Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Indian Journal of Geo Marine Sciences Vol. 48 (08), August 2019, pp. 1258-1266
Assessment of wetland change dynamics of Chennai coast, Tamil Nadu, India, using satellite remote sensing
T. German Amali Jacintha1, S. R. Radhika Rajasree2,3*, J. Dilip Kumar4, & J.Sriganesh5 1Centre for Remote Sensing and Geoinformatics, Sathyabama Institute of Science & Technology, Jeppiaar Nagar,
Rajiv Gandhi Road, Chennai, Tamil Nadu, India 2Centre for Ocean Research, Sathyabama Institute of Science & Technology, Jeppiaar Nagar, Rajiv Gandhi Road,
Chennai, Tamil Nadu, India 3Faculty of Fisheries, Kerala University of Fisheries and Ocean Studies, Cochin, Kerala, India.
4Marine Biotechnology Div., ESSO-National Institute of Ocean Technology, Chennai, Tamil Nadu, India 5Department of Ocean Engineering, IIT- Madras, Chennai, Tamil Nadu, India
*[E-mail: [email protected]]
The coastal wetlands of Chennai are increasingly being affected by anthropogenic factors, such as urbanization, residential, and industrial development. This study helps to monitor and map the dynamics of the coastal wetlands of Chennai using Landsat satellite images of 1988, 1996, 2006, and 2016 by following a supervised classification method. Post-classification wetland change detection was done in three temporal phases, that is, 1988-1996, 1996-2006, and 2006-2016. Change detection matrix analysis was performed to identify the from-to changes. Ground truthing was carried out to validate the wetland classes. The overall accuracy of the classified image was 79.29% and the kappa coefficient was 0.7600. These results were imported into a GIS environment for further analysis. It was found that the wetlands have decreased to an alarming extent in the past 28 years from 23.14% in 1988 to 15.79% in 2016 of the total study area, owing to conversion of wetlands into industrial development, urban expansion, and other developmental activities.
[Keywords: Coastal wetlands; Change detection; Landsat imageries; Supervised classification]
Introduction Wetlands are essential ecological features in any
landscape. They perform vital functions by providing habitat, improving water quality, recharging groundwater aquifers, reducing erosion, and mitigating flood severity; they are often characterized by high rates of primary production1,2. Environmental researchers in recent years have identified a trend in deterioration of the extent and health of wetland areas3,4. During the past three decades, the coastal regions of Chennai have been rapidly developed for human habitation and other development projects in the form of Special Economic Zones being carried out in large scales, leading to the exploitation of wetlands5,7. Documenting losses and gains in wetland areas and monitoring changes in wetland types are the need of the hour. A variety of remote sensing methods are available to map wetland areas8,14. Supervised and unsupervised classifications are common techniques used for wetland studies across the globe. Clement et al.8
used ASTER and Landsat satellite data to classify and monitor the coastal wetlands in north-eastern
New South Wales and used a supervised classification method with maximum likelihood standard algorithm. Bakeret al.15 used stochastic gradient boosting (SGB) to classify the image and change vector analysis (CVA) to identify locations where changes occurred. High-resolution satellite data, such as Quick bird, World View, and IKONOS are also used for detailed mapping of wetlands16,17. By acquiring suitable temporal data, it is possible to identify each wetland class and to assess net gain or loss of each wetland classification13,15. El-Hattab14
used a post-classification approach based on the comparative analysis of independently produced classification images of the same area at different dates to detect and assess land cover changes of the Abu Qir Bay zone. They also used socio-economic data along with satellite data to indicate relationship between land cover and land use changes which reflected the changes in occupation status of settlers in specific areas. The present study utilized a post-classification change detection method to assess the changes of wetlands along the coastal regions of Chennai using the time series of remotely sensed data.
