmedabad Municipal Corporation (AMC) consists of forty-three census wards
(191 km2 ).(1) The area beyond the city's municipal boundary of, which is very much
part of the city in terms of employment and other linkages and is experiencing high
growth pressure, consists of forty-seven census wards (319 km2 ). This peripheral area
together with the AMC is known as the Ahmedabad Urban Complex (AUC) area (see
figure 2).

3 Density gradients (1951)


Clark (1951) in his seminal paper ``Urban population densities'' introduced the very
useful concept of a density gradient. His study revealed two very important general-
sations: (a) that density falls exponentially as one moves away from the centre of the city
and (b) as time goes by, the density tends to fall in the most populous inner suburbs and
to rise in the other suburbs, and the whole city tends to `spread itself out'.
Clark expressed the mathematical relation between residential density and distance
from the city centre as follows:
Dx ˆ D0 exp ÿgx† , (1)

(1) The AMC boundary was extended in 2006 to cover an area of 464 km2. However, this study used

pre-2006 AMC boundary to avoid spatial incompatibilities with 2001 Census data.
852 B Adhvaryu

Figure 2. [In colour online.] Ahmedabad Municipal Corporation (AMC) and Ahmedabad Urban
Complex (AUC).
Analysing evolution of urban spatial structure 853

where
Dx is the density of resident population (persons per unit area);
x is the distance from the city centre;
D0 is a coefficient that indicates the theoretical density (or in other words, the degree
of overcrowding, which the citizens are prepared to `tolerate' at the centre of the
city);
g is a coefficient that measures the rate of decline of density.
Coefficients D0 and g are determined using linear regression by transforming equation (1)
into a linear form as follows:
ln Dx ˆ ln D0 ÿ gx . (2)
Clark's study of nearly twenty cities at various points in time revealed that the above
relationship holds true at all times. A high (absolute) value of g means that the density
will decline sharply with increasing distance from the city centreöthat is, a compact
city; and a low value of g means that the density declines more slowlyöthat is, the city
is more `spread out'. He concluded that g is dependent largely on the intraurban
transport costs, or more precisely the cost of travelling in relation to the average
citizen's income (observations on this assumption are discussed at the end of sec-
tion 4). If the transport costs are low, the city can afford to `spread out'. It should be
noted that D0 is a hypothetical rather than an actual figure, because in fact the centre
of the city is occupied by businesses with few or no residents. He maintains, however,
that D0 remains a useful figure as it shows the point to which the densities are tending,
should the densities of the inner residential suburbs be extrapolated to reach the centre
of the city.
Clark concludes that looking at the values of g and D0 enables us to give a
simplified classification of cities and their trend of growth. If a city has fairly high
transport costs then it will have a fairly high value of g and, as its total population
grows, the value of D0 (that is, the degree of overcrowding at the centre) must
necessarily rise. Conversely, if a city has to have high total population, it must either
put up with a considerable degree of overcrowding or it must spread out itself.
Density gradient is a powerful tool to describe trends in the spatial pattern of a city
with regard to population. Clark maintains that looking at the value of g over time for
a city enables us to study the manner in which a city's spatial structure is influenced by
changes in transport costs. To understand this point better consider a hypothetical
situation of a city with population increasing over time, with the only other variable

7.5

7.0

6.5
ln Dx

6.0

5.5

5.0
0 5 10 15 0 5 10 15
Distance from centre, x (km) Distance from centre, x (km)

Time t ‡ n Time t Time t ‡ n Time t


(g ˆ 0:08† (g ˆ 0:11) (g ˆ 0:10) (g ˆ 0:11)
(a) (b)
Figure 3. Hypothetical density gradients over time: same city, two scenarios: (a) transport costs
reduce, city spreads out; (b) transport costs increase, density increases at all points.
854 B Adhvaryu

being the transport costs. If we imagine a scenario where transport costs reduce over
time the city would spread out with pivoting of density curves [see figure 3(a)]. On the
other hand [figure 3(b)], if transport costs increase over time then density at all points
increases, causing the curve to shift upwards. Of course, in this example it is assumed
that all other things in the two scenarios, except transport costs, are the same.
However, in the real world, in addition to transport costs, other topographic, societal,
and cultural factors could also be responsible for the shape of a city. The point is
revisited at the end of section 4.
Density gradients g along with the R 2 -values and D0 for the AMC area are presented
in table 1 and figure 4. It can be observed that density gradients in Ahmedabad are
becoming flatter over the years, and have a reasonable fit with the exponential model
as suggested by Clark (1951). Also, as predicted by Clark, the pivoting of density
gradients is evident in Ahmedabad. In other words, this indicates a gradual decline
in city centre densities: that is, people moving out from more central areas to suburbs
as the city spreads out. This is supported by observed data (see table 1), wherein the
actual densities for the ward considered as the centreöthat is, ward 2 (see figure 2) or
Table 1. Density gradients and densities for the Ahmedabad Municipal Corporation area.

