ISSN (Print) : 0974-6846

Indian Journal of Science and Technology, Vol 9(18), DOI: 10.17485/ijst/2016/v9i18/92089, May 2016 ISSN (Online) : 0974-5645

Assessment of the Temporal Variations of Surface

Water Bodies in and around Chennai
using Landsat Imagery
Nikita Roy Mukherjee1 and Christopher Samuel2*
1
Remote Sensing and Geographic Information System, SRM University, Kattankulathur – 603 203, Kanchipuram
District, Tamil Nadu, India
2
Department of Civil Engineering, SRM University, Kattankulathur – 603 203, Kanchipuram District, Tamil Nadu,
India; [email protected]

Abstract
Urban India is witnessing a rapid growth that has an impact on the fate of water bodies. Chennai, one of the fastest growing
metropolitan cities in India, is not an exception to this urbanization process. Chennai’s precious water resources are under
severe stress. Further, the recent, 100-year flood disaster that occurred in November–December 2015 over Chennai has
had tragic consequences on its population and has brought back the focus on its water bodies. Remote sensing combined
with GIS techniques enable accurate spatiotemporal change detection of natural resources. Landsat and multispectral
imageries provide reliable and accurate data for the detection of changes in the areal extents of surface water bodies.
Spatiotemporal change analyses on the surface water bodies in Chennai and its surrounding areas were conducted through
water extraction techniques by using indexes, such as WRI, NDWI and MNDWI using GIS software. The changes in the
surface water bodies were estimated by analyzing and using Landsat images for the pre- and post-monsoon periods of
the years1977–78, 1988–89, 1997–98, 2006–07 and 2015–16. The changes and their variations in the indexes and the
interpretation of these changes are discussed. NDWI is most suitable for extracting surface water bodies. WRI and MNDWI
can be used for the extraction of surface water bodies only from those satellite imageries that have the MIR band. However,
MNDWI is suitable for inundated areas within a largely urban-covered region. Hence, urban planners can utilize these
results for better urban planning and flood-disaster management.

Keywords: Change Detection, Chennai, Disaster Management, Landsat Imagery, MNDWI, NDWI, Remote Sensing and
GIS, Surface Water Bodies, Urban Planning, Water Extraction Techniques, WRI

1. Introduction and hydrology are important for human beings and are
critical for their survival1.
Water resources are under stress globally and this stress Remote sensing combined with GIS techniques enable
has reached a stage of crisis in many developing countries. accurate spatiotemporal change detection of natural
he surface water bodies are under severe stress caused by resources. Landsat and multispectral imageries provide
anthropogenic activities such as intensive urban growth reliable and accurate data for the detection of changes in
and rapid industrialization. Physical changes in the urban the areal extents of surface water bodies.
environment degrade the quality of the water bodies. Spatiotemporal change analyses on the surface
he Chennai Metropolitan Area (CMA) is home to a water bodies in Chennai and its surrounding areas were
network of rivers, ponds and lakes. hese surface water conducted through water extraction techniques by using
bodies have since time immemorial served as sources indexes, such as WRI, NDWI and MNDWI by using GIS
for the irrigation and domestic needs of the population sotware. he changes in the surface water bodies were
around them. Water resources management, land use estimated by analyzing and using Landsat images for the

* Author for correspondence

 Assessment of the Temporal Variations of Surface Water Bodies in and around Chennai using Landsat Imagery