JACINTHA et al.: WETLAND CHANGE DYNAMICS USING REMOTE SENSING
1259
Materials and Methods
Study area The study region (Fig. 1), bound by Kalanji village
in the north (13° 22' 2.22"N, 80° 20' 16.79"E) and Kadalur village in the south (12° 26' 53.55"N, 80° 8' 38.43"E), spreads across three coastal districts of Tamil Nadu, namely, Tiruvallur, Chennai, and Kancheepuram and covers Chennai urban and semi-urban areas. The annual rainfall of the study area varies between 1100 and 1250 mm. There are several public sector and private industries, such as the North Chennai Thermal Power Station, EID Parry fertilizer plant, Ennore Thermal Power Station, Chennai Petroleum Corporation, and Tamilnadu Petro Product in north Chennai, and Indira Gandhi Centre for Atomic Research, Nemmeli Seawater Desalination Plant located in the south. Beach resorts, farm houses, theme parks, tourism hotspots, and amusement parks are mainly located south of Chennai. There are several wetland ecosystems, such as the Ennore creek, Adyar estuaries, Pallikaranai marshland, Great Salt Lake (near Nemmeli), and Muttukadu creek. The Pallikaranai marshland is one of the freshwater ecosystems in Chennai, which plays a vital role for aquatic species and also provides the habitat for many migratory birds. Recently, the Government of Tamil Nadu declared the Pallikaranai marshland as a reserved wetland forest.
Data used We used temporal Landsat images of 30 m
resolution of multispectral images in this study.
The Survey of India toposheets of 66C7, 66C8, 66C4, 66D1&5, 66d2, 66D3&4 were used as reference data set. The ground truth Global Positioning System (GPS) points were collected for accuracy assessment. Methodology
Image pre-processing The detailed methodology is shown in the
flowchart in Figure 2. The Landsat satellite images of specific dates of 1988, 1996, 2006, and 2016 (Table 1) were processed by converting digital numbers to reflectance values and further subjected to dark object subtraction, which assumes that reflectance from dark objects includes a substantial component of atmospheric scattering. This enables measurement of reflectance from a dark object, such as a deep lake, and that value is subtracted from the image3,18,19,21. Image classification
Supervised classification was performed on the reflectance images using a user-defined training site which requires digitizing polygons based on knowledge of the wetland classes obtained from regular field visits. A supervised maximum likelihood classification algorithm was used to detect any change in wetland area. The entire study area was classified
Table 1 — Satellite data used
Year Sensor Date
1988 LANDSAT 5 TM Feb 5, 1988 1996 LANDSAT 5 TM June 3, 1996 2006 LANDSAT 5 TM June 15, 2006 2016 LANDSAT 8 OLI Apr 23, 2016
Fig. 2 — Detailed methodology flowchart for this study
Fig. 1 — Study area map
INDIAN J. MAR. SCI., VOL. 48, NO. 08, AUGUST 2019
1260
into 10 categories, such as agriculture (paddy field, current fallow land), forest/plantations (including natural and planted forest), urban (land for settlement, industry), shrubs/dry grassland/open land, mudflats, beach/dunes/sand plains, marshy land/wet grassland, mangroves, aquaculture/saltpan, and water bodies (rivers, canals, creeks, water tanks, and sea)22.
Post-classification processing The classification process sometimes misclassifies
pixels in neighboring areas that are not an accurate representation. To avoid that, a majority/minority analysis is performed on a classification image to change false pixels within a large single class. Clump and Sieve operations are performed for generalizing classification images23. The classified maps were later exported to Geographic Information System ArcGIS software for analysis of changes in wetland size and extent.