1971 1981 1991 2001

g 0.50 0.40 0.26 0.19


R2 0.56 0.63 0.60 0.42
D0 theoretical 769 798 641 565
D0 actual (central ward, 2) 672 603 513 455
D0 actual (city centre) 658 645 551 515
Note: Densities are in persons per hectare.

4
ln (Dx )

0
0 2 4 6 8 10
Distance from centre, x (km)

ln (Dx )71 ln (Dx )81 ln (Dx )91 ln (Dx )01


1971 1981 1991 2001
y ˆ ÿ0:4996x ‡ 6:6455 y ˆ ÿ0:393x ‡ 6:6826 y ˆ ÿ0:2583x ‡ 6:4633 y ˆ ÿ0:1854x ‡ 6:336
2 2 2 2
R ˆ 0:5586 R ˆ 0:6262 R ˆ 0:5959 R ˆ 0:4228

Figure 4. Density gradients for Ahmedabad Municipal Corporation area.


Analysing evolution of urban spatial structure 855

indeed even for the entire city centre,(2) öhave been consistently declining. It should be
noted that the theoretical densities would be greater than actual values as a substantial
proportion of the centre has nonresidential activities.
By this measure, Ahmedabad has been exhibiting a tendency toward `spreading
out'. In general, this could be attributed to a reduction in the generalised cost of travel.
Mills and Tan (1980) looked at density gradients for several cities in developed and
developing countries and concluded that indeed cities in developing countries also exhibit
a decline in the density gradient, although not as fast a decline as cities in developed
countries owing to lower incomes and poorer transport infrastructure. The other strong
observation that can be made from figure 4 is that the points above the density gradient

1971 1981

1991 2001

Persons per hectare (gross)

< 200 AUC boundary (510 km2 )


200 ^ 399
400 ^ 599 AMC boundary (191 km2 )
600 ^ 799
5 800

Figure 5. Population density for Ahmedabad Urban Complex (AUC), 1971 to 2001 (AMCö
Ahmedabad Municipal Corporation).

(2) the city centre is the area covered by wards 1 ^ 6 (about 7:2 km2 ; see figure 2) popularly known

as the Walled City (or Old City) of Ahmedabad.


856 B Adhvaryu

lines in the 4 ^ 6 km band from the CBD correspond mostly to the poorer wards in
Eastern Ahmedabad, where households are generally observed to be living at higher
densities (see maps of density for 1971 to 2001 in figure 5). It should be noted that some
of the data points in figure 4 beyond 6 km for 1971 have ln Dx values in the region of 1.0.
A key reason for this is the change in census ward boundaries creating a discrepancy in
the data for subsequent years.
Clark observes that the density gradient is dependent largely on the intraurban
transport costs, or more precisely the cost of travelling in relation to the average
citizen's income, and that a decrease in the latter would mean a flattening out of
density gradients. The increase in vehicle ownership (expressed by a proxy measure,
vehicles per 1000 population) could be attributed to a decrease in the generalised cost
of travel. Therefore, its relationship with density gradient should be strong. And,
indeed, it can be seen from figure 6 that density gradient [that is, g in equation (1)
and table 1] and vehicle ownership exhibit a very strong relationship.

y ˆ 0:483 exp ( ÿ 0:0043x)


0.50 1971 R 2 ˆ 0:9409

0.40 1981
Density gradient

0.30

1991
2001
0.20

0 50 100 150 200 250


Private vehicles per 1000 population

Figure 6. Density gradient versus vehicle ownership for Ahmedabad Municipal Corporation.

