Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
8
, No
.
6
,
Decem
ber
201
8
, p
p.
4111
~
4119
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
4111
-
41
19
4111
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Extracti
on of W
ater
-
b
ody Area f
ro
m
Hig
h
-
r
esolut
ion
Landsat
Image
ry
B. C
h
andra
babu
Na
ik
,
B.
A
nuradh
a
Dep
ar
t
m
ent
o
f
E
le
c
troni
cs
and
C
om
m
unic
at
ion,
SV
U Col
le
ge
of
Engi
ne
eri
ng,
SV
Univer
sit
y
,
Ind
i
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
16
, 201
8
Re
vised
Ju
l
1
0
,
201
8
Accepte
d
Aug
2
, 2
01
8
Ext
ra
ct
ion
of
wa
te
r
bodie
s
from
sate
llit
e
imager
y
has
bee
n
broa
dl
y
expl
or
ed
in
the
cur
r
ent
de
ca
de
.
So
m
an
y
t
ec
hniqu
es
were
i
nvolve
d
in
de
te
c
ti
ng
of
th
e
surfac
e
wat
er
bo
die
s
from
sate
lli
te
data.
To
de
tect
and
ext
ra
ct
in
g
of
surfac
e
wate
r
bod
y
cha
n
ges
in
Nag
arj
un
a
Sagar
Reser
vo
ir,
Andhra
Prade
sh
from
the
per
iod
1989
to
2017,
were
calc
ula
t
ed
using
Lands
at
-
5
TM,
an
d
La
ndsat
-
8
OLI
dat
a
.
Uns
uper
vised
c
la
ss
ifica
t
ion
and
spe
ctral
wat
er
ind
exing
m
et
hods,
inc
ludi
ng
the
Norm
al
iz
ed
Diff
er
enc
e
Veg
et
a
ti
on
Inde
x
(ND
VI),
N
orm
al
iz
ed
Diffe
ren
c
e
Moi
sture
Inde
x
(N
DM
I),
Norm
al
ized
Diffe
r
enc
e
W
at
er
Inde
x
(ND
W
I),
and
M
odifi
ed
Norm
al
i
ze
d
Diffe
ren
ce
W
at
er
Inde
x
(MN
DW
I),
were
used
to
det
ec
t
a
nd
ext
racti
on
of
the
surfac
e
wat
er
bod
y
from
sate
llite
da
ta.
Instea
d
of
al
l
i
ndex
m
et
hods,
th
e
MN
DW
I
w
as
per
form
ed
bet
t
er
result
s
.
The
Reser
voir
wate
r
area
was
ext
racte
d
usin
g
spec
tra
l
wat
e
r
inde
xing
m
et
hods
(ND
VI
,
ND
W
I,
MN
DW
I,
and
ND
MI)
in
1989
,
1997
,
2007,
an
d
2017.
Th
e
shor
el
in
e
shrunk
in
the
twenty
-
ei
g
ht
-
y
ea
r
dura
t
ion
of
images.
The
Reser
vo
i
r
Naga
rjuna
Sag
ar
lost
nea
r
l
y
aro
u
nd
one
-
fourth
of
it
s
surfac
e
wate
r
are
a
compare
d
to
1989.
H
oweve
r,
the
R
ese
rvoir
has
a
cr
it
i
ca
l
posi
ti
o
n
in
rec
e
nt
y
e
ars
due
to
cha
ng
es
i
n
surfac
e
wat
er
and
get
t
ing
high
er
m
ud
and
sand.
Maximum
wate
r
surfac
e
area
of
the
Re
servoir
wil
l
lose
if
suc
h
dec
re
asing te
nde
nc
y
fol
lows c
ont
inuousl
y
.
Ke
yw
or
d:
Inde
x
m
et
ho
d
Lan
ds
at
MND
WI
ND
M
I
NDVI
NDWI
Su
r
face
water
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
B.
