International Journal of Electrical and Computer Engineering (IJECE)

Vol. 8, No. 6, December 2018, pp. 4111~4119

ISSN: 2088-8708, DOI: 10.11591/ijece.v8i6.pp4111-4119  4111

Extraction of Water-body Area from High-resolution

Landsat Imagery

B. Chandrababu Naik, B. Anuradha

Department of Electronics and Communication, SVU College of Engineering, SV University, India

Article Info ABSTRACT

Article history: Extraction of water bodies from satellite imagery has been broadly explored
in the current decade. So many techniques were involved in detecting of the
Received Apr 16, 2018 surface water bodies from satellite data. To detect and extracting of surface
Revised Jul 10, 2018 water body changes in Nagarjuna Sagar Reservoir, Andhra Pradesh from the
Accepted Aug 2, 2018 period 1989 to 2017, were calculated using Landsat-5 TM, and Landsat-8
OLI data. Unsupervised classification and spectral water indexing methods,
including the Normalized Difference Vegetation Index (NDVI), Normalized
Keyword: Difference Moisture Index (NDMI), Normalized Difference Water Index
(NDWI), and Modified Normalized Difference Water Index (MNDWI), were
Index method used to detect and extraction of the surface water body from satellite data.
Landsat Instead of all index methods, the MNDWI was performed better results.
MNDWI The Reservoir water area was extracted using spectral water indexing
NDMI methods (NDVI, NDWI, MNDWI, and NDMI) in 1989, 1997, 2007, and
NDVI 2017. The shoreline shrunk in the twenty-eight-year duration of images.
NDWI The Reservoir Nagarjuna Sagar lost nearly around one-fourth of its surface
Surface water water area compared to 1989. However, the Reservoir has a critical position
in recent years due to changes in surface water and getting higher mud and
sand. Maximum water surface area of the Reservoir will lose if such
decreasing tendency follows continuously.
Copyright © 2018 Institute of Advanced Engineering and Science.
All rights reserved.

Corresponding Author:
B. Chandrababu Naik,
Department of Electronics and Communication,
SVU College of Engineering, SV University, India.
Email: [email protected]

1. INTRODUCTION
The remote sensing knowledge is mainly used in different areas, such as the lake, coastal zone
management, shoreline change and erosion monitoring, forest for monitoring of changes, forest and
vegetation changes [1], [2], disaster monitoring [3], [4], flood prediction and evaluation of water resources
[5]. It is essential for agriculture (food crops), day to day life of humans, and ecosystems [6]. To accomplish
the information about open surface water is most important in different scientific areas, those are surface
water analysis, watershed analysis, dynamic changes of rivers, environment monitoring, present and future
estimations of water resources, and flood mapping [7]-[10]. Remote sensing satellites are having 30m
resolutions and they offer a huge amount of data, which is widely used for detecting and extraction of surface
water areas and its dynamic changes in recent decades [11]-[17].
Identification of water is very significant for various precise estimations and human life.
To detecting and extraction of surface water area from satellite data has been introduced at many more image
processing techniques in the current decades. A single-band and multi-band methods were widely used in
Landsat imagery for detecting and extraction of surface water area along with selected threshold value, either
positive or negative value [9]. Compared with a single-band method, multi-band method was extensively
used for enhancing the surface water bodies [9]. Four different satellite multi-band methods were used for
extraction of surface water bodies, those are water indexing methods, including the NDVI, NDMI, NDWI,

Journal homepage: http://iaescore.com/journals/index.php/IJECE

4112  ISSN: 2088-8708

and MNDWI. Thus, these methods were introduced to an extraction of surface water bodies with specified
threshold values being either less than zero (negative value) or greater than zero (positive value). The NDVI,
NDMI, NDWI, and MNDWI for the detecting and extraction of surface water bodies from satellite imagery.
By using change detection, water features can extract separately for the different years of satellite data.
The Nagarjuna Sagar Reservoir is located in Nalgonda district in Telangana state. In recent decades,
there are tremendous changes in its capacity. Therefore, the dynamic changes of the Reservoir surface water
area need to monitor continuously. In this study, the spatiotemporal changes of Reservoir Nagarjuna Sagar
from 1989 to 2017 are investigated based on satellite-multiband water indexed methods, including NDVI,
NDMI, NDWI, and MNDWI using Landsat-5 TM and Landsat-8 OLI images for extraction of the surface
water body. Overall, the MNDWI were found superior to other indexes. These method is highly important for
time-series analyses of extracting shorelines using any number of Landsat images in different time intervals,
and it provides an important contrast that can be used to investigate shoreline changes.