4 Dispersion index
In 1999 Bertaud and Malpezzi (Bertaud, 2001) proposed a measure to describe the
`shape performance' of a city. Bertaud (2001) argues that the spatial structure of a city
can be defined by two complementary components: (a) the distribution of population
over space and (b) the pattern of trips made by people from their residences to any
other destination. The spatial distribution of population can be simply mapped using
census data for the wards in a city. This graphical representation could be further
enhanced by creating a three-dimensional map wherein the height of each of the wards
is scaled to its population (eg, see Bertaud, 2001, page 3, figure 1).
Bertaud (2001) maintains that the pattern of trips could be encapsulated in the
average distance per person to the centre.(3) This is a weighted average using popula-
tion of each ward as the weight. Bertaud (2001) argues that all else being equal, in a
city with a small built-up area the distance per person to the centre will be shorter
than in a city with a larger built-up area. Therefore, in order to have a comparative
measure of shape between cities, it is necessary to have a measure independent of the
(3) It is suggested that the centre in the case of a monocentric city is the CBD, and in the case of a

polycentric city, the centre of gravity (ie, the geometric centre of the entire city).
Analysing evolution of urban spatial structure 857

area of the city. This could be achieved by taking the ratio of the average distance per
person to the centre and the average distance per person to the centre of a circle
whose area would be equal to the built-up area. Such a measure, called the dispersion
index (4) r can be mathematically expressed as:
X   1=2 X 
A
r ˆ d i wi 2
, or r ˆ d w
i i
2
r , (3)
i
3
p i
3

where
di is the distance of the centroid of the ith tract (or ward or zone) from the CBD,
weighted by the tract's share of population wi ;
A is the built-up area of the city;
r is the radius of a circle with area A.
The numerator (that is, the actual distance) in equation (3) is the average distance
per person to the centre (CBD or the geometric centre as the case may be) and the
denominator (that is, the theoretical distance) is the average to the centre of a circle
(or cylinder with unit height) of equivalent area and uniform population density [for a
graphical description of this see Bertaud and Malpezzi (2003, page 64, figure 6)].
The dispersion index reflects the `shape performance' of a city: that is, how com-
pact or dispersed a city is. Since it is independent of the area and the density of a city, it
is a useful measure to compare cities of very different shapes, sizes, and densities at a
point in time or the same city over time. Bertaud (2001), of course, argues that it is not
suggested that a circular city is somewhat optimal, but merely suggests that some cities
will be more compact (those which have a lower value of r) and some will be less
compact (those which have a higher value of r), implying the value of 1.0 as the
watershed between compactness and dispersion.
Dispersion indexes for the AUC area are shown in table 2. The denominator of
equation (3), which is the average distance per person to the centre of equivalent radius
of the AUC area, turns out to be 8.49 km. Thus, although in absolute terms the city is
compact, it has indeed exhibited a tendency of dispersing over the thirty years to 2001.
Obviously, this trend is likely to continue in the future, but the rate is decreasing as
seen from the percentage change. The growth rate in terms of area from 1972 to 1996
turns out to be 3.1% per annum, while the population growth rate for the period 1971
to 2001 is around 2.7% per annum (spatial comparison of the two is shown in figure 7).
The ratio of these two numbers is defined as the Rusk sprawl index,(5) which turns out
to be 1.15, corroborating the tendency of Ahmedabad towards dispersal. For compar-
ison, dispersion indexes (along with other relevant items) for seven megacities of the
world are presented in table 3. Although the exact date of population values used in
the analysis is not mentioned in Bertaud's (2001) paper, from the date of publication
it could be assumed to be for or before 2001, and therefore are comparable with
the Ahmedabad 2001 value of 0.63. However, compared with other megacities of the

(4) The dispersion index (earlier known as compactness index) was first defined by Bertaud and
Malpezzi in 1999 as: ``All else being equal, a city shape which decreases the distance between
people's residence and the main place of work and consumption will be more favourable to the
functioning of labour and consumer markets. For a given built-up area, the shorter the average
distance per person to the main place of work or the main commercial areas, the better would be
the performance of the city shape'' (Bertaud, 2001, pages 7 ^ 8).
(5) ``The Rusk sprawl index, defined as the percentage change in urbanized land area divided by the

percentage change in urbanized area population, is easy to calculate and seeming straightforward
to interpret. On its face, the index being greater than one seems to be clear evidence of declining
population density, leapfrog development, transformation of rural to urban landöwell, sprawl''
(Bogart, 2006, page 57).
858 B Adhvaryu

Table 2. Dispersion index for Ahmedabad Urban Complex (AUC).