Cha
ndra
babu Nai
k
,
Dep
a
rtm
ent o
f El
ect
ro
nics
and C
omm
un
ic
ation
,
SVU Colle
ge o
f
E
ng
i
neer
i
ng, SV
U
niv
er
sit
y, I
nd
ia
.
Em
a
il
:
bab
unai
kb@
gm
ail.co
m
1.
INTROD
U
CTION
The
rem
ote
sensing
knowle
dge
is
m
ai
nly
us
e
d
in
di
ff
e
re
nt
area
s,
s
uc
h
as
the
la
ke
,
c
oastal
zo
ne
m
anag
em
ent,
sh
orel
ine
cha
ng
e
a
nd
er
os
i
on
m
on
it
or
i
ng,
forest
for
m
on
it
or
ing
of
changes,
f
orest
an
d
veg
et
at
io
n
c
ha
ng
e
s
[
1
]
,
[
2
]
,
disaste
r
m
on
it
or
i
ng
[
3
]
,
[
4
]
,
flo
od
pr
e
dicti
on
an
d
e
valuati
on
of
water
re
so
urces
[
5
]
.
It
is
essent
ia
l
fo
r
agr
ic
ultur
e
(
foo
d
cr
ops),
day
to
day
li
fe
of
hum
ans,
and
ec
os
yst
em
s
[6
]
.
To
acc
om
pl
ish
the
in
form
at
io
n
a
bout
open
su
r
face
water
is
m
os
t
i
m
po
rtant
in
dif
fer
e
nt
sci
entifi
c
are
as,
th
o
se
a
re
s
urface
water
a
naly
sis,
waters
he
d
a
na
ly
sis,
dynam
i
c
cha
ng
e
s
of
r
ivers,
en
vir
on
m
ent
m
on
it
or
i
ng,
pr
ese
nt
a
nd
f
uture
est
i
m
ation
s
of
water
resou
r
ces,
and
flo
od
m
app
ing
[
7
]
-
[
10]
.
Rem
ote
sensing
sat
el
li
te
s
are
hav
in
g
30
m
reso
l
utions
an
d
they
offer
a hug
e
am
ount
of d
at
a,
w
hich
is widely
us
e
d
f
or
detect
in
g
an
d
extracti
on
of
s
urface
water a
reas a
nd it
s
dynam
ic
c
hanges
in rece
nt d
eca
des
[11
]
-
[
17
]
.
Id
e
ntific
at
ion
of
water
is
ver
y
sig
nifica
nt
f
or
var
i
ous
pr
eci
se
est
i
m
at
ion
s
an
d
hu
m
an
li
fe.
To
detect
ing
a
nd
ext
r
act
ion
of
s
urface
water
area
from
sat
e
ll
it
e
data
has
be
en
int
rod
uced
at
m
any
m
or
e
i
m
age
processi
ng
te
chn
i
qu
e
s
in
the
cu
rr
e
nt
decade
s.
A
sin
gle
-
ba
nd
an
d
m
ulti
-
band
m
et
hods
wer
e
wi
dely
use
d
i
n
Lan
ds
at
i
m
ager
y
for
detect
in
g
an
d
extracti
on
of
surface
w
at
er
are
a
al
ong
with
sel
ect
ed
thres
hold
val
ue,
ei
ther
po
sit
ive
or
ne
gative
val
ue
[
9].
Com
par
ed
with
a
sing
le
-
band
m
et
ho
d,
m
ul
ti
-
band
m
e
thod
was
e
xtensiv
el
y
us
e
d
f
or
e
nh
a
nc
ing
the
s
urfac
e
water
bo
dies
[9
]
.
F
our
different
sat
el
li
te
m
ul
ti
-
band
m
e
thods
we
re
use
d
f
or
e
xtracti
on
of
su
r
face
water
bodies
,
tho
se
a
r
e
water
ind
e
xi
ng
m
et
ho
ds,
in
cl
ud
in
g
the
N
DVI,
N
DMI
,
NDWI,
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4111
-
41
19
4112
and
M
NDWI.