Change detection analysis The classified maps of 1988, 1996, 2006, and 2016
were overlaid using the intersect tool in ArcGIS software and a pivot table analysis was performed. A post-classification change detection matrix was derived to identify from-to changes in the study area. To determine the change in the rate of land use categories over the study periods, the land use dynamic degree (LUDD) was calculated24.The computational equation is given by
LUDD (%) = (Yb − Ya) / (Ya*T)*100,
where Yb and Ya are the areas of cover type in time b and a, respectively and T is the interval between b and a. Results and Discussion
In the present study, the land use was analyzed for four different years of three decades (Fig. 3). The summary of landuse/landcover classification statistics of 1988, 1996, 2006, and 2016 are discussed in Table 2. The total agricultural area of 1988, 1996, 2006, and 2016 is 24.84%, 19.60%, 17.24%, and 12.64%, respectively, which showed a gradual decline. The loss of agricultural land is attributed mainly to rapid expansion in urbanization5. Owing to exponential growth of human population as well as rapidly increasing land value of real estate, most of the agricultural lands in semi-urban areas are converted into residential areas25. Aquaculture sites and saltpans showed an increase in the coastal wetland area (Fig. 4a) due to demand for shrimps and
fishes and salt products that provide economic benefits. The total area of aquaculture, which was 2.25% in 1988, increased to 2.77%, 4.19%, and 4.46% in 1996, 2006, and 2016, respectively. On the other hand,the total area of beaches and sandy area, which was 5.75% of the total area in 1988, reduced to 5.55% in 1996 and 4.28% in 2006 (Fig. 4b). In 2016, the beaches showed a drastic reduction to 2.42%, which may be attributed to the developmental activities such as industrialization, new ports in north Chennai and beach resorts, theme parks, and farm houses in south Chennai26. Forest/plantations include forest trees and casuarina plantations. The decadal variation showed an increasing trend over the period of 28 years, namely, 2.75%, 3.96%, 5.99%, and 5.34% in 1988, 1996, 2006, and 2016, respectively. A notable increase in forest/plantations observed in the present study is due to planting of casuarina trees promoted by state and central governments as bioshield28.The total area of mangroves in 1988 was 0.21%, which was reduced to 0.13% in 1996 (Fig. 4c). However, an increase was noticed from 0.23% to 0.39% between 2006 and 2016 because of mangrove restoration activity undertaken by the Tamil Nadu forest department. This increase was mainly noticed after the 2004 tsunami in which mangroves acted as shelter belt and reduced the impact of waves and wind in the coastal region27,28. During 1998, the marsh land covered 5.99% of the total study area and this was reduced to 3.10% (Fig. 4d). This is mainly because of the wetland being used as the City Corporation’s waste dumping yard and many developmental activities, such as the Mass Rapid Transit System, Special Economic Zone, and State Industries Promotion Corporation of Tamilnadu Ltd., (SIPCOT)29,31. There was significant decrease in mud flats from 8.93% to 5.42% due to conversion of aquaculture and development of industries such as the North Chennai Thermal Power Station (Fig. 4e). The area of shrubs (including scrubland, grassland, herbs, and wasteland) in 1988, 1996, 2006, and 2016 were 19.31%, 23.13%, 19.91%, and 18.33%, respectively. Much area in this land cover decreased owing to conversion to other land use classes. Urban area showed a dramatic increase from 21.32% to 39.21%. Chennai is the fourth largest metropolitan city in India, the city limits having expanded rapidly owing to increased population and industrialization. Water bodies including river canals, streams, water tanks, creeks, estuaries, etc., which covered 8.64% of the study area in 1988, reduced to 7.94% in 2006. In
JACINTHA et al.: WETLAND CHANGE DYNAMICS USING REMOTE SENSING
1261
2016, there was marginal increase in water bodies from 8.64% in 1988 to 8.69% in 2016 due to heavy rainfall during the year 2015. Accuracy assessment
To evaluate the accuracy of the classification, accuracy assessment was performed for the year 2016. For each class of 10 ground control points (GCPs), the sum up to a total of 100 GCPs was collected from the entire study area by using field
GPS points as well as high-resolution reference images. The accuracy assessment was performed with an error matrix with producer’s and user’s accuracy. The highest producer’s accuracy was 93.36 for forest and plantations, whereas the lowest producer’s accuracy was 54.04 for mudflats which was misclassified as beaches or some other class that could be rectified in post-classification editing. The highest user accuracy was 94.04 for urban class and lowest value was 50.01 for beaches and sand dunes.
