National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER

Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)

Assessment of Urban Morphology of Delhi using Geospatial Techniques

Parul Singh*, Sudha Ravindranath†, Vidya A†, Ganesha Raj K†
*
Banasthali Vidyapeeth, Tonk, Rajasthan, 304022.
†
Regional Remote Sensing Centre - South, Bengaluru, Karnataka, 560037.
Corresponding Author emails:parulsinghkasyap@gmail.com*,
ravindranath_s@nrsc.gov.in† vidya_a@nrsc.gov.in† ganeshraj@nrsc.gov.in†
, ,
Corresponding author’s Ph. No.: +91-8765626478, +91-9844373606

Abstract

In the present study urban growth and pattern of Delhi has been studied using geospatial techniques.
Medium resolution satellite data covering two time periods viz., 1989 (Landsat 5), 2019 (Landsat 8) and
other ancillary data such as GIS layers, population reports has been used for the study. Supervised
classification using Maximum Likelihood Classifier (MLC) has been used for generating broad urban
land use/land cover and FRAGSTATS software for computing class and landscape metrics. It is seen that
Delhi city is undergoing urban sprawling with aggregation at the core area and conversion of rural areas
into urban patches at the periphery.

Key words: Land use/ Land cover (LU/LC), Spatial metrics, Morphology

1.Introduction
Urbanization is a rampant phenomenon at all
scales of development around the world. It is
considered as an index of transformation from
traditional rural economies to modern industrial
one. Kingsley Davis has explained urbanization
as process of switch from spread out pattern of
human settlements to one of concentration in
urban centers (Davis, 1962) [6]. Cities constantly
undergo structural change, improvement and
growth; such processes also involve the change
in urban relationship with the surrounding
environment. In recent years, cities have
witnessed rapid urbanization and urban
population growth resulting in haphazard
sprawling of cities.
According to United Nations report (The
World’s Cities 2018) Delhi will overtake
Tokyo as the world’s largest city by 2030 [9].
Due to rapid growth of population in Delhi, the
region is facing many problems associated with
housing, waste disposal, air pollution, traffic
congestion, shortage of electric power and
security. Urban planning is essential because
cities are becoming overpopulated and
resources are depleting at alarming stage. The
level of pollution i.e. air, water and land has
increased due to poor environmental
management. This has its direct impact on
quality of urban environment, affecting
efficiency of the people and their productivity
in the overall development [3]. Preservation and
protection of the environmentally sensitive
areas are mandatory. The study of land use and
land cover changes is useful for managing
natural resources and monitoring environmental
changes [1].
Use of Geospatial techniques involves
visualization, modeling, analysis, computation,
manipulation and interpretation of geospatial
data using remote sensing, GIS, spatial
statistics and GPS. These techniques play an
important role in monitoring, quantifying and
modeling of spatial temporal fluctuations in the
urban region over a period, identifying the
pattern of urban sprawl, and plan for urban
development [5].
Delhi being the capital city of India and rapidly
undergoing urbanization, there is an urgent
need to study the urban growth and its patterns
by analysis of spatial and temporal datasets
using geospatial techniques. This will help in
proper planning of the city and its
infrastructure.
2. Scope & objectives:
 National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER
Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)

The major scope of the study is to evaluate the
urban morphology dynamics from 1989 to 2019
which provides vital information on the urban
growth and its impact on the surrounding
environment. To realize the scope, the study is
divided into 3 main objectives which is listed
below:
city witnesses a large level of migration from all
parts of the country. According to the 2011
census Delhi city population was around 16
million [4] and the present estimated population
is nearly 29 million [10]. The increasing
population is one of the major driver for urban
expansion.

1. Generation of Urban landuse /

landcover map of Delhi using satellite
data for 2 time periods (1989 and 2019)
2. Generation of spatial metrics to
describe the urban growth pattern.
3. Change detection analysis of urban
landuse.