2. STUDY AREA AND DATA SETS

The Nagarjuna Sagar Reservoir was built across the Krishna river, which is located in
between Guntur District and Nalgonda District in Telangana and its geographical area surrounded flanked by
latitude 16°34'55.60"N to 16°56'44.95"N and longitude 78°24'13.97"E to 78°47'06.07"E as shown in
Figure 1. Water spread area is 285 km2 at Full Reservoir Level (FRL), The catchment Area is
214,185 km2 (82,697 sq mi) and the gross storage capacity 405 TMC at FRL, In India, the Nagarjuna Sagar
Reservoir is the second biggest water reservoir and it’s also one of the most basic multi-purpose irrigation
and hydro-electric projects in India.

Figure 1. Location of nagarjuna sagar reservoir

The Landsat-5 TM and Landsat-8 OLI data obtained in January 1989, 1997, 2007, and 2017. These
data were taken from the US Geological Survey (USGS) portal, but collected all Landsat images were
cloud-free data. Table 1 represents the specification of Landsat-5 TM and Landsat-8 OLI images. The water
body was extracted from satellite images with a 30m spatial resolution and different band with different
wavelengths as shown in Table 1.

Table 1. Landsat-5 TM and Landsat-8 OLI data with its Specifications

Satellite Year Path/Row Resolution (m) Wavelengths (μm)
Band1:0.45-0.52
Band2:0.52-0.60
Landsat-5 1989 Band3:0.63-0.69
TM 1997 143/49 30 Band4:0.76-0.90
2007 Band5:1.55-1.75
Band7:2.08-2.35
Band1:0.433-0.453

Int J Elec & Comp Eng, Vol. 8, No. 6, December 2018 : 4111 - 4119
Int J Elec & Comp Eng ISSN: 2088-8708  4113

Table 1. Landsat-5 TM and Landsat-8 OLI data with its Specifications

Satellite Year Path/Row Resolution (m) Wavelengths (μm)
Band2:0.450-0.515
Band3:0.525-0.600
Landsat-8 2017 143/49 30 Band4:0.630-0.680
OLI Band5:0.845-0.885
Band6:1.560-1.660
Band7:2.100-2.300
Band9:1.360-1.390

3. METHODOLOGY
Considered the period of 1989 to 2017, that can show the changes of water area in Reservoir.
Introduces the methodology and its performances of different satellite –multiband indexes, including the
Normalized Difference Vegetation Index (NDVI) [18], Normalized Difference Moisture Index (NDMI) [19],
Normalized Difference Water Index (NDWI) [20], and Modified Normalized Difference Water Index
(MNDWI) [12], were used for detecting and extraction of surface water bodies from Landsat imagery as
shown in Table 2. Thus, four different years of satellite images (Landsat-5 data from 1989, 1997, 2007 and
Landsat-8 data from 2017) were performed and extraction of surface water bodies using different indexing
methods (NDVI, NDMI, NDWI, and MNDWI). Table 2. Represent the Landsat-5 data, indexes used for
water feature extraction (B2: band2=green, B3: band3=red, B4: band4=near infrared, B5: band5=middle
infrared. Similarly, in Landsat-8 data (B3: band3=green, B4: band4=red, B5: band5=near infrared,
B6: band 6=middle infrared).

Table 2. Satellite-multiband Indexes used for Water Feature Extraction

Index Equation Remark Reference
Normalized Difference Vegetation Index NDVI=(B4-B3)/(B4+B3) Water would be a negative value [28]
Normalized Difference Moisture Index NDMI=(B4-B5)/(B4+B5) Water would be a positive value [19]
Normalized Difference Water Index NDWI=(B2-B4)/(B2+B4) Water would be a positive value [20]
Modified Normalized Difference Water Index MNDWI=(B2-B5)/(B2+B5) Water would be a positive value [12]