pre- and post-monsoon periods of the years 1977–78, with a total water-spread area of 63.8 km2 approximately
1988–89, 1997–98, 2006–07 and 2015–16. he changes 5.4% of the area of the CMA3. Many of these water bodies
and their variations in the indexes and the interpretation are a part of the system of tanks (Eri system) within the
of these changes are discussed. Urban planners can utilize CMA. hese are water harvesting systems that consist
these results for better planning and disaster management. of a series of connected tanks that are situated within
one of the basins of the rivers lowing through the area.
hey meet the needs of the people for both drinking and
2. Materials and Methods irrigation. hey form the major aquifer recharge areas.
2.1 Study Area
Chennai city is one of the largest urban agglomerations
in India. he areal extent of Chennai Metropolitan Area
(CMA) is 1189 km2 (Figure 1)2. he CMA area has a
longitudinal extension of 80° 1΄ 16" E to 80° 19΄ 18˝ E
and a latitudinal extension of 12° 51΄41˝ N to 13° 15΄ 49˝
N approximately (Figure 1). It has an average elevation
of around 6.7 m from the mean sea level and its highest
elevation is 60 m. Chennai contains a diverse population
of ethnoreligious communities. he CMA covers three
districts of the state of Tamil Nadu: Chennai District, part
of Tiruvallur District and part of Kanchipuram District.
he district-wise details of the areal extents of the CMA
is given in Table 12.
Chennai city is drained by two meandering rivers, the
Cooum River in the central region and the Adyar River in Figure 1. Map of the Chennai Metropolitan Area (CMA).
the southern region. Both the rivers are heavily polluted
with eluents from commercial and domestic sources. 2.2 Materials
he Buckingham canal travels parallel to the coast linking he analyses in this study were conducted by using
the two rivers, Adyar and Cooum. he OtteriNullah an Landsat satellite images. Five sets of satellite images were
east–west stream runs through Buckingham canal. he acquired for the years 1977 to 20164. he imageries for
Red Hills, Cholavaram and Chembarambakkam lakes the pre-monsoon seasons are for the years 1977, 1988,
supply Chennai with potable water3. 1997, 2006 and 2015. he corresponding post-monsoon
here are 124 tanks spread throughout the CMA area imageries are for the years 1978, 1989, 1998, 2007 and
2016. he details of the imageries are provided in Table 2.

Table 1. Areal extent of CMA

DISTRICT TALUK AREA in km2
Chennai Fort-Tondiarpet, Perambur-Purasawalkam, Egmore-Nungambakkam, Mambalam-Guin- 176 (total area)
dy, Mylapore-Triplicane
Kanchipuram Tambaram, Sriperumbudur, Chengalpattu 376 (part of the district)
Tiruvallur Ambattur, Tiruvallur, Ponneri, Poonamallee 637 (part of the district)

Table 2. Details of satellite imageries

Satellite Sensor Bands Spectral Scene Size Pixel Year
(number) Range (µm) (km) Resolution (m)
Landsat 1–45 Multi-Spectral Sensor (MSS) 1, 2, 3, 4 0.5–1.1 170 × 185 79 1977, 1978 and 1989
Landsat 4–55 hematic Mapper - TM 1, 2, 3, 4, 5, 0.45–2.35 170 × 183 30 1988, 1997 1998,
and 7 2006, and 2007
Landsat 86 Operational Land Imager (OLI) 1–11 0.43–12.51 170 × 183 30 2015 and 2016

2 Vol 9 (18) | May 2016 | www.indjst.org Indian Journal of Science and Technology
 Nikita Roy Mukherjee and Christopher Samuel

2.3 Methodology Figure 2. he computed corresponding WRI values for

he Landsat satellite images s were processed using GIS these years are presented in Table 3. he WRI indexes that
sotware, ERDAS Imagine and ArcGIS. he satellite were computed for the post-monsoon seasons of the years
images were converted to the Universal Transverse 1998, 2007 and 2016 and the identiied water bodies are
Mercator (UTM) projection prior to processing. he shown in Figure 3. he computed WRI values for these
surface water bodies were analyzed by using the Water years are presented in Table 4.
Ratio Index (WRI), Normalized Diference Water Index
(NDWI) and Modiied Normalized Diference Water
Index (MNDWI). hese indices provide a quantiication
of the temporal changes in the surface water bodies.

3. Results
3.1 Water Extraction Techniques
Figure 2. WRI images showing extracted surface water
bodies for the pre-monsoon years (a) 1988, (b) 1997, (c)
3.1.1 Water Ratio Index (WRI) 2006 and (d) 2015.
he Water Ratio Index as highlighted by Shenand Li7:
WRI = (GREEN + RED) / (NIR+MIR)
WRI is the ratio between the total of the spectral
relectances in the RED and GREEN bands to that of the
total of the Near Infra-Red (NIR) and Middle Infra-Red
(MIR) bands. In this study, this technique was applied
over the L 4–5TM and L 8 images as only these have the
required band characteristics. herefore, WRI was not
calculated for the years 1977, 1978 and 1989 because
of the limitations in the number of bands provided by
Landsat MSS. he WRI indexes that were computed for Figure 3. WRI images showing extracted surface water
the pre-monsoon seasons of the years 1988, 1997, 2006 bodies for the post-monsoon years (a) 1998, (b) 2007 and
and 2015 and the identiied water bodies are shown in (c) 2016.