3. Study Area
The present study has been carried out within Figure1: Study area
NCT- Delhi, the capital city of India located
between the 28º24′17″ and 28º53′00″N latitudes 4.Data Used
and 76°45′30″ and 77º21′30″E longitudes
(Figure1). The study area covers a geographical In the present study, two Landsat series satellite
area of 1,490 km² and Yamuna river divides the imageries have been utilized for LULC change
city into two parts East and West. Delhi city is detection. Satellite imageries have been
surrounded by the Himalayas in the North and downloaded from the USGS website
Aravalli range in South-West. The Delhi Ridge (http://glovis.usgs.gov) for temporal study
is a remnant of the Aravalli ranges and functions (1989, 2019), (Figure:3). Both the datasets are
as the green lung space of the city and protects referenced in the same referencing system i.e.
the city from the hot winds coming from the UTM projection with zone number 43 and
deserts of Rajasthan to the west. The Delhi ridge WGS 84 datum. The details of the satellite data
is divided into four zones namely The Northern used is given in Table 1. Along with satellite
Ridge (The Old Delhi), Central Ridge (The New datasets, different GIS layers namely, drainage,
Delhi), South- Central Ridge (The Mehrauli) road and railways etc. and Census report of
and Southern Ridge (The Tughlaqabad). The India have been used.
ridge has many forest and sanctuaries, of which
Southern ridge forest, Sanjay van, Jahanpapa Table 1: Details of satellite data used
city forest, Rajokri Protected Forest, Mangar
Satellite Sensor Spatial Acquisition
Bani Forest are ecologically important areas Name resolution
which needs to be protected. There are many (meters)
Landsat 5 Thematic 5 Dec
historically significant monuments and heritage Mapper 30 1989
sites showing the rich history through its (TM)
architecture viz., Humayun’s Tomb, Lodhi Landsat 8 Operational 20 May
land Imager 30 2019
Gardens, Qutub Minar, Safdarjung’s Tomb, (OLI)
Jantar Mantar, Red Fort, Jama Masjid etc. which
are global tourist spots. There is an urgent need 5.Methodology
to protect and preserve these culturally rich
monuments from unplanned urban growth.
Apart from this, Delhi is also the hub of Landsat series satellite images of 1989 and 2019
education, employment, business and commerce have been used for the generation of Landuse /
generates enormous job opportunity for skilled landcover maps (Figure: 3). The satellite data has
and unskilled personnel. In the Proximity of been georeferenced to the same referencing
Delhi there are satellite towns like Gurugram, system i.e. UTM projection with zone number 43
Noida, Faridabad and Ghaziabad which are the and WGS 84 datum. The supervised classification
financial and industrial hubs due to which the using Maximum Likelihood classifier (MLC) has
 National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER
Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)

been adopted for generating broad urban land PD depends on

cover classes. The classification scheme adopted NP (Direct
relationship)
has the following landuse/ landcover classes viz.,
urban (built-up and transportation), water, NP decreases, Infilling & edge expansion happens
vegetation, scrub land, open/ fallow land and PD decreases
others. The overall methodology adopted for the
NP increases, PD scattered growth happens
study is given in Figure 2. The broad steps increases
involved are acquisition of temporal datasets If SHDI is 0 No Diversity
(1989 and 2019) from the USGS website, layer
stacking, study area extraction and georeferencing If SHDI > 0 Diversity of a particular patch type.
to the same referencing scheme. Supervised
classification has been adopted for delineating the If SHEI = 0 Patch type unevenly distributed
broad LU/LC classes. Post classification (Dispersed)
refinement has been carried out to improve the
If SHEI =1 Perfectly even distribution
accuracy of classification [9]. (Compact)
In order to understand the changing urban CONTAG
landscape pattern, spatial metrics have been used decreases Landscape pattern is aggregated
for quantifying and categorizing complex
landscape structure into simple and identifiable CONTAG Landscape pattern is clumped
patterns. The use of these metrics in landscape increases
studies helped in shifting of environmental ecology
from a qualitative to quantitative analysis. This has
been carried out using FRAGSTATS, a spatial
pattern analysis program which uses the landuse/ DATA ACQUISITION
(Landsat 5 – 1989, Landsat 8 – 2019)
landcover images as the input. Seven metrics have
been selected based on the literature review which
has the potential to best describe urban growth PRE-PROCESSING

pattern. They are Class Area (CA), Number of

Patches (NP), Patch Density (PD), Largest Patch
LAYER STACKING AND
Index (LPI), Contagion Index (CONT), Shannon’s SUBSETTING
Diversity Index (SHDI) and Shannon’s Evenness
Index (SHEI). Class Area (CA) is a measure of
landscape composition, specifically, how much of SUPERVISED
CLASSIFICATION
the landscape is comprised of a particular patch (MLC)