The NDWI has introduced to detect and extracting the surface water bodies with a specified
threshold value. The positive threshold values for water and negative threshold values for nonwater
bodies [20]. The MNDWI has introduced a powerful index to detect and extracting the surface water bodies.
Because band5 (middle infrared) has replaced by the band4 (near infrared). Hence band5 reflectance’s more
compared with band4 [12]. The MNDWI widely used for suppressing errors from vegetation, soils, and built-
up areas. The threshold values of MNDWI have positives and negatives for water and nonwater bodies.
The NDVI has introduced mainly used for extracting green vegetation from other wetland surface areas.
Thus, NDVI also extracts the surface water much better than NDMI indexed method, and its threshold values
for water would be negative values [18]. The NDMI has introduced mainly for extracting vegetation, and
water liquid but it’s not much more capable of extracting water bodies as compared to others index methods
(NDVI, NDWI, and MNDWI). Thus, the NDMI method was not efficient for extraction of water bodies. The
NDMI threshold value of water would be a positive. Based on these analyses, the MNDWI method has
performed slightly better than other index methods (NDWI, NDVI, and NDMI).
The best significance of water body extraction techniques was recognized and employed to
spatiotemporal changes of the Nagarjuna Sagar Reservoir in the period 1989 to 2017. To detect and
extracting Reservoir surface water bodies in four different years, such as Landsat-5 (1989, 1997, 2007) and
Landsat-8 (2017) images. Out of four-year analyses of extracting Reservoir surface water area, the maximum
changes occur in the period of 2007-2017. In order to get the efficiency of detection and extraction of surface
water area, different accuracy analyses were performed. By using accuracy assessment analyses, calculate the
parameters are overall accuracy, producer’s accuracy, user’s accuracy, and kappa coefficient. Those
parameters were performed over the changes of water body in the period 1989 to 2017.

4. RESULTS AND ANALYSIS

Once the required satellite data obtained, The image processing techniques have been involved for
further processing. The analyses were performed before applying water indexed methods as shown in
Figure 2 (a-d). And after applying water indexed methods, obtained the results as shown in Figures 3,4,5, and
Figure 6 (a-d). Different satellite-multiband water indexed methods, including NDVI, NDMI, NDWI, and

Extraction of Water-body Area from High-resolution Landsat Imagery (B. Chandrababu Naik)
4114  ISSN: 2088-8708

MNDWI were used to detect and extract the surface water body from the Landsat-5 (1989, 1997, 2007) data
and Landsat-8 (2017) data.
The four spectral water indexed methods (NDVI, NDMI, NDWI, and MNDWI) are applied to the
Reservoir water area to highlight the differences between water and non-water bodies as shown in
Figures 3,4,5, and 6 (a-d). Out of all water indexed methods, the MNDWI has the better method for
separating water bodies as compared to other indexed methods (NDVI, NDWI, and NDMI). Generally,
threshold values of water areas having greater than zero values and vegetation areas having negative values.
First, calculate surface water area and changed surface water area for four different years with selected
Reservoir as shown in Table 4. Similarly, calculate the surface perimeter and changed surface perimeter for
four different years with same Reservoir ass hwon in Table 5.

Figure 2. Landsat-5 TM image from 1989 (a); Landsat-5 TM image from 1997 (b); Landsat-5 TM
image from 2007 (c); Landsat-8 OLI image from 2017 (d)

Figure 3. NDVI-based extracts water body from the Landsat-5 TM image 1989,1997, and 2007 (a-c);
Landsat-8 OLI image 2017 (d)

Int J Elec & Comp Eng, Vol. 8, No. 6, December 2018 : 4111 - 4119
Int J Elec & Comp Eng ISSN: 2088-8708  4115

Figure 4. NDMI-based extracts water body from the Landsat-5 TM image 1989,1997, and 2007 (a-c);
Landsat-8 OLI image 2017 (d)

Figure 5. NDWI-based extracts water body from the Landsat-5 TM image 1989,1997, and 2007 (a-c);
Landsat-8 OLI image 2017 (d)

Extraction of Water-body Area from High-resolution Landsat Imagery (B. Chandrababu Naik)
4116  ISSN: 2088-8708

Figure 6. MNDWI-based extracts water body from the Landsat-5 TM image 1989,1997, and 2007 (a-c);
Landsat-8 OLI image 2017 (d)

Figure 7. Changes in the area of Reservoir Nagarjuna Sagar generated using (a) MNDWI 1989; (b) MNDWI
and NDVI 1989 and 1997; (c) MNDWI, NDVI and NDWI 1989, 1997, and 2007; (d) MNDWI, NDWI and
NDVI 1989, 1997, 2007 and 2017.