1971 1981 1991 2001

Weighted average distance per person to centre 3.87 4.37 4.93 5.39
(DA) (km)
Weighted average distance per person to centre 8.49 8.49 8.49 8.49
(DT) of circular area equal to AUC (km)
Dispersion index r (DA/DT) 0.46 0.51 0.58 0.63
Percentage change 13.1 12.7 9.4
Population (millions) 1.87 2.65 3.46 4.47
Average density (persons per ha) 37 52 68 88

Table 3. Dispersion index for seven megacities of the world [source: Bertaud (2001), except
Ahmedabad values, which are from table 2].

City DA DT r Population Built-up Average


(km) (km) (DA/DT) (millions) area density
(km2 ) (persons
per ha)

Berlin 12.65 12.90 0.98 4.212 1 176 36


Jakarta 17.38 20.40 0.85 14.908 2 942 51
(metropolitan area)
London 12.63 12.26 1.03 6.626 1 062 62
Moscow (municipality) 10.57 8.44 1.25 8.497 503 169
New York 18.26 19.45 0.94 10.753 2 674 40
(metropolitan area)
Paris 10.24 11.51 0.89 7.998 937 85
(metropolitan area)
Shanghai (city proper) 4.70 5.88 0.80 7.397 244 303
Ahmedabad 5.39 8.49 0.63 4.465 510 88
(urban complex)
Note: DA is the weighted average distance per person to the centre. DT is the average distance
to the centre of a circle (or cylinder with unit height) of equivalent area and uniform
population density.

world, Ahmedabad was still (in 2001) much less dispersed. An important conclusion
Bertaud (2001) draws from table 3 is that a high-density city does not always result in
lower average trip lengths (to the centre). For a graphic of the densities profiles of the
cities listed in table 3, see Bertaud (2001, page 14, figure 6).
Clark (1951) theorised an exponential relationship between residential population
density and distance from the city centre. In the seven megacities studied by Bertaud
(2001) the profiles do follow a general pattern of density decline as one moves away
from the city centre. However, this is not always true. For example, New York has a
subpeak at about 13 km from the city centre and so does Berlin at about 21 km.
In London the peripheral areas seem to have higher density. More radically, Moscow
exhibits a completely reversed pattern. An important point that can be drawn from
such patterns is that the relationship between (population) density and distance
(from the city centre) is influenced by the topographical conditions, political milieu,
sociocultural aspects, and planning policy constraints (including nature of land
market), in addition to transport costs as Clark assumed. However, the current study
does not investigate the combination of such societal factors and its influence on the
structure of the relationship between density and distance, which could itself be taken
up as a major research project.
Analysing evolution of urban spatial structure
1972 (93.67 km2 ) 1972 ^ 75 (110.70 km2 )
2
1975 ^ 82 (134.40 km ) 1982 ^ 89 (161.13 km2 )
1989 ^ 96 (196.03 km2 )
(a) (b)
Figure 7. [In colour online.] Map of change in (a) urban footprint (source: AUDA, 1997) and (b) population (source: author's calculation).

859
n:/psfiles/epb3805w/
860 B Adhvaryu

5 Concentration ^ deconcentration measure


Gayda et al (2003),(6) introduced a measure called the concentration ^ deconcentration
measure (referred to as the H indicator) to quantify concentration or deconcentra-
tion effects in the urban system. They opine that urban sprawl has a spatial and
temporal dimension and therefore it is worthwhile to analyse the process of creation
of sprawl by introducing a general spatiotemporal typology. The H indicator is based
on the observation that the process of densification is rather inhomogeneous in space
and depicts quite different growth rates in time and space. Furthermore, the densifica-
tion and the effect of urban sprawl is a phenomenon occurring rather on the margin of
a city, than homogeneously distributed over the total area under consideration.
Inspired by physics, the H indicator (in discrete terms, that is, if the area under study
is divided into n zones) is defined as:
X X
H ˆ ri ri2 Ai , or H ˆ Pi ri2 , (4)
i i