Th
us
,
these
m
et
hods
we
re
in
tro
du
ce
d
to
an
extracti
on
of
su
r
face
wate
r
bodies
with
s
pe
ci
fie
d
thres
ho
l
d
valu
es
bein
g
ei
the
r
le
ss
than
zer
o
(n
e
gati
ve
valu
e)
or
gr
eat
er
t
ha
n
zer
o
(positi
ve
value).
T
he
NDVI,
ND
M
I,
N
D
WI,
an
d
MN
D
WI
f
or
the
detect
ing
an
d
extra
ct
ion
of
s
urface
water
bodies
f
r
om
sat
el
l
it
e
i
m
ager
y.
By
u
sin
g
c
hange
detect
ion,
w
at
er f
eat
ur
es
ca
n
e
xtract se
parat
el
y fo
r
t
he di
ff
e
ren
t y
ear
s
of sa
te
ll
it
e d
at
a.
The
Na
garj
un
a
Saga
r
Re
se
r
voir
is
locat
e
d
in
Nalg
onda
distr
ic
t
in
Tel
an
ga
na
sta
te
.
I
n
rece
nt
de
cades
,
there
a
re
trem
e
ndous
c
hanges
in
it
s
capaci
ty
.
The
refore
,
th
e
dynam
ic
chan
ge
s
of
the
Re
servoir
s
urface
wate
r
area
nee
d
to
m
on
it
or
c
on
ti
nu
ou
sly
.
In
t
his
s
tud
y,
the
sp
at
i
otem
po
ral
cha
ng
e
s
of
Re
ser
voir
Nag
a
rju
na
Saga
r
from
19
89
to
2017
are
i
nv
es
ti
gated
base
d
on
sat
el
li
te
-
m
ulti
ban
d
water
i
nd
e
xe
d
m
et
ho
ds
,
in
cl
udin
g
NDVI,
ND
M
I,
NDWI
,
an
d
M
NDWI
us
in
g
Lan
ds
at
-
5
TM
an
d
La
nd
s
at
-
8
OL
I
i
m
ages
f
or
e
xtr
act
ion
of
the
s
urface
water
body. O
ver
al
l, the M
N
D
WI
w
e
re fo
und
s
uperi
or
t
o othe
r
in
dex
es
. T
hese m
et
ho
d i
s h
ig
hly im
po
r
ta
nt f
or
tim
e
-
series
ana
ly
ses
of
e
xtract
ing
s
horeli
ne
s
us
in
g
a
ny
num
ber
of
La
ndsat
i
m
ages
in
dif
f
eren
t
ti
m
e
inter
vals
,
and it
provides
an
im
po
r
ta
nt c
on
t
rast that ca
n be
us
e
d
t
o
in
ve
sti
gate shoreli
ne
c
hanges
.
2.
STUDY
A
RE
A AND
D
ATA
SET
S
The
Nag
a
r
j
un
a
Saga
r
Re
s
ervoir
was
bu
il
t
acro
ss
th
e
K
rishna
ri
ve
r
,
wh
ic
h
is
locat
ed
i
n
betwee
n
Guntur
Distric
t
a
nd
Nalg
onda
Distric
t
in
Tel
an
ga
na
an
d
it
s
ge
og
raphical
area
s
urrou
nd
e
d
flan
ked
by
la
ti
tud
e
16°
34
'
55.60"
N
to
16°
56
'
44.
95
"
N
and
lo
ng
it
ude
78°
24
'
13.
97
"
E
to
78°47'
06.07"E
as
s
ho
wn
i
n
F
ig
ure
1.