4.1. Evaluation of the changes

The Reservoir water area was extracted using Unsupervised classification and spectral water
indexing methods in 1989, 1997, 2007, and 2017. The shoreline shrunk in the twenty-eight-year duration of
images. However, the Reservoir has been in a critical situation in recent years due to changes in surface water
and getting higher mud and sand. According to the NDVI, NDMI, NDWI, and MNDWI, between 1989 and
2007 there were little bit changes in Reservoir water area, and the changes between 2007 and 2017 there was
maximum changes in Reservoir water area as shown in Figure 7. The NDVI, NDMI, NDWI, and MNDWI

Int J Elec & Comp Eng, Vol. 8, No. 6, December 2018 : 4111 - 4119
Int J Elec & Comp Eng ISSN: 2088-8708  4117

results expose that the surface area of Reservoir Nagarjuna Sagar in January 1989, 1997, 2007, and 2017 was
approximately 205 km2, 203 km2, 196 km2, and 160 km2 as shown in Table 3. The result show that the
Reservoir surface water body changes around 2.131 km2 between 1989 to 1997, 6.494 km2 between 1997 to
2007, 36.016 km2 between 2007 to 2017. The overall changes in surface water around 44.641 km2 between
1989 to 2017 as shown in Table 4.

Table 3. Performance Evaluation of The Satellite-multiband Indexes used for Surface Water Extraction
Min Max Standard
Indexes Year Mean Perimeter (km) Area (km2)
Threshold Threshold Deviation
1989 -0.555 0.716 0.034 0.132 258.095 205.574
1997 -0.290 0.613 0.068 0.128 257.113 203.443
NDVI
2007 -0.375 0.672 0.040 0.155 252.296 196.949
2017 -0.582 0.416 0.119 0.083 212.021 161.529
1989 0.311 0.745 0.186 0.145 264.112 201.555
1997 0.304 0.774 0.282 0.152 260.436 201.101
NDMI
2007 0.334 0.778 0.318 0.182 254.346 192.115
2017 0.323 0.782 0.433 0.173 216.170 160.933
1989 0.607 0.924 0.446 0.188 261.423 204.229
1997 0.635 0.918 0.505 0.187 259.029 201.630
NDWI
2007 0.613 0.923 0.475 0.207 252.034 195.679
2017 0.623 0.987 0.544 0.198 211.232 161.411
1989 0.603 0.936 0.214 0.256 263.827 203.784
1997 0.601 0.950 0.250 0.236 258.978 201.879
MNDWI
2007 0.610 0.948 0.299 0.269 252.152 194.783
2017 0.617 0.988 0.407 0.264 213.627 161.335

Table 4. Statistics of the Reservoir Surface Area Changes

Year Surface water area (km2) Surface water area change (km2) Total surface water area changes (km2)
1989 205.574
1997 203.443 -2.131
2007 196.949 -6.494
-44.641
2017 160.933 -36.016

Similarly, The NDVI, NDMI, NDWI, and MNDWI results expose that the surface perimeter of
Reservoir Nagarjuna Sagar in January 1989, 1997, 2007, and 2017 was approximately 264 km, 260 km, 254
km, and 216 km as shown in Table 3. The result show that the Reservoir surface perimeter changes around
3.676 km between 1989 to 1997, 6.090 km between 1997 to 2007, 38.176 km between 2007 to 2017.
The overall changes in surface perimeter around 47.942 km between 1989 to 2017 as shown in Table 5.
The result show that decreasing surface water area in Reservoir from the period 1989 to 2017, and by
observing the resulting analyses of maximum changes in the surface water body, as well as surface perimeter,
occurred in the period 2007 to 2017. The Reservoir Nagarjuna Sagar lost nearly around one-fourth of its
surface area compared to 1989. The maximum surface water area changes are observed in the western region
and western - north parts of the Reservoir.