where
Pi is the population of the ith zone;
Ai is the area of the ith zone;
ri is the population density of the ith zone (ie, Pi =Ai );
ri is the straightline distance of centroid of the ith zone from the centroid of the CBD
(or centroid of the study area).
By definition, if H is increasing over time, it can be concluded that the city is
dispersing and vice versa. Further to this, a relative concentration measure H rel
is introduced to assess how outer areas P are changing in relation to central areas.
H rel is calculated by using Pi =[(1=n) i Pi ] instead of Pi in equation (4). If H rel is
increasing over time, it indicates that the urban periphery is growing faster than the
urban centre, in relative terms. On the other hand, if H rel is decreasing over time, this
indicates that the urban centre is growing faster than the urban periphery. If all zones
in a study area increase by a uniform rate (that is, if the growth rate is independent of
the distance), then H rel will be identical for those two time periods. H rel is therefore an
important measure for separating general growth effects of the city from sprawling
(or compaction) effects).
Lastly H^ (or H^ rel ) can be calculated as the percentage difference between H (or H rel )
for the year under question and the start year of analysis. If H^ > 0, then dispersion is
likely to occur, and for H^ < 0 concentration effects may dominate. It should be noted
that employment data can also be used in equation (4) to draw conclusions on con-
centration ^ deconcentration of employment. However, the employment H indicator for
Ahmedabad has not been calculated as time-series data were not available.
Table 4 shows the H and H rel for AUC, which indicate that Ahmedabad does
exhibit a dispersing tendency. The increase in H rel indicates that outer areas have
been growing at a faster rate than inner areas. In addition, both H^ and H^ rel turn out
to be positive and have been increasing, implying that dispersion is likely to continue in
the future. Gayda et al (2005) studied six major cities in Europe (Milan, Bristol,
Stuttgart, Brussels, Helsinki, and Rennes). Out of these, the strongest population
decentralisation effect was exhibited in Milan with highest per annum percentage change
in H rel (about 0.9%) over a thirty-year period to 2001. Ahmedabad, by comparison, in
the same period has shown a per annum rate of 1.6%.

(6) Thepaper mentions that this work draws from Deliverable 3 of SCATTER ``Statistical Analysis
in the six Case Cities'', December 2002, whose main authors are Guenter Haag and Jan Binder,
from STASA (Stuttgart).
Analysing evolution of urban spatial structure 861

Table 4. H and H rel indicators for Ahmedabad Urban Complex.

Item 1971 1981 1991 2001

H indicator (millions) 42 71 109 161


H rel indicator 2 027 2 398 2 825 3 244
H^ (%) 0.0 67.6 157.6 281.4
H^ rel indicator (%) 0.0 18.3 39.4 60.0

6 Application to formulating alternative planning policies


Apart from providing a quantitative understanding of the evolution of the spatial
structure of an urban area, the spatial analysis tools discussed here could also be
used in formulating inputs for testing alternative spatial planning policies. For example,
this analysis of Ahmedabad was used to inform the process of refining and fine-tuning
alternative planning policies, which were tested using a simplified planning model
called SIMPLAN.(7) Dispersion indexes and H^ rel were calculated for estimated popu-
lation (outputted from SIMPLAN) at a future date. Two radically different policies
for 2021 were developed: compaction and dispersal, in addition to a trend policy.
Given certain zone-wise policy inputs (in this case dwelling floorspace and employ-
ment), the residential location module of SIMPLAN calculates residential population
by zones. These future (modelled) values are used to calculate the spatial indicators for
2021 (see figure 8). Of course, it is obvious that, if the inputs of an alternative policy
0.9
(a) (b)
Population dispersion index

0.8

0.7

0.6

0.5

0.4

Future

200
Population H^ rel (%) indicator

150

100

50

0
1971 1981 1991 2001 2021
Trend Compaction Dispersal
Figure 8. Spatial indicators for alternative spatial policies.

(6) Details of SIMPLAN are available in Adhvaryu and Echenique (forthcoming) and SIMPLAN's

application to Ahmedabad is covered in Adhvaryu (2010).


862 B Adhvaryu

are geared towards dispersing the city, the spatial indicators will show dispersion and
vice versa. However, the advantage of using such an approach is that quantifying past
trends and projecting them for future alternative policies allows policy designers to see
deviations of alternative policies from an observed trend as well as amongst alternative
policies, ultimately providing a better methodology for refining and fine-tuning the
modelling inputs for testing alternative policies. For example, if trend analysis is not
carried out [that is, part (a) of figure 8, the shaded region] would not exist, in which
case part (b) of figure 8 by itself is not of much value in the context of the city being
modelled. On the other hand, part (a) of the figure by itself is a learning that is useful
regardless of it being used further in a model-based planning exercise for the future.
This approach of studying past trends and then projecting them on the basis of
modelled outputs complements spatial urban analysis and modelling well. Lastly, as
a decision-support tool, because of its graphical simplicity in showing the various
directions spatial trends could take, such an approach could be useful in refining and
fine-tuning planning inputs, especially when such decisions are influenced by top-level
(and busy) political and civil service personnel.