Table 3. Areal extent of the surface water bodies for pre-monsoon years (in km2)
SURFACE WATER DETECTION YEAR YEAR YEAR YEAR YEAR
ALGORITHM 1977 1988 1997 2006 2015
WRI - 56.68 35.84 59.83 58.79
NDWI 42.11 59.55 40.06 61.64 41.20
MNDWI - 94.24 40.74 65.29 40.16

Table 4. Areal extent of the surface water bodies for post-monsoon years (in km2)
SURFACE WATER DETECTION YEAR YEAR YEAR YEAR YEAR
ALGORITHM 1978 1989 1998 2007 2016
WRI - - 55.55 49.13 177.94
NDWI 68.59 8.35 92.87 83.23 91.62
MNDWI - - 121.09 96.09 686.02

Vol 9 (18) | May 2016 | www.indjst.org Indian Journal of Science and Technology 3
 Assessment of the Temporal Variations of Surface Water Bodies in and around Chennai using Landsat Imagery

3.1.2 Normalised Diference Water Index (NDWI) monsoon seasons of the years 1988, 1997, 2006 and 2015
and the identiied water bodies are shown in Figure 6.
he NDWI takes advantage of the fact that water has a
he computed corresponding MNDWI values for these
high relectance in the green band while it has strong
years are presented in Table 3. he MNDWI indexes that
absorptions in the Near Infra-Red bands. herefore, it is
were computed for the post-monsoon seasons of the years
based on the ratio as highlighted by McFeeters8:
1998, 2007 and 2016 and the identiied water bodies are
NDWI = (GREEN-NIR) / (GREEN+NIR)
shown in Figure 7. he computed MNDWI values for
he NDWI algorithm was applied on all the ive sets
these years are presented in Table 4.
of images, that is, 1977–78, 1988–89, 1997–98, 2006–2007
MNDWI = (GREEN-MIR) / (GREEN+MIR)
and 2015–16. he resultant water bodies mapped by this
technique are shown in Figures 4 and 5. he computed
NDWI values for the entire CMA region using all the
imageries are provided in Tables 3 and 4.

Figure 6. MNDWI images showing extracted surface

water bodies for the pre-monsoon years (a) 1988, (b) 1997,
Figure 4. NDWI images showing extracted surface water (c) 2006 and (d) 2015.
bodies for the pre-monsoon years (a) 1977, (b) 1988, (c)
1997, (d) 2006 and (e) 2015.

Figure 5. NDWI images showing extracted surface water

bodies for the post-monsoon years (a) 1978, (b) 1989, (c)
1998, (d) 2007 and (e) 2016. Figure 7. MNDWI images showing extracted surface
water bodies for the post-monsoon years (a) 1998, (b)
3.1.3 Modiied Normalized Diference Water Index 2007 and (c) 2016.
(MNDWI)
3.1.4 Rainfall
he MNDWI as highlighted by Xu9 replaces the Near
Infra-Red band with the Middle Infra-Red band and he manifestation of water bodies through the extraction
thus, it eliminates the noises caused by the built-up techniques as described in the previous sections are
area and vegetation. he noises result in negative values directly related to the rainfall that occurs over the CMA
and hence, they are easily eliminated by the use of the area during the speciic years of study. Table 5 provides
algorithm. he positive values relect the extracted water data on the rainfall that occurred in the three districts
bodies. herefore, similar to the WRI technique, the use of Chennai, Kanchipuram and Tiruvallur districts that
of MNDWI is restricted to only those imageries that have constitute the CMA10,11. Based on the ratio of the areal
the Middle Infra-Red band. Hence, MNDWI is applicable extents of each of these districts within the CMA, the
only for those images pertaining to the years 1977, 1987 total rainfall for the speciic years of the entire CMA area
and 1989. was computed and this is provided in the last column of
he MNDWI indexes that were computed for the pre- Table 5. he rainfall data for the year 2005 is included
as the precipitation was high that year and this fact can