Fig. 3 — Land use/land cover map of the study area in the years: (a) 1988, (b) 1996, (c) 2006 and (d) 2016
INDIAN J. MAR. SCI., VOL. 48, NO. 08, AUGUST 2019
1262
Table 2 — Summary of land use/land cover classification statistics of 1988, 1996, 2006, and 2016
Class Area of 1988 (ha)
Area of 1996 (ha)
Area of 2006(ha)
Area of 2016 (ha)
% of total area in 1988
% of total area in 1996
% of total area in 2006
% of total area in 2016
Agriculture 15629.97 12332.99 10845.61 7949.63 24.84 19.60 17.24 12.64 Aquaculture/Saltpan 1418.22 1741.83 2638.88 2808.38 2.25 2.77 4.19 4.46 Beaches/Dunes/ Sand plains 3615.14 3494.15 2692.27 1521.04 5.75 5.55 4.28 2.42 Forest/Plantations 1728.72 2494.63 3769.21 3357.14 2.75 3.96 5.99 5.34 Mangroves 133.19 83.97 144.51 247.63 0.21 0.13 0.23 0.39 Marshland/Wet Grassland 3771.75 2628.32 2319.49 1949.91 5.99 4.18 3.69 3.10 Mudflats 5621.21 4744.41 3463.07 3410.63 8.93 7.54 5.50 5.42 Shrubs/Dry Grassland 12146.11 14553.67 12526.92 11535.30 19.31 23.13 19.91 18.33 Urban 13415.00 17148.67 19520.85 24668.29 21.32 27.26 31.03 39.21 Water bodies 5437.56 3694.23 4996.07 5468.93 8.64 5.87 7.94 8.69 Total 62916.87 62916.87 62916.87 62916.87 100.00 100.00 100.00 100.00
Fig. 4 — Changes in wetland dynamics: (a) Aquaculture/saltpans, (b) Beaches/dunes/sand plains (c) Mangroves, (d), Marshland/wet grassland, Changes in wetland dynamics: (e) Mudflats
The overall 79.2966% an Change dete
A land usprepared tooccurred in 1996-2006, areas in the of two datesthe matrix. Tcolumn repr(Table 3), thland which shrub land (
1 9 8 8
AG
AQ BS
FP
M
ML MF SH U W
GT 1996 Note: AG-A
Marshland,
A
1 9 9 6
AG 808AQ 11BS 20FP 15M —
ML 12MF 50SH 137U 37W 14GT
2006 108
Note: AG-AMarshland, S
JACI
accuracy fond the kappa
ection results se/land cover understand each categoryand 2006-20change matri
s which are pThe rows reprresents the lahere is a perce
is distribute3359.30 ha) a
AG
9344.90
33.34 237.31
48.82
—
222.27 532.09 1272.43 232.47 405.98
12329.60
Agriculture, AQSH-shrubs, MF-
AG AQ
80.89 213.011.40 948.96
08.80 126.986.16 1.08 — 1.71
20.89 — 03.28 898.6277.22 120.71
70.77 3.96 4.96 323.74
844.36 2638.7
Agriculture, AQSH-shrubs, MF-
NTHA et al.: W
or the classifcoefficient w
change detecthe from-to
y for the peri016 (Tables 3ix represent uplaced along resent the earlater year. Dueptible changed to other cand urban are
Table 3 —
AQ BS
24.21 222.2
703.35 6.301.17 1939
— 59.2
0.99 —— 6.75
492.20 414.964.46 568.51.08 53.4
454.38 222.51741.83 3494
Q-Aquaculture/s- mudflats U- U
Table 4 — C
BS
1 126.19 6 2.88 8 1717.65
91.73 —
10.62 2 108.10 1 458.07
89.57 4 87.36 78 2692.18 3
Q-Aquaculture/smudflats, U- Ur
WETLAND CHA
fied image wwas 0.7600.
ction matrix wo changes tiods 1988-193-5). Theshadunchanged arthe diagonal
lier year and uring 1988-19ge in agricultuclasses such ea (1575.26 h
— Change matrix
S FP
29 358.35
0 — .94 124.72
26 864.34
— 2.07
5 104.37 97 3.96 56 423.07
45 590.80 59 22.71 2.12 2494.38
saltpans, BS-BeUrban, w-water b
Change matrix of
FP M
329.27 — 0.30 3.5135.38 0.27
1467.68 2.221.62 35.50
127.04 0.0955.80 33.8
574.39 28.51160.62 23.0
16.85 17.503768.94 144.5
saltpans, BS-Berban, w-water bo
ANGE DYNAMI
was
was that 96, ded reas l of the
996 ural
as ha).
