Urban sprawl assessment Entropy approach

Sulochana Shekhar
Lecturer in geography
National Defence Academy, india
sulogis@yahoo.com

The Remote Sensing data, GIS combined with Shannon's entropy has proved to be efficient in
monitoring and measuring the urban sprawl of Pune, a developing city in the Indian state of
Maharashtra

The monitoring of urban development is mainly to find out the type, the amount and location of land
conversion for future planning. Although various studies have been dedicated to measure and monitor
the growth and sprawl of urban form they have limitations in capturing the characteristics of urban
sprawl. Here is an alternative technique - the Shannon's entropy - specially to measure the sprawl is
developed within the integration of remote sensing and GIS. The advantages of the entropy method
are its simplicity and easy integration with GIS.

Hence an attempt has been made to make use of entropy method to analyse urban sprawl of Pune
city, in the state of Maharashtra, which is a part of fast growing urban agglomeration in India and to
prove that the technique is useful in mapping and monitoring the urban sprawl patterns.

Study Area
The study area (Figure 1), the Pune city lies between latitudes 1825'N and 1837'N and longitudes
between7344'E and 73 57'E and cover an area of 243.96 sq km. Pune is located in a saucer shaped
basin at an average altitude of 560m from mean sea level and surrounded almost on all sides by hills
at different altitudes. The ground slope rises gradually from the river, with RL 530 metre (1,750ft)
towards the hills with the highest RL 697 metre (2,300 ft) at the Vetal Hill. The slopes become steeper
close to the hills.

The city is very fortunate in having in addition to the beautiful hills a number of rivers flowing through it.
The Mutha River enters the city from southwest, the Mula River from northwest and the Pawna River
from the north. After their confluence to the west of Dapodi, the Mula and the Pawna join the Mutha
River near the Sangam Bridge and these rivers then take an almost eastward course and leave the
study area on the east of the Mundhawa village. The climate of Pune is typical monsoon. The
temperature ranges from 15C to 35C. The average rainfall is 70cm with more rain during southwest
monsoon period. Pune is one of the fast developing urban agglomerations in Asia and ranks eigth at
national level (census 2001). It has grown quite haphazardly. The present growth is due to various
factors such as industrialisation, location of various Central and state Government establishments.

Database
Digital data of IRS 1D LISS III of 5th December 2001, pertaining to path-095 and row 059, Survey of
India toposheets (Index No. 47F/14, 47F/15) on 1:50,000 scale were used for the purpose. Maps and
other collateral data were procured from Pune Municipal Corporation (PMC), Town Planning Office,
and Mashal NGO group (who were involved in the preparation of city development plan). The
demographic details were taken from census abstracts.

Methodology
The digital remote sensing data was processed and geo-referenced in Erdas 8.3.1 software. Initially
the toposheets were scanned and geo-referenced and used as base for image registration. The geo-
referenced FCC image (Figure 2) was further enhanced by using necessary enhancement techniques.
After that visual interpretation of image was done to identify the major land use classes. The enhanced
image was classified on the basis of samples collected from different classes. Maximum likelihood
classifier of supervised classification (Figure 3) in Erdas was used to classify the image into four major
classes and again they remerged into two main classes as built-up and Non- built-up area.
Fig 1. Study Area

Fig 3. Classified Area

Fig 2. FCC Image of the study area

The percentage of an area covered by impervious surfaces and concrete is a straightforward measure
of development (Barnes et al, 2001). Modern urbanisation results in profound changes to the
landscape, specifically the proliferation of asphalt and concrete along with the displacement of
agriculture and forestland.

Hence, considering the built-up area as a potential and fairly accurate parameter of urban sprawl gives
better knowledge for understanding the behavior of such sprawls. Therefore the classified image and
the merged toposheets were brought into Arc View3.2a environment and the built-up area was
digitised from the image and the toposheets and the area under built-up was calculated for further
analysis.

Measuring Urban Sprawl

Urban sprawl over the period of 1980 to 2001, almost a period of two decades was determined by
computing the area of all the settlements from the digitised toposheets of 1979-80 and comparing it
with the area obtained from the classified IRS 1D mage. Shannon's entropy was used to capture the
amount of sprawl in these time periods. Shannon's entropy (E) can be used to measure the degree of
spatial concentration and dispersion exhibited by geographical variable. (Theil, 1967: Thomas, 1981)

This measure is based on the notion that landscape entropy or disorganisation increases with sprawl.
Urban land uses are viewed as interrupting and fragmenting previously homogenous rural landscape,
thereby increasing landscape disorganisation. Entropy value varies from 0 to 1. If the distribution is
maximally concentrated in one region, the lowest value 0 will be obtained. Conversely an evenly
disperse distribution across space will give a maximum value of 1.