type. Number of Patches (NP) measure the patches

in the landscape. Patch Density (PD) measure the ACCURACY ESTIMATION AND
number of patches per unit area. Largest Patch POST CLASSIFICATION
REFINEMENT
Index (LPI) is the percentage of the landscape
comprised of the single largest patch (at the class
or landscape level). LPI is the measure of
dominance. CONTAG measure the tendency of LAND USE/ LAND COVER
MAP
FRAGSTATS
SOFTWARE
patch type to be spatially aggregated. SHDI and
SHEI measures the diversity of the landscape [8].
The interpretation of the selected spatial metrics is CHANGE GENERATED
given in Table 2 [2]. DETECTION AND
STATSISTICS
CLASS
AND LANDSCAPE
GENERATION METRICS

Table 2: Spatial metrics Interpretation table

SPATIAL PATTERN OF URBAN GROWTH ANALYSIS

ANALYSIS
METRICS

CA increases, Scattered growth of city

NP increases (Sprawling)
Figure 2: Overall methodology for the
CA increases, study
NP decreases Infilling & Edge expansion
(Urban Growth) 6. Result and Discussion
 National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER
Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)

Open/
i)Landuse/Landcover (LU/LC) map generation 690 450
fallow 16
46% 30%
Land
Supervised classification has been carried out on
both the satellite datasets (1989 and 2019) by 26 24
taking appropriate training signatures for broad Water 0.1
1.7% 1.6%
LU/LC classes viz., built-up, vegetation, open
land/fallow land, water bodies, scrub etc. The
136 100
results of the landuse / landcover is given in Scrub land 2.4
9.1% 6.7%
Table 3 and the classified outputs are given in
Figure 4. The classified map of 1989 shows
concentration of built-up in the centre and 358 311
Vegetation 3
eastern part of Delhi. But from the 2019 landuse 24% 21%
classified map it is seen that the direction of the
urban growth is toward the south east, south 280 604
west and north direction. The reason may be due Built-up 22
19% 41%
to the presence of technological and industrial
hubs like Gurugram, Faridabad, Noida and
Ghaziabad. From the Table 3, it is seen that the
urban area which was 280 km2 in 1989, has
increased to 605km2 in 2019. The open /fallow
land which was 690km2 in 1989 has decreased to
450 km2. There is 16 % decrease in the fallow/
open land which has been converted into new
residential layouts like Dwarka, Rohini,
Janakpuri. Marginal decrease in area is noticed
1989 2019
in the vegetation and scrubland categories. Thus,
an overall 22% increase is observed in the built-
up category over 30 years of span. The accuracy Figure 3: Satellite images of study area
assessment of both the classified images were
carried out by taking 50 points which were
cross-verified, and the classification accuracy
was 90% for 1989 and 92% for 2019 classified Urban Landuse/ Landcover Map

images. Post classification refinement was

carried to improve the accuracy of LU/LC
maps.
Change detection analysis has been carried out
using both the LU/LC maps and comparison is
given in Figure 5. The 1989 classified image is
considered as the base image for the change
detection analysis. In this map, the areas
showing
more than 10% increase is seen in green color,
the areas with decrease of 10% in red and the
areas which have no change is seen in grey.
From the Figure 5 it is seen that the central and Figure 4: Classified images of Delhi
eastern part of Delhi is already urbanized and
shows no change.
Change analysis 1989-2019

Table 3: Land Use/ Land cover Statistics

Area in Area in Chang

Class
km2 and % km2 and % e
Name
(1989) (2019) %
 National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER
Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)