Table 5. Statistics of the Reservoir Surface Perimeter Changes

Year Surface perimeter (km) Surface perimeter change (km) Total surface perimeter change (km)
1989 264.112
1997 260.436 -3.676
2007 254.346 -6.090 -47.942
2017 216.170 -38.176

4.2. Accuracy assessment analyses

The probability (%) that the classifier has labeled an image pixel into the ground truth Class. It is the
probability of a reference pixel being correctly classified. Producer’s accuracy: It represent pixels that belong
to the truth class but fail to be classified into the proper class (omission error = number of correctly classified
pixels in each category per total number of classified pixels in that category (column total)), User’s accuracy:
Reliability, probability a pixel class on the map represents the category on the ground (commission
error=number of correctly classified pixels in each category per total number of classified pixels in that
category (row total)), Overall accuracy: The overall accuracy=Total number of correctly classified pixels
(diagonal) per total number of reference pixels, Kappa coefficient (Khat): It measure of agreement between
Extraction of Water-body Area from High-resolution Landsat Imagery (B. Chandrababu Naik)
4118  ISSN: 2088-8708

the classification map and the reference data. a discrete multivariate technique of use in accuracy assessment.
Khat>0.80 represent strong agreement and good accuracy. Khat=0.40-0.80 is middle, Khat<0.40 is poor.

(TS X TCS)− ∑(COL.TOTAL X ROW.TOTAL)

Kappa Coefficient = X100 (1)
TS2 − ∑(COL.TOTAL X ROW.TOTAL)

Where, TS=Total number of Samples, TCS=Total number of Correct Samples.

Statistical parameters of the accuracy assessment result show that MNDWI achieved 1.551km2 has
the absolute error, 96.06% has user’s accuracy, 93.18% has producer’s accuracy, 96.43% has overall
accuracy, and 0.8906 is kappa coefficients as shown in Table 6. Though, the index methods, including
MNDWI, NDWI, and NDVI provided good accuracies as compared with NDMI method. Instead of all
indexed methods, MNDWI provides better results than other methods (NDWI, NDVI, and NDMI) for
detecting and extracting surface water body in the Reservoir. Here, the Reservoir provides a lot of benefits
for agricultural (food crops), industrial, domestic purpose, and human existing in its surrounding. Thus, it
required correcting measurements to avoid any obstacles in the Reservoir to get its original conditions for
restoring the Reservoir.

Table 6. Statistical Parameters of Accuracy Assessment of Changes in Surface Water Body

Changed water Absolute Producer's User's Overall Kappa
Method
body (km2) Error (km2) accuracy (%) accuracy (%) accuracy (%) coefficient
Reference 44.000 0.000 100 100 100 1
NDVI 44.045 0.045 89.57 91.50 92.86 0.8096
NDWI 42.818 1.182 92.31 88.91 94.64 0.8100
MNDWI 42.449 1.551 93.18 96.06 96.43 0.8906
NDMI 40.622 3.378 87.59 88.96 91.67 0.7625

5. CONCLUSION
The main aim of this study is to detect the spatiotemporal changes of surface water area in
Nagarjuna Sagar Reservoir from the period of 1989 to 2017, by the satellite image interpretation and GIS.
Using satellite images to extract information regarding Reservoir water area change is faster and more
accurate than other observation methods. Several index methods (NDVI, NDWI, MNDWI, and NDMI) were
used for detecting and extracting surface water area. The result shows that decreasing surface water area
maximum changes occurred about 44.641 km2 in the period 1989 to 2017. Especially, from 2007 to 2017 the
Reservoir lost its surface water area about 36.016 km2. If such decreasing tendency follows continuously the
Reservoir will lose its maximum surface water area in near future. The statistical parameters of the accuracy
assessment result show that MNDWI provides better results as compared to other index methods (NDVI,
NDWI, and NDMI) so that the MNDWI method has good efficient in detecting and extracting surface water
body in Nagarjuna Sagar Reservoir. The future scope of these methods could be useful for detecting and
extracting (decreasing or increasing) open water surface area in the world with different band wavelengths
and different satellite data.

ACKNOWLEDGEMENTS
We would like acknowledge to the Centre of Excellence (CoE) for utilization of resources and
availability of software tools. The grateful thanks to the USGS portal for downloading Landsat data with a
free of cost.