The dispersal of built-up areas from a city centre will lead to an increase in the entropy value. This
gives a clear idea to recognise whether land development is towards a more dispersed or compact
pattern.
 Table 1 Pune - Population (Ward Wise)

% of Total Area in Built-up in Pop A Pop B

Name Pop2001
Built-up km km density density
Aundh 25.28 179886 44.77 11.32 4018.003 15890.99
Ghole road 42.99 204316 12.05 5.18 16955.68 39443.24
Karve road 50.1 201527 10.38 5.2 19414.93 38755.19
Warje 63.72 116386 13.81 8.8 8427.661 13225.68
Vishrambagh 85.77 154425 4.71 4.04 32786.62 38224.01
Kasba peth 95.85 100059 3.37 3.23 29691.1 30978.02
Bhavani peth 99.53 205009 2.15 2.14 95353.02 95798.6
Sangam wadi 17.31 213718 23.98 4.15 8912.344 51498.31
Yerwada 26.02 251100 35.32 9.19 7109.287 27323.18
Dhole patil
36.57 218306 9.98 3.66 21874.35 59646.45
road
Hadapsar 30.19 239370 29.94 9.04 7994.99 26478.98
Bibvewadi 42.41 239530 21.95 9.31 10912.53 25728.25
Entropy (E) can be calculated by using this formula.

DENi is the density of land development. That equals to the amount of land development (Built-up
area) divided by the total amount of land in the ith zone in a total of n zones.
Since entropy can be used to measure the distribution of a geographical phenomenon, the difference
on entropy between two different periods of time can also be used to indicate the change in the
degree of dispersal of land development or urban sprawl. (Anthony gar-on YEH.et al).

To analyse the sprawl of Pune city, the study area was divided into 14 zones based on the
administrative wards of Pune City. Further to quantify the sprawl, the villages which were recently
merged with PMC were also taken into the analysis. The ward boundaries were digitised from the
scanned map of PMC and similarly the newly added 23 village boundaries were also digitised. Two
types of thematic layers are needed for calculation of densities of land development in each zone. The
ward and village boundaries formed one layer and the built-up area formed the other layer. The
overlay of boundaries layer on the built-up layer can clip the built-up area of individual wards and
villages. The density of built-up area was calculated by dividing the built-up area of each village from
its total area.

In general, the sprawl is affected by some location factors, such as distance to urban centers and
roads. Entropy was calculated based on these factors by creating buffer zones from the core as well
as from the highway. Three zones were created from the city centre, namely: core, middle and the
fringe zones, based on their location, which followed their administrative boundaries. Then the road
buffer zones of 500m, 1000m, 1500m and 2000m from national highway were created in the Arc View
GIS and these buffer zones were overlaid on image built-up area theme. By clipping out the built-up
area of each buffer zone, the density of land development in each zone was computed. Finally entropy
values were calculated for all these zones.

Results And Discussion

Remote sensing and GIS are time and again proving that they are highly capable in urban studies
particularly in urban management and planning. With these modern tools it is quite possible to analyse
the spatial pattern of urban sprawl over different time periods and it can be systematically mapped,
monitored and accurately assessed from remotely sensed data along with conventional ground data.
(Lata et al, 2001). Pune, is experiencing high rate of population growth (Figure 4). Between 1976 and
1981, the population of Pune city grew by 16.7%, and from 1981 to 1991 it grew by 30.2%. Between
1991 and 2001, the growth was more than two times i.e., 62.17%, where as Pune district has the
growth rate of 30.58% and the state experienced the growth rate of 22.5%. This shows that the
increasing population of Pune city exercise lot of pressure on available land in the city and the city's
limit had been pushed towards the adjacent villages. Further to understand this situation the
population density for each ward was calculated in two aspects, such as Population density A and
Population density B. Population density A (Table1) is nothing but the population of the ward divided
by the total area of the ward and the Population density B was calculated by dividing the population of
the ward to the built-up area of the ward. This will show (Figure 5) how much pressure has been given
to the available land. Recently, 23 villages had been added to the old municipal limit and the total area
of Pune city has increased from 145.92 sq km to 243.96 sq km. The Pune Municipal Corporation has
also prepared the development plan for these villages. Our present study does not have any aim to
analyse the development plan of fringe villages. But to help the planners to identify the villages which
have to be given more priority and where exactly the sprawl takes place and to understand the degree
and intensity of such sprawl for better management. As the urban phenomenon (residential, industrial,
commercial, public and semi public uses etc) has uniform reflectance throughout electromagnetic
spectrum (Pathen et al, 1991), it is not possible to identify and delineate urban land use classes using
digital analysis techniques as these techniques employ spectral characteristics of the objects for the
classification.