4 Vol 9 (18) | May 2016 | www.indjst.org Indian Journal of Science and Technology
 Nikita Roy Mukherjee and Christopher Samuel

inluence the real areal extents of the water bodies that 4.2 Normalized Diference Water Index
were studied for the year 2006 and 2007. (NDWI)
Table 5. Rainfall data for the study area10,11 he NDWI for the entire study years of 1977–1978, 1988–
YEAR Chennai Kanchipuram Tiruvallur CMA
1989, 1997–1998, 2006–2007 and 2015–2016 indicate the
mm mm mm mm following (Figure 9):
1977 215 511 712 1438 • here is a 56.47% increase in the area of the surface
1988 205 453 674 1333 water bodies in the post-monsoon season of 1978
1997 222 474 717 1412 when compared to the pre-monsoon season of 1977.
2005 367 785 1074 2226 his corresponds to a rainfall of 1438 mm.
2006 208 444 623 1275 • here is a decrease of 85.64% in the areal extent of the
2015 4382 5740 7822 17944 surface water bodies in CMA in the post-monsoon
season of 1989. his corresponds to a rainfall of 1333
4. Discussion mm in the pre-monsoon season of 1988.
• here is an unusual increase of 52.81% in the surface
4.1 Water Ratio Index (WRI) water bodies of the post-monsoon season of 1998
he WRI for the pairs of years 1997–1998, 2006–2007 and owing to the occurrence of a high amount of rainfall
2015–2016 indicate the following (Figure 8): in the pre-monsoon season of 1997. his corresponds
• here is a 55% increase in the area of the surface to a rainfall of 1412 mm.
water bodies in the post-monsoon season of 1998 • here is a 21.59% increase in the areal extent of the
when compared to the pre-monsoon season of 1997. surface water bodies in the post-monsoon season of
his corresponds to a rainfall of 1412 mm. the year 2007 when compared to the pre-monsoon
• here is a decrease of 17.9% in the area of water bodies season of 2006, which corresponds to a rainfall of
in the post-monsoon season of 2007 as compared to 1275 mm and the previous year monsoon season of
the pre-monsoon season of 2006. his corresponds to 2005, when the precipitation was 2226 mm.
a rainfall of 1275 mm. However, there was an unusual • here is an unusual increase of 122.37% in the total
increase in the area of the water bodies in the CMA areal extent of the surface water bodies of the post-
in the 2006 pre-monsoon period owing to an excess monsoon season of 2016 owing to the unprecedented
of rainfall in the 2005 monsoon season when the rainfall of 17944 mm in the monsoon season of 2015.
precipitation was 2226 mm.
• here is an increase of 202.7% in the total areal extent
of all surface water bodies situated within the CMA
during 2016 owing to the unprecedented rainfall of
17944 mm in the monsoon season of 2015.
• However, the WRI values could not be calculated for
the years 1977, 1978 and 1989 as these images did not
have the required MIR band.

Figure 9. NDWI indexes for pre- (in green) and post-

monsoon (in red) seasons.

4.3 Modiied Normalized Diference Water

Index (MNDWI)
he MNDWI for the pairs of years 1997–1998, 2006–2007
and 2015–2016 indicate the following (Figure 10):
• here is an increase of 197.22% in the areal extent of
Figure 8. WRI indexes for pre- (in green) and post- the surface water bodies in the post-monsoon season
monsoon (in red) seasons.

Vol 9 (18) | May 2016 | www.indjst.org Indian Journal of Science and Technology 5
 Assessment of the Temporal Variations of Surface Water Bodies in and around Chennai using Landsat Imagery