7 Conclusions
All three measures discussed in this paper show that Ahmedabad is gradually dispersing.
Reduction in the density gradients indicate that population is moving from the centre
to peripheral areas, which is also supported by the change in the H rel indicator.
This movement of people is likely to be further accentuated by migrant workers who
mainly choose to settle in peripheral areas (high-paid workers in western Ahmedabad
and low-paid workers in Eastern Ahmedabadöa trend allegedly observed by local
planning officials). In terms of the shape performance of the city, though in 2001
Ahmedabad was relatively compact, it is very likely that in the future this would
change, as shown by the dispersion index analysis. The 2011 Census figures are
expected to be released with the next couple of years and I am very keen on extending
this analysis by a decade and reporting it.
This paper shows that in a situation where only time-series population data are
available at a reasonable spatial disaggregation level, very simple but useful analysis of
evolution of the urban spatial structure of a city can be carried out. Typically, in
developing countries where detailed spatially disaggregated data are limited, such an
analysis can be very informative and useful on two accounts: first, it can be used by
city planning officials to improve the understanding of the evolution of the city's spatial
structure and, second, it could inform the process of formulating and testing implica-
tions of urban planning policies for the future, especially when `what if ' scenarios are
being built and tested to arrive at a final development/master plan for city-regions.
Acknowledgements. Gratitude is expressed to Cambridge Commonwealth Trust, Hinduja Cambridge
Trust, Churchill College, and Kettle's Yard Travel Fund for part funding support. Advice from
Professor Marcial Echenique is duly acknowledged. I thank the three anonymous referees for their
constructive criticisms, addressing which have enriched this paper. All figures and tables were
created by the author, unless stated otherwise. The views expressed in this paper are solely those
of the author.
References
Adhvaryu B, 2009 Enhancing Planning in Developing Countries: Urban Modelling for Ahmedabad,
India unpublished PhD dissertation, Department of Architecture, University of Cambridge
Adhvaryu B, 2010, ``Enhancing urban planning using simplified models: SIMPLAN for
Ahmedabad, India'' Progress in Planning 73 113 ^ 207
Adhvaryu B, Echenique M, forthcoming, ``SIMPLAN: A SIMplified PLANning model''
Environment and Planning B: Planning and Design (in press)
Analysing evolution of urban spatial structure 863

AMC, Ahmedabad Municipal Corporation


1979 Statistical Outline of Ahmedabad City 1977 ^ 78 Municipal Commissioner
1990 Statistical Outline of Ahmedabad City 1988 ^ 89 Municipal Commissioner
2007 Statistical Outline of Ahmedabad City 2006 ^ 07 http://www.egovamc.com/amc budget/
Outline%202006-07.pdf
AUDA, 1997 Revised Draft Development Plan of AUDA-2011 AD, Part-II: Policies, Proposals and
Programmes Ahmedabad Urban Development Authority, Ahmedabad
Bertaud A, 2001, ``Metropolis: a measure of the spatial organization of 7 large cities'',
http://alain-bertaud.com/images/AB Metropolis Spatial Organization.pdf
Bogart W, 2006 Don't Call it Sprawl: Metropolitan Structure in the Twenty-first Century (Cambridge
University Press, New York)
Clark C, 1951, ``Urban population densities'' Journal of the Royal Statistical Society 114 490 ^ 496
Gayda S, Boon F, Batty M, Besussi E, Chin N, Haag G, Binder J, Lautso K, Martino A, Noe«l C,
Dormois R, 2003 The SCATTER Projectö Sprawling Cities and Transport: from Evaluation to
Recommendations http://www.casa.ucl.ac.uk/scatter/download/Barcelona scatter Martino.pdf
Gayda S, Haag G, Besussi E, Lautso K, Noe«l C, Martino A, Mollanen P, Dormois R, 2005
SCATTER: Sprawling Cities and Transport: from Evaluation to Recommendations final report,
version 2.0, http://www.casa.ucl.ac.uk/scatter/download final.html
Mills E, Tan J, 1980, ``A comparison of urban population density functions in developed and
developing countries'' Urban Studies 17 313 ^ 321

ß 2011 Pion Ltd and its Licensors


Conditions of use. This article may be downloaded from the E&P website for personal research
by members of subscribing organisations. This PDF may not be placed on any website (or other
online distribution system) without permission of the publisher.

View publication stats

You might also like