In thconveurbanconveduringwere lands develoand tr
Duagricu1029.periodand, 2develo
of the study area
1 9 9 6
M ML
– 286.01
4.50 0.18 — 5.40
1.26 15.17
33.97 —
1650.90 11.38 102.86 11.89 135.44 0.09 132.21
20.88 299.84 83.97 2628.01
eaches/Dunes/Sabodies, and GT-G
f the study area
2006
ML
253.75 1 — 7 4.68 2 31.46 0 1.71 9 1219.93 1 106.67 5 416.13 5 274.84 0 10.16 51 2319.33
eaches/Dunes/Saodies, GT- Gran
ICS USING REM
he same pererted to shrub. The area o
ersion of 492g the same peconverted towere convert
opment of carees in parks. uring 1996-2ultural land w74 ha went ind, 409.82 ha o281.36 ha mopment such
a for the period
MF
414.67 33
401.72 155.32 7
1.44 4
59.29 1197.85 10
2368.14 8774.02 6818.00 4453.95 6
4744.41 14
and plains, FPGrand total –
for the period 19
MF S
376.64 182338.96 8.287.57 525
1.35 28123.78 3.17.64 306
1703.81 649559.75 808232.23 790
121.21 553462.94 1252
and plains, FPnd total, and –
MOTE SENSIN
riod, 1067.08b land and 51of aquacultur2.20 ha of meriod, while 5 mudflats anted into foresanopies along
2006 (Table was convertento urban devof beaches, 49
mudflats were h as construc
1988-1996
SH U
359.30 1575.
45.69 8.3456.94 459.2
28.77 306.1
12.78 4.50067.08 512.993.06 237.8816.78 1948.12.90 1189158.75 203.9552.05 17147
P-Forest plantat No change
996-2006
H U
3.32 1029.783 4.23
5.25 409.821.09 425.5051 0.09
6.39 492.719.20 281.362.80 2507.6
0.32 14256.6.13 112.61
25.83 19520.3
P-Forest plantat No change
NG
8 ha marshl12 ha of marre increased
mudflats to aq59.29 ha of mnd 590.80 ha st/plantations g the major
4), 1823.3ed to shrublavelopment. D92.71 ha of m
converted inction of ports
W
.26 43.58
4 114.81 27 34.84
14 3.52
0 19.59 90 9.43 82 564.73 .38 130.35 .39 79.39 90 2693.92
7.90 3694.18
tions, M-Mangre
W
74 98.66 422.77
2 177.65 0 36.23
16.05 1 332.76 6 403.76 61 427.23 67 146.20 1 2934.67 33 4995.98
ions, M-Mangre
1263
land was rshland to owing to
quaculture mangroves
of urban owing to roadsides
2 ha of and, while During this marshland, nto urban s, MRTS,
GT-1988
15628.57
1418.22 3614.91
1728.72
133.19 3771.56 5621.21
12145.38 13411.78 5436.91
62910.45
roves, ML-
GT-1996
12331.47 1741.83 3494.06 2494.50
83.97 2628.07 4744.41
14552.46 17148.23 3694.19
62913.18
roves, ML-
1264
and SIPCOTmudflats we2006-2016 (1187.03 ha oof shrublandthe same sandplains, mudflats, aconverted in
Tables 3-5bodies, indicChennai overthat water boaccretion alon Change rate
The rate shown in Tahighest rate period of plantations, wmangroves hin the peri+7.21 Y−1 anand 2006-2had a positi28 years. Wpositive tresandplains, shrubs, and dthe wetlandmudflats arehad high dec(−)1.40 % Marshland/w
2006
AGAG 6515AQ 19.7BS 142.FP 183.M —
ML 12.8MF 95.6SH 700.U 268.W 9.54GT
2016 7947
Note: AG-AgrSH-shrubs, M
T. During theere converte(Table 5), 81of forest and
d were converperiod, 668685.80 ha o
and 328.40 nto urban area5 show the concating erosionr a period of 2odies were cong the coast in
e of change isable 6. The of positive c1988-2016
which had thehad a negativiod 1988-19nd +7.14% Y
2016, respective change o
Water bodies aends, while
marshland/dry grassland
d types, beae the most domcreasing chanY−1 (Table
wet grassland
G AQ .92 160.57 72 1629.66 38 24.73 16 3.15
— 0.94 89 — 65 358.30 59 61.54 13 8.46 4 560.90 .98 2808.25
riculture, AQ-AMF- mudflats U-
INDIAN
e same period into aqua9.35 ha of ag
d plantations, rted into urba.26 ha of
of marshlandha of wat
as. nversion of ben along the c28 years. It wonverted to bn the same per
s determinedaquaculture/s
change of +3followed b
e rate of + 3e rate of chan96. But thi
Y−1 in the pertively. Overf +3.07% Y−
and urban lanagriculture,
/wet grasslad had negativeaches/dunes/saminant wetlannge rate of (−)6) over the s have reduce
Table 5 — C
BS 30.99 378.66 0
951.40 1822.04 171.44 01.35 3
13.14 2370.85 4972.09 4749.01 1
1520.97 33
Aquaculture/saltpUrban, w-water
N J. MAR. SCI.