W
at
er
spread
a
rea
is
28
5
km
2
at
Fu
ll
Re
servoir
Le
v
el
(F
RL
),
T
he
cat
chm
ent
Ar
ea
is
214,1
85
km
2
(
82,69
7
sq
m
i)
and
t
he
gr
os
s
s
t
or
a
ge
capaci
ty
405
TMC
at
FRL,
I
n
I
ndia
,
the
Na
garj
un
a
Saga
r
Re
servoir
is
th
e
seco
nd
bigge
st
water
re
ser
voir
a
nd
it
’s
al
so
one
of
t
he
m
os
t
basic
m
ulti
-
pu
r
pose
ir
rigati
on
and h
y
dro
-
el
ec
tric
p
r
oject
s in
India.
Fig
ure
1. Locat
ion
of
nag
a
rju
na
sag
ar
reserv
oi
r
The
Lan
ds
at
-
5
TM
and
La
nd
s
at
-
8
OL
I
data
ob
ta
ine
d
in
Ja
nuary
19
89,
1997,
20
07,
an
d
2017.
T
hese
data
wer
e
ta
ke
n
from
the
US
Ge
ologica
l
Su
r
vey
(USG
S)
portal
,
but
colle
ct
ed
al
l
Landsat
i
m
ages
were
cl
oud
-
fr
ee
data
.
Ta
ble
1
repre
sents
t
he
s
peci
fic
at
ion
of
La
ndsat
-
5
TM
an
d
Lan
ds
at
-
8
OL
I
im
ages.
T
he
water
body
was
e
xtr
act
ed
f
r
om
satell
it
e
i
m
ages
with
a
30m
sp
at
ia
l
reso
luti
on
an
d
dif
fer
e
nt
b
an
d
with
dif
f
eren
t
wav
el
e
ng
t
hs
as
shown i
n
Ta
bl
e 1
.
Table
1.
L
an
dsa
t
-
5
TM
and L
andsat
-
8 O
LI d
at
a w
it
h
it
s S
pe
ci
ficat
ion
s
Satellite
Year
Path
/Ro
w
Res
o
lu
tio
n
(
m
)
W
av
elen
g
th
s (μ
m
)
Land
sat
-
5
TM
1989
1997
2007
1
4
3
/4
9
30
Ban
d
1
:0
.4
5
-
0
.52
Ban
d
2
:0
.5
2
-
0
.60
Ban
d
3
:0
.6
3
-
0
.69
Ban
d
4
:0
.7
6
-
0
.90
Ban
d
5
:1
.5
5
-
1
.75
Ban
d
7
:2
.0
8
-
2
.35
Ban
d
1
:0
.4
3
3
-
0
.453
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Ext
ra
ct
io
n
of
Water
-
body
Ar
ea
fr
om H
i
gh
-
r
esolutio
n La
nd
sa
t I
m
ag
er
y (
B
. Cha
nd
r
abab
u N
aik)
4113
Table
1.
L
an
dsa
t
-
5
TM
and L
andsat
-
8 O
LI d
at
a w
it
h
it
s S
pe
ci
ficat
ion
s
Satellite
Year
Path
/Ro
w
Res
o
lu
tio
n
(
m
)
W
av
elen
g
th
s (μ
m
)
Land
sat
-
8
OLI
2017
1
4
3
/4
9
30
Ban
d
2
:0
.4
5
0
-
0
.515
Ban
d
3
:0
.5
2
5
-
0
.600
Ban
d
4
:0
.6
3
0
-
0
.680
Ban
d
5
:0
.8
4
5
-
0
.885
Ban
d
6
:1
.5
6
0
-
1
.660
Ban
d
7
:2
.1
0
0
-
2
.300
Ban
d
9
:1
.3
6
0
-
1
.390
3.
METHO
D
OL
OGY
Con
si
der
e
d
t
he
pe
rio
d
of
1989
t
o
2017,
t
hat
can
s
how
the
cha
nges
of
wate
r
a
rea
i
n
Re
se
rvoir.