Hence the visual analysis plays an important role for the classification of urban land uses. In this study
also, the basic elements such as tone, size, shape, texture, pattern, location and association were
used to differentiate the built-up area from non-built-up. From the image, the built-up area was
digitised and built-up area for each administrative ward and village were taken individually and the
area under built-up was calculated. In a similar way, the built-up area from the toposheet also been
taken up and ward wise area under built -up was calculated. By using the formula referred above, the
entropy values for Pune city for both the periods were calculated.

The entropy value for the year 1979-80 (from the topo sheet) was 0.7976 and for the year 2001, (from
the IRS 1D LISSIII image) it was 0.9553 for the administrative wards and 0.9430 for the merged
villages. Even then the entropy values in both the cases were above 0.5 and indicating higher rate of
sprawl, the entropy value for the year 2001, shows a highly dispersed development from the centre or
core of the city comparatively during 1979-1980, where it was less dispersed. In order to identify the
villages and the wards which are showing higher rate of sprawl, the entropy values are reclassified into
three classes by using Arc View Spatial analyst. Each class represented a zone namely: zone of
higher level, and moderate and lower level. Out of 23 villages, Balewadi, Baner, Bavdhankhurd, Kalas,
Dhanori, Mohammadwadi, Undri and Katrej are showing higher degree of sprawl. At Administrative
ward level, Aundh, Sangamwadi, and Yerwada are experiencing high sprawl.

The measurement of the difference on entropy between time (t) and t +1 was also obtained by using
the above said formula. It showed the increase of entropy as high as 20% between 1980 and 2001.
When we compare the existed land use of Pune during 1979 to the built-up area of 2001, it shows a
substantial increase of nearly 40% that is 27.9 sq km. The land under agricultural use was 24.63 sq
km during 1979 and it became reduced into 15.33 sq km during 2001 with a loss of 37.75% of
cultivable land. Thus, this analysis revealed that the city is under highest rate of sprawl, and the sprawl
takes place at the cost of adjacent agricultural, forestlands.

The dispersed development along the highways or surrounding city and in rural countryside is often
referred as sprawl (Theobald, 2001). In general, the urbanisation takes place either in radial direction
from the core or linearly along the highways. Therefore in this study, these two aspects were taken
care of. In this context, firstly, study area was divided into three zones from the core and the entropy
values were calculated for both the periods. The entropy values resulted from the analysis showed a
similar trend of previous analysis. For the year 1979-80 (from toposheet) the entropy value was 0.71
and it was 0.98 for the year 2001 (from the image).

The difference on entropy between 1979-80 and 2001 was 38.5%, which is quite high. Secondly, the
buffer function of GIS was used to calculate entropy from the highway. Four buffer zones of 500m
each were created along the National highway and density of land development in each buffer zone
was extracted from the image built-up area. By applying the formula, the entropy value of 0.9714, for
the road buffer zones was obtained and it revealed the same fact that the city has the highest degree
of sprawl along national highway. It also showed the density of land development was declining while
the distance from the highway increases. There was a high negative correlation between these two
variables. The low density in the fourth buffer zone means the dispersed settlement and occurrence of
sprawl.

Conclusion
This study shows that the city of Pune is experiencing a leapfrog pattern of urban sprawl due to the
hills, and ribbon sprawl along the highways. The higher entropy values indicate that the rate of sprawl
is quite high and needs proper management to attain sustainable development. It also proves that
entropy is a good indicator to identify, measure and monitor the spatial distribution of urban
phenomenon. With the integration of remote sensing and GIS, it is easy to implement entropy
approach in the study of Pune city and quite sure that the results will be definitely useful for
implementing and managing the development plan of Pune.

ACKNOWLEDGEMENTS
The Author is grateful to Shri P L N Raju IIRS, Dehradun, Dr Raghavasamy, Dr Badrinath, K M Latha
and other faculty of Urban & Land Use Department, NRSA for their guidance and encouragement. The
author is also thankful to Prof.Jeymala Diddee (Rtd), Prof K R Dixit (Rtd), Department of Geography,
University of Pune and Prof V G Dixit, NDA for their encouragement and support.