d, 898.62 haculture. Durgricultural laand 1187.03
an areas. Durbeaches/dun
d, 346.44 ha terbodies w
eaches into wacoastal region
was also observeaches owingriod.
d by the LUDsaltpans had .5% Y−1 for by forest a3.36% Y−1. Tnge −4.62% Yis increased riods 1996-20rall, mangrov−1 over the pnd use also h, beaches aand, mudflae trends. Amoand plains and groups wh)2.07 % Y−1 apast 28 yea
ed at the rate
Change matrix of
FP M 71.66 0.09 0.72 4.11 81.00 —
781.60 5.53 0.81 70.90 5.63 7.73 2.52 56.92 92.66 40.35 73.71 2.52 6.48 59.49 56.79 247.63
pans, BS-Beacher bodies, GT-Gra
, VOL. 48, NO.
a of ring and,
ha ing
nes/ of
were
ater n of ved g to
DD the the and The Y−1
to 006 ves
past had and ats, ong and
hich and ars. e of
(−) 1.dominper antotal s2016
f the study area
2 0 1 6
ML 50.25 10.09 57.11 4
134.19 —
853.91 64.89 15400.83 4283.26 5155.29 4
1949.83 34
es/Dunes/Sand pand total, and
Class
AgricuAquacuBeached plainForest/MangroMarshlGrasslaMudflaShrubs/GrasslaUrban darea Water b
Table
Class N
AquacuSaltpanBeacheSand plMangroMarshlGrasslaMudflaTotal
08, AUGUST 2
73% Y−1 (Tanant class winnum. The tostudy area, wh(Table 7). O
for the period 20
MF SH179.43 2544562.05 62.344.38 490.9.09 399.
14.88 0.85837 414.573.67 426.427.75 637450.49 737.490.51 87.0410.61 11537
lains, FP-Forest – No chan
Table 6 —
19
lture ulture/Saltpan es/Dunes/Sans Plantations oves land/Wet and ats /Dry and development
bodies
e 7 — Quantific
Name % wetlan
198
ulture/ n
2.2
es/Dunes/ lains
5.7
oves 0.2land/Wet and
5.9
ats 8.923.
2019
able 6). The uith an increastal wetland ahich was redu
Overall, this s
006-2016
H U 4.08 819.35 38 28.63 43 668.26 44 1187.031 9.17 44 685.80 68 346.44
4.68 3146.0207 17446.7
03 328.40 7.05 24665.8
plantations, M-nge
— Rate of chang
988-1996 in %
1996-2in %
-2.64 -1.22.85 5.1-0.42 -2.2
5.54 5.1-4.62 7.2-3.79 -1.1
-1.95 -2.72.48 -1.3
3.48 1.3
-4.01 3.5
cation of wetlandperiod 1988-20
of nd in 88
% of wetland
1996
25 2.77
75 5.55
21 0.13 99 4.18
93 7.54 14 20.17
urban area issing rate of
area was 23.14uced to 15.79study shows
W 172.03 322.75 182.49
3 43.63 45.57 249.03 524.74
2 510.71 8 178.02
3239.37 9 5468.34
Mangroves ML
e (LUDD) Y−1
2006 %
2006-2016in %
21 -2.67 5 0.64
29 -4.35
1 -1.09 21 7.14 18 -1.59
70 -0.15 39 -0.79
38 2.64
52 0.95
d change dynami016
in % of
wetland in 2006
4.19
4.28
0.23 3.69
5.50 17.89
the most 3.0% Y−1 4% of the 9% during that there
GT-2006 10844.37 2638.78 2692.18 3768.86 144.51
2319.16 3462.94 12525.99 19520.53 4996.01 62913.34
-Marshland,
6 1988-2016 in %
-1.75 3.50 -2.07
3.36 3.07 -1.73
-1.40 -0.18
3.00
0.02
ics for the
% of wetland in
2016
4.46
2.42
0.39 3.10
5.42 15.79
JACINTHA et al.: WETLAND CHANGE DYNAMICS USING REMOTE SENSING
1265
was a gradual change found in the wetlands along the coastal regions of Chennai during 1988-2016. Conclusion