In
tr
oduce
s
the
m
et
ho
do
l
og
y
and
it
s
perfor
m
ances
of
dif
f
eren
t
sat
el
li
te
–
m
ulti
ban
d
in
dex
e
s,
inclu
di
ng
the
Norm
al
iz
ed
Di
ff
e
ren
ce
Ve
get
at
ion
Inde
x
(NDVI)
[
18
]
,
Nor
m
al
iz
ed
Diff
er
ence
Moist
ur
e
Inde
x
(NDMI
)
[1
9],
Norm
al
iz
ed
Diff
ere
nce
Wat
er
I
nde
x
(
N
D
WI)
[
20
]
,
a
n
d
Mo
dified
N
orm
alized
Dif
f
eren
ce
W
at
e
r
I
nd
e
x
(MN
D
WI
)
[
12]
,
wer
e
us
e
d
f
or
detect
ing
a
nd
e
xtracti
on
of
s
urface
wat
er
bo
dies
f
ro
m
Lan
ds
at
im
ag
ery
as
sh
ow
n
in
Tabl
e
2
.
T
hus,
f
our
diff
e
re
nt
ye
ar
s
of
sat
el
li
te
i
m
ages
(La
ndsa
t
-
5
data
from
1989,
1997,
20
07
a
nd
Lan
ds
at
-
8
data
from
20
17)
w
ere
pe
rfor
m
ed
and
e
xtracti
on
of
s
urface
water
bo
dies
us
i
ng
di
ff
e
ren
t
in
de
xing
m
et
ho
ds
(ND
VI
,
N
DMI,
N
D
WI
,
a
nd
M
N
D
WI
)
.
Table
2.
Re
prese
nt
the
Lan
ds
at
-
5
data,
in
dex
es
us
e
d
for
wate
r
feat
ur
e
extracti
on
(B
2:
band2
=
gr
e
en
,
B3:
ba
nd3=
r
ed,
B
4:
band
4=
near
in
frare
d,
B5:
ba
nd5=
m
idd
le
infr
a
re
d.
Sim
i
la
rl
y,
in
Land
sat
-
8
data
(
B3:
band3=g
r
een,
B4:
ba
nd4=
re
d,
B5:
band
5=
nea
r
inf
rar
e
d,
B6:
ba
nd
6=
m
i
dd
le
i
nfrar
e
d).
Table
2.
Sate
ll
i
te
-
m
ult
iband
I
nd
e
xes
u
se
d f
or
Water Feat
ure Extracti
on
Ind
ex
Equ
atio
n
Re
m
ark
Ref
erence
No
r
m
aliz
ed
Dif
f
erence Vegetatio
n
I
n
d
ex
ND
VI=
(B4
-
B3
)
/
(B
4
+
B3
)
W
ate
r
wo
u
ld
be a
n
eg
ativ
e valu
e
[
2
8
]
No
r
m
aliz
ed
Dif
f
erence M
o
istu
re
Ind
ex
NDM
I=(
B4
-
B5
)/
(
B4
+
B5
)
W
ate
r
wo
u
ld
be a
p
o
sitiv
e valu
e
[
1
9
]
No
r
m
aliz
ed
Dif
f
erence Water
Ind
ex
NDWI=(B2
-
B4
)/
(
B2
+
B4
)
W
ate
r
wo
u
ld
be a
p
o
sitiv
e valu
e
[
2
0
]
Mod
if
ied
Nor
m
ali
zed Dif
f
erence Wa
ter
Ind
ex
MND
W
I
=(B2
-
B5
)
/(B2
+
B5
)
W
ate
r
wo
u
ld
be a
p
o
sitiv
e valu
e
[
1
2
]
The
N
D
WI
ha
s
introduc
ed
to
detect
and
extracti
ng
the
su
r
face
water
bodies
with
a
sp
eci
fied
thres
ho
l
d
val
ue
.