201
1989 9
Open /
Fallow
10794 13713 4.06 5.15 16.8 5.39
Land

Water 890 1009 0.33 0.37 0.34 0.32

Scrub land 2027 2341 0.76 0.88Increase

Decrease
2.36 1.93
No change
Vegetatio
22670 16322 8.52 6.13 2.00 1.83
n
Built-up 9115 12573 3.42 4.72 2.52 9.96
Figure 5: Map showing LU/LC Change analysis

ii)Spatial metrics computation

Using the landuse/landcover classified maps of

1989 and 2019 in Fragstats software, the Table 5: Landscape metrics for LULC Classes
landscape and class spatial metrics were
computed for 7 parameters as mentioned in the LANDSCAPE
1989 2019
methodology. The class spatial metrics for Built- METRICS
up area indicated that, the total Class Area (CA) 48.9185 49.4313
CONTAG
has increased by 22% while the Number of
Urban Patches (NP) has increased by 38% SHDI 1.4272 1.4349
indicating the scattered growth of the region.
The same observation is confirmed from the SHEI 0.7965 0.8008
increase in patch density value from 3.4 to 4.7
along with increase in number of patches, which
indicates that the urban expansion is undergoing 7. Conclusion
in a scattered manner. Large Patch Index (LPI) From the study, it is seen that the urban growth
has increased by 295% which indicates has increased by nearly 22% over 30-year span
aggregation at the urban core area. with most of the increase is seen due to the
The class metrics for landuse/ landcover class conversion of open land / fallow land into built-
is given in Table: 4. From the landscape up. The spatial metric analysis has indicated
analysis with computation of spatial metric that the city is sprawling with aggregation at
indicators, Contagion value (CONTAG) is the core. It is evident that the combined
increased resulting in the tendency of patch approach of using geospatial techniques and
types to be spatially clumped and the increase spatial metrics has helped in capturing the
in Shannon’s Diversity Index (SHDI), and physical dimension of urban growth and pattern
Shannon’s Evenness Index (SHEI) indicate the leading to better understanding and
richness in the number of different patches type representation of the spatiotemporal dynamics
and dispersed pattern of the classes of Delhi city.
respectively.
The landscape metrics for the LU/LC 8. References
From the analysis of landuse and landcover map [1] Anas, A, Arnott, R., & Amall, K. (1998).
and spatial metrics it is seen that Delhi city is Urban spatial structure. Journal of Economic
undergoing urban sprawling with aggregation at Literature, American Economic Association,
the core area and conversion of rural areas into 36(3), 1426-1464.
urban patches at the periphery.
[2] Abebe, G. A. (2013): Quantifying Urban
growth Pattern in Developing countries using
Table 4: Class metrics for landuse / landcover
Remote Sensing and Spatial Metrics: a case
classes
study in Kampala, Uganda. Master of Science
CLASS PD thesis. University of Twente.
NP (Number LPI
NAME per 100 (percent) [3] Bhatta, B. (2010). Analysis of urban
hectares) Growth and sprawl from remote sensing Data.
1989 2019 1989 2019 Springer Heidelberg Dordrecht, London, New
 National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER
Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)

York. DOI10.1007/978-3- 642- 05299-6.

[4] Census of India (2011). Population Census,
New Delhi.

[5] Coppin, P., Jonckheere, I., Nackaerts, K.,

Muys, B. and Lambin, E., (2004), Digital
change detection methods in ecosystem
monitoring: a review. International journal of
remote sensing, 25(9), pp 1565-1596.

[6] Davis Kingsley (1962): "Urbanization in

India – Past and Future", in Turner, R. (ed.)
India's Urban Future, University of California
Press, Berkley.

[7] Herold, M., Couclelis, H. Clarke, K. C.

(2005). “The role of spatial metrics in the
analysis and modeling of urban land use
change.” Computers, Environ. & Urban
Systems, 29(2005), 369-399.

[8] McGarigal, K., Cushman, S. A., Neel, M.C.,

& Ene, E. (2002). FRAGSTATS: Spatial
pattern
analysis program for categorical maps.
Amherst: Computer software produced by the
authors at the University of Massachusetts.

[9] Taubenböck, H., Pengler, I., Schwaiger, B.,

Cypra, S., Hiete, M., Roth, A (2007): A multi-
scale urban analysis of the Hyderabad
Metropolitan area using remote sensing and
GIS. In: Urban Remote Sensing Joint Event,
Paris, France. p.6.

[10] United Nations (2018). World urbanization

prospects, the 2018 revision, New York
National Symposium on Innovations in Geospatial Technology for Sustainable Development with special emphasis on NER
Shillong, November20-22, 2019. Paper No. – (To be filled by organizers)