The coastal wetlands have been reduced in Chennai city owing to rapid growth of urbanization during the past three decades. The beaches/dunes/sand plains, marshland and mudflats in the study area show a decreasing trend, whereas other wetland types such as aquaculture and saltpans, mangroves have increased significantly. Overall, the wetland areas have decreased owing to developmental activities and urbanization of the Chennai metropolitan city. We show that satellite remote sensing is an effective tool for detecting coastal landuse changes and the open source Landsat 5 thematic mapper and Landsat 8 OLI sensors with multi-spectral bands having 30 m resolution are very useful in the detection of the dynamics of wetland. The post-classification change detection is a very useful method in identifying changes in wetlands, which is done after ground truth verification. In future, high-resolution images from unmanned aerial vehicle (UAV) can be used for detailed mapping of wetlands. Acknowledgment
The authors are thankful to Sathyabama Institute of Science and Technology for providing necessary facilities for the research work. References 1 Ma, C., Zhanga, G.Y., Zhanga, X.C., Zhaoa, Y.J., Lib, H.Y.,
Application of Markov model in wetland change dynamics in Tianjin Coastal Area, China, Procedia Environ. Sci.,13(2012) 252-262.
2 Ozesmi, S.L., Bauer, M.E., Satellite remote sensing of wetlands, Wetl. Ecol. Manage.,10(5)(2002)381-402,https://doi.org/10.1023/ A:102090843248.
3 Ekstrand, S., Assessment of forest damage with Landsat TM: Correction for varying forest stand characteristics, Remote Sens. Environ., 47(1994) 291-302.
4 Munyati, C., Wetland change detection on the Kafue flats, Zambia, by classification of a multitemporal remote sensing image dataset, Int. J. Remote Sens.,21(2000) 1787-1806.
5 Bharath, H., Aithal, T.V., Ramachandra visualization of urban growth pattern in Chennai using geoinformatics and spatial metrics, J. Indian Soc. Remote Sens.,44(4)(2016) 617–633,doi: 10.1007/s12524-015-0482-0.
6 Chandramohan, D.B., Bharathi, D., The role of public governance in conservation of urban wetland system: A study of Pallikaranai marsh, paper presented at the Proceedings of The Indian Society for Ecological Economics (INSEE), 5th Biennial Conference, Ahmedabad, India, 21- 23 January 2009.
7 Sujatha, V., Role of Pallikaranai marsh in moderating the flood, M.E thesis, Anna University, India, 2010.
8 Akumu, C.E, Pathirana, S., Baban, S., Bucher, D., Monitoring coastal wetland communities in north-eastern NSW using ASTER and Landsat satellite data, Wetl. Ecol. Manage.,18(2010) 357-365, doi: 10.1007/s11273-010-9176-0.
9 Huang, N., Wang, Z., Liu, D., Niu, Z., Selecting sites for converting farmlands to wetlands in the Sanjiang Plain, Northeast China based on remote sensing and GIS, Environ. Manage., 46(2010) 790-800, doi: 10.1007/s00267-010-9547-6.
10 Zhao, H., Cui, B., Zhang, H., Fan, X., Zhang, Z., Lei, X., A landscape approach for wetland change detection (1979–2009) in the Pearl River Estuary, Procedia Environ. Sci.,2(2010) 1265-1278.
11 Jensen, J.R., Introductory digital image processing: A remote sensing perspective, 3rd edition,(Prentice Hall, USA) 2005., 337-399
12 Rokni, K., Ahmad, A., Selamat, A., Hazini, S., Water Feature Extraction and Change Detection Using Multitemporal Landsat Imagery. Remote Sens. 2014, 6, 4173-4189. doi:10.3390/rs6054173.
13 El-Hattab, M.M., Change detection and restoration alternatives for the Egyptian Lake Maryut, Egypt. J. Remote Sens. Space Sci.,18(2015) 9-16.