The
po
sit
iv
e
thres
ho
l
d
va
lues
f
or
water
an
d
ne
gative
thre
shold
values
for
no
nw
at
e
r
bodies
[
20
]
.
T
he
M
NDWI
ha
s
intr
oduce
d
a
powe
rful
in
de
x
to
detect
a
nd
extracti
ng
th
e
su
r
face
water
bodies
.
Be
cause
ba
nd5
(m
idd
le
inf
rare
d)
has
re
place
d
by
the
band
4
(n
ea
r
inf
rar
e
d).
He
nce
band
5
re
flect
ance’s
m
or
e
com
par
ed
with
band
4
[
12]
.
T
he
MN
D
WI
w
idely
us
e
d
f
or
s
uppressi
ng
er
r
or
s f
r
om
veg
et
at
ion
,
s
oils,
a
nd
buil
t
-
up
areas
.
T
he
thres
ho
l
d
valu
es
of
MN
D
WI
ha
ve
posit
ive
s
an
d
ne
gative
s
for
water
a
nd
no
nwat
er
bodies.
The
N
DVI
ha
s
intr
oduce
d
m
ai
nly
us
ed
f
or
extr
act
ing
gr
ee
n
ve
getat
ion
f
ro
m
oth
e
r
wetl
an
d
s
urfa
ce
areas
.
Th
us
,
N
DVI
al
so
extracts
the su
r
face
water m
uch
bette
r
t
han
ND
M
I
ind
e
xed
m
et
ho
d,
a
nd
it
s
thres
ho
l
d
values
for
wa
te
r
w
ould
be
ne
gative
values
[
18
]
.
The
N
DMI
has
introd
uced
m
a
inly
fo
r
extrac
ti
ng
ve
getat
ion,
an
d
water
li
quid
but
it
’s
not
m
uch
m
or
e
capa
ble
of
ext
racti
ng
w
at
er
bo
dies
as
com
par
ed
t
o
ot
her
s
in
de
x
m
eth
ods
(NDVI,
N
D
WI,
an
d
MN
D
WI).
T
hus,
the
N
DMI
m
et
ho
d
was
no
t
e
ff
ic
ie
nt
f
or
e
xtracti
on
of
water
bodi
es.
Th
e
ND
M
I
thre
sho
ld
value
of
w
at
er
woul
d
be
a
po
sit
ive.
B
ased
on
the
se
analy
ses,
the
MND
WI
m
et
ho
d
has
perform
ed
sli
gh
tl
y bett
er th
a
n othe
r
i
nd
e
x
m
et
hods
(ND
WI, ND
VI
,
and
N
DMI).
The
best
sig
ni
ficance
of
w
at
er
body
e
xtracti
on
te
c
hn
i
qu
e
s
was
rec
ognized
an
d
e
m
plo
ye
d
to
sp
at
iotem
po
ral
changes
of
the
Na
garj
un
a
Sagar
Re
ser
voir
in
the
pe
r
iod
19
89
to
2017.
T
o
detec
t
and
extracti
ng
Re
s
ervoir
surface
water
bodies
in
fo
ur
diff
e
re
nt
ye
ars,
su
c
h
a
s
La
nd
sat
-
5
(
1989,
1997,
20
07)
a
nd
Lan
ds
at
-
8
(20
17)
im
ages.
Ou
t
of
fou
r
-
ye
ar
a
naly
ses
of
ext
r
act
ing
Re
ser
voir
su
r
face
wate
r
area,
the
m
ax
i
m
u
m
change
s
occ
ur
in
the
pe
rio
d
of
2007
-
2017.
I
n
order
to
get
t
he
e
ff
ic
ie
ncy
of
detect
ion
an
d
extracti
on
of
s
urface
water
a
rea,
dif
f
eren
t
acc
uracy
an
al
yse
s
we
re p
er
form
ed.