REFERENCES
[1] S. Kaliraj, et al., “Application of remote sensing in detection of forest cover changes using geo-statistical change
detection matrices-A case study of devanampatti reserve forest, tamilnadu, India,” Nat. Environ. Polluti. Technol.,
vol. 11, pp. 261-269, 2012.
[2] V. Markogianni, et al., “Land-use and vegetation change detection in plastira artificial lake catchment (Greece) by
using remote-sensing and GIS techniques,” Int. J. Remote Sens., vol. 34, pp. 1265-1281, 2013.
[3] M. Volpi, et al., “Flooding extent cartography with Landsat TM imagery and regularized Kernel Fisher’s
discriminant analysis,” Comput. Geosci., vol. 57, pp. 24-31, 2013.
[4] B. Brisco, et al., “Sar polarimetric change detection for flooded vegetation,” Int. J. Digit. Earth, vol. 6, pp. 103-
114, 2013.

Int J Elec & Comp Eng, Vol. 8, No. 6, December 2018 : 4111 - 4119
Int J Elec & Comp Eng ISSN: 2088-8708  4119

[5] Y. O. Ouma and R. Tateishi, “A water index for rapid mapping of shore-line changes of five East African Rift
Valley lakes: an empiricalanalysis using Landsat TM and ETM+ data,” Int. J. Remote Sens, vol. 27, pp. 3153-3181,
2006.
[6] S. Lu, et al., “Water body mapping method with HJ-1A/B satellite imagery,” Int. J. Appl. Earth Obs. Geoinf., vol.
13, pp. 428-434, 2011.
[7] P. J. J. Desmet and G. Govers, “A GIS procedure for automatically calculating the USLE LS factor
on topographically complex landscape units,” J. Soil Water Conserv., vol. 51, pp. 427-433, 1996.
[8] W. Zhou and B. Wu, “Assessment of soil erosion and sediment delivery ratio using remote sensing and GIS: A case
study of upstream chaobaihe river catchment, north China,” Int. J. Sediment Res., vol. 23, pp. 167-173, 2008.
[9] Z. Du, et al., “Estimating surface water area changes using time-series Landsat data in the qingjiang river basin,
China,” J. Appl. Remote Sens., vol. 6, 2012.
[10] F. Sun, et al., “Comparison and improvement of methods for identifying waterbodies in remotely sensed imagery,”
Int. J. Remote Sens., vol. 33, pp. 6854-6875, 2012.
[11] Water Body Extraction from Multi-Source Satellite Images. Available
online: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.301.8033&rep=rep1&type=pdf.
[12] H. Xu, “Modification of normalised difference water index (NDWI) to enhance open water features in remotely
sensed imagery,” Int. J. Remote Sens., vol. 27, pp. 3025-3033, 2006.
[13] Water Body Extraction And Change Detection Based on Multi-Temporal SAR Images. Available online:
http://adsabs.harvard.edu/abs/2009SPIE.7498E..96Z.
[14] H. Zhou, et al., “Landscape and water quality change detection in urban wetland: A post-classification comparison
method with IKONOS data,” Procedia Environ. Sci., vol. 10, pp. 1726-1731, 2011.
[15] Z. Tang, et al., “Extraction of water body based on Landsat TM5 imagery-A case study in the Yangtze river,” Adv.
Inf. Comm. Technol., vol. 393, pp. 416-420, 2013.
[16] W. Li, et al., “A comparison of land surface water mapping using the normalized difference water index from
TM, ETM+ and ALI,” Remote Sens., vol. 5, pp. 5530-5549, 2013.
[17] S. K. McFeeters, “Using the normalized difference water index (NDWI) within a geographic information system to
detect swimming pools for mosquito abatement: a practical approach,” Remote Sens., vol. 5, pp. 3544-3561, 2013.
[18] J. W. Rouse, et al., “Monitoring Vegetation Systems in the Great Plains with ERTS (Earth Resources Technology
Satellite),” Proceedings of Third Earth Resources Technology Satellite Symposium, Greenbelt, ON, Canada, vol.
SP-351, pp. 309-317, 1973.
[19] E. H. Wilson and S. A. Sader, “Detection of forest harvest type using multiple dates of Landsat TM imagery,”
Remote Sens. Environ., vol. 80, pp. 385-396, 2002.
[20] S. K. McFeeters, “The use of the normalized difference water index (NDWI) in the delineation of open water
features,” Int. J. Remote Sens., vol. 17, pp. 1425-1432, 1996.

Extraction of Water-body Area from High-resolution Landsat Imagery (B. Chandrababu Naik)