14 El-Hattab, M.M., Applying post classification change detection technique to monitor an Egyptian coastal zone (Abu Qir Bay),Egypt. J. Remote Sens. Space Sci.,19(2016) 23-36.
15 Baker, C., Rick L. Lawrence, Montagne, C., and Patten, D., Change detection of wetland ecosystems using landsat imagery and change vector analysis, Wetlands, 27(3)(2007) 610-619.
16 Maxa, M., Bolstad, P., Mapping northern wetlands with high resolution satellite images and LiDAR,Wetlands,29(2009) 248, https://doi.org/10.1672/08-91.1.
17 Ahmad, S.S., and Erum, S., Remote sensing and GIS application in wetland change analysis: Case study of Kallar Kahar, Sci., Tech. Dev., 31(3) (2012) 251-259.
18 Chavez, P.S., Jr., Mackinnon, D.J., Automatic detection of vegetation changes in the south-western United States using remotely sensing images, Photogramm. Eng. Remote Sens., 60(5)(1994) 571-583.
19 Conghe, Song, Curtis, E., Woodcock Karen, C., Seto, Mary Pax Lenney, and Scott A. Macomber, Classification and change detection using landsat TM data: When and how to correct atmospheric effects? Remote Sens. Environ., 75(2001)230-244.
20 Jakubauskas, M.E., Thematic mapper characterization of Lodgepole pine seral stages in Yellowstone National Park USA, Remote Sens. Environ., 56(1996) 118-132.
21 Spanner, M.A., Pierce, L.L., Peterson, D.L., Running, S.W., Remote sensing of temperate coniferous forest leaf area index: The influence of canopy closure, understory vegetation and background reflectance, Int. J. Remote Sens.,11(1)(1990) 95-111.
22 SAC Application Centre (ISRO), Ahmedabad, 2012, Coastal Zones of India, Retrievedfrom: http://www.moef.nic.in/ sites/default/files/Coastal_Zones_of_India.pdf.
23 Harris Geospatial Solutions, Sieve Classes, 2018, retrieved from https://www.harrisgeospatial.com/docs/SievingClasses.html
INDIAN J. MAR. SCI., VOL. 48, NO. 08, AUGUST 2019
1266
24 Mozumder, C., and Tripathi, N.K., Use of multiple satellite images in multiple scales for feature extraction and image classification: A case study of Ramsar Wetland in North East India, paper presented at The 33rd Asian Conference on Remote Sensing, Thailand Nov 26–30, 2012
25 Purushothaman, Pratheep. (2014, December 12). ‘Buy Land Today and Become Lord Tomorrow’ Land Governance in the Peri-Urban Region of Chennai. Governance, Policy and Political Economy (GPPE). Retrieved from http://hdl.handle.net/2105/17443.
26 Bassi N., Kumar M.D., Sharma A., Pardha-Saradhi P. Status of wetlands in India: A review of extent, ecosystem benefits, threats and management strategies, J. Hydrol. Reg. Stud., 2(2014) 1-19, https://doi.org/10.1016/j.ejrh.2014.07.001.
27 Danielsen, F., Sørensen, M.K., Olwig, M.F., Selvam, V., Parish, F., Burgess, N.D., Hiraishi, T., Karunagaran, V.M., Rasmussen, M.S., Hansen, L.B., Quarto, A., Suryadiputra, N.,
The Asian tsunami: A protective role for coastal vegetation, Science, 310(5748)(2005) 643.
28 Kathiresan, K., Rajendran, N., Coastal mangrove forests mitigated tsunami, Estuar. Coast. Shelf Sci., 65(2005) 601-606.
29 Vencatesan, J., Daniels, R.J.R., Jayaseelan, S. S. Karthick, M.N., Comprehensive management plan for Pallikaranai marsh, Technical Report submitted to the Conservation Authority for Pallikaranai Marsh, Tamilnadu Forest Department,2014, p.264.
30 Venkatachalam, L., Jayanthi, M., Estimating the economic value of ecosystem services of Pallikaranai marshin Chennai City: A Contingent Valuation Approach, MIDS Working Paper No. 220, Chennai, 2016.
31 Surya. S., Landscape ecological urbanism for restoration of Pallikaranai marsh land, Chennai, Tamil Nadu, Procedia Technol., 24(2016)1819-1826.