of 1998 when compared to the pre-monsoon season 5. Conclusions

of 1997. his corresponds to a rainfall of 1412 mm.
• here is an increase of 30.8% in the total area of the Based on the results and discussions provided in Sections
surface water bodies situated in CMA in the post- 3 and 4, the following conclusions can be arrived at:
monsoon season of 2007 when compared to the • he indices for the delineation of the areal extent
pre-monsoon season of 2006. his corresponds to a of water bodies such as, WRI, NDWI and MNDWI
rainfall of 1275 mm. However, there was an unusual when applied on Landsat images through ArcGIS
increase in the area of the water bodies in the CMA and ERDAS Imagine sotware provide reliable
in the 2006 pre-monsoon period owing to an excess spatiotemporal information.
of rainfall in the 2005 monsoon season when the • he NDWI can be utilized for interpretation of the
precipitation was 2226 mm. areal extent of water bodies in a wide variety of
• here is an unusual increase of 1608.21% in the areal imageries.
extent of the surface water bodies situated in the • WRI and MNDWI are restricted for use only in
CMA in the post-monsoon season of 2016 owing images that have the MIR band.
to the unprecedented rainfall of 17944 mm in the • he CMA shows good correlation of the areal extent
monsoon season of the year 2015. of surface water bodies with that of the rainfall data.
• However, the MNDWI values could not be calculated • MNDWI provides for a sophisticated and reined
for the years 1977, 1978 and 1989 as these images did estimation of the extent of surface water bodies
not have the required MIR band. within the CMA when compared to WRI and NDWI
(Figures 6, 7, and 11). his is particularly evident for
the lood afected years 2015–2016. he extracted
images can be utilized by planning authorities in order
to understand the inundated areas within the CMA,
analyse the causes for the same, and take suitable
decisions for the prevention of such occurrences in
the future.

6. Acknowledgement
he authors are grateful for the imageries provided by the
USGS Earth Explorer. We would also like to thank the
Figure 10. MNDWI for pre- (in green) and post- Department of Civil Engineering, SRM University for all
monsoon (in red) seasons. the help extended.

7. References
1. Sudha MC, Ravichandran S, Sakthivadivel R. Water bod-
ies protection index for assessing the sustainability status
of lakes under the inluence of urbanization: A case study
of south Chennai, India. Environment, Development and
Sustainability. 2013 Oct; 15(5):1157–71.
2. Chennai Metropolitan Development Authority (CMDA).
Available from: http://www.cmdachennai.gov.in/
3. A. Anbarasan. Development of operational guidelines for
equitable distribution of surface water to Chennai city
[PhD hesis]. Chennai: Faculty of Civil Engineering, Anna
University; 2010. Available from: http://shodhganga.inlib-
net.ac.in:8080/jspui/bitstream/10603/27024/8/08_chap-
Figure 11. Pre- and post-monsoon changes in areal ter%203.pdf
extent of water bodies within CMA.

6 Vol 9 (18) | May 2016 | www.indjst.org Indian Journal of Science and Technology
 Nikita Roy Mukherjee and Christopher Samuel

4. USGS [Internet]. Available from: http://earthexplorer.usgs. 9. Xu H. Modiication of Normalised Diference Water Index
gov/ (NDWI) to enhance open water features in remotely sensed
5. Landsat 7 Science Data Users Handbook [Internet]. imagery. International Journal of Remote Sensing. 2006 Jul;
2012. Available from: https://pubs.er.usgs.gov/publica- 27(14):3025–33.
tion/7000070 10. Climate Madras/Minambakkam. Climate data: 1957-2016.
6. Zanter K. Landsat 8 (L8) data users handbook. LSDS-1574 [Internet]. Reported date by Weather Station: 432790
Version 1.0 [Internet], 2005. Available from: landsat.usgs. (VOMM). 2016. Available from:http://en.tutiempo.net/cli-
gov/l8handbook.php mate/ws-432790.html
7. Shen L, Li C. Water body extraction from Landsat ETM+ 11. Indian Metrological Department. Posted by: Nisreen
imagery using adaboost algorithm. IEEE. 2010 18th Inter- Ahmed, posted on: 1st September, 2012. 2016. Avail-
national Conference on Geoinformatics; Beijing. 2010. Jun able from:http://www.indiawaterportal.org/articles/dis-
18-20. p. 1–4. trict-wise-monthly-rainfall-data-list-raingaugestations-in-
8. McFeeters SK. he use of the Normalized Diference Wa- dia-meteorological-department
ter Index (NDWI) in the delineation of open water fea-
tures. International Journal of Remote Sensing. 1996 May;
17(7):1425–32.

Vol 9 (18) | May 2016 | www.indjst.org Indian Journal of Science and Technology 7