By
u
sin
g
acc
uracy
assessm
ent
an
al
yse
s,
cal
culat
e
the
par
am
et
ers
are
ov
e
rall
accuracy,
produc
er’
s
acc
ur
acy
,
us
er
’s
accu
r
acy
,
and
kappa
coe
ff
ic
ie
nt.
Tho
s
e
par
am
et
ers
were pe
rfor
m
ed
ov
er th
e
ch
a
nges
of w
at
er
bo
dy in
the
p
e
rio
d 1
989 t
o 2
017.
4.
RESU
LT
S
A
ND AN
ALYSIS
On
ce
t
he
re
quired
sat
el
li
te
da
ta
ob
ta
ine
d,
T
he
im
age
pr
oc
essing
te
c
hn
i
ques
ha
ve
bee
n
involve
d
f
or
furthe
r
proces
sing.
T
he
a
na
ly
ses
wer
e
pe
rfor
m
ed
be
f
ore
ap
plyi
ng
wa
t
er
in
dex
e
d
m
et
hods
a
s
s
ho
wn
in
F
igure 2
(a
-
d)
.
A
nd
a
fter
a
pp
l
yi
ng
w
at
er
in
de
xed
m
et
ho
ds
, o
b
ta
ine
d
t
he
re
su
lt
s
as
s
how
n
in
F
ig
ur
e
s
3,4
,
5,
a
nd
Figure
6
(a
-
d).
Dif
fer
e
nt
sat
e
ll
it
e
-
m
ult
iband
water
i
nd
e
xe
d
m
et
ho
ds
,
inc
lud
in
g
N
DVI,
ND
M
I,
NDWI
,
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4111
-
41
19
4114
MND
WI
wer
e
us
e
d
to
de
te
ct
and
e
xtract
the
su
r
face
wate
r
body
f
ro
m
the
Lan
ds
at
-
5
(
1989,
1997,
2007)
data
and La
nd
sat
-
8
(20
17) data
.
The
f
our
s
pect
ral
water
in
de
xed
m
et
ho
ds
(
NDVI,
N
DMI
,
NDWI,
an
d
MND
WI)
are
app
li
ed
t
o
the
Re
servoir
water
area
to
hi
gh
li
ght
the
di
ff
e
ren
ce
s
bet
ween
water
a
nd
non
-
w
at
er
bodies
as
s
how
n
in
F
ig
ure
s
3,4,5,
and
6
(a
-
d)
.
Ou
t
of
al
l
wa
te
r
ind
e
xe
d
m
et
hods
,
t
he
MND
WI
has
t
he
bette
r
m
et
h
od
f
or
separ
at
in
g
wat
er
bodies
as
c
om
par
ed
t
o
oth
er
in
dex
e
d
m
et
hods
(ND
VI,
N
D
WI,
a
nd
ND
M
I)
.
G
ene
rall
y,
thres
ho
l
d
va
lu
es
of
wate
r
are
as
ha
ving
gr
ea
te
r
than
zer
o
va
lues
an
d
ve
ge
ta
ti
on
areas
ha
ving
neg
at
ive
values
.
First,
cal
culat
e
su
rf
a
ce
water
area
and
c
ha
ng
e
d
s
urface
water
area
f
or
fo
ur
di
ff
e
ren
t
ye
ars
with
sel
ect
ed
Re
servoir
as
s
how
n
in
Table
4
.
Sim
i
la
rly
,
cal
culat
e
the
s
urface
per
im
et
er
and
cha
nged
s
urface
per
im
eter
f
or
four d
i
ff
e
ren
t y
ears
with sam
e Rese
r
vo
ir
ass
hwo
n
in
Ta
ble
5
.
Fig
ure
2. Lan
dsa
t
-
5 TM
im
ag
e from
1
989 (a
); Landsat
-
5 T
M im
age f
r
om
1
99
7 (b); La
ndsat
-
5 TM
i
m
age
from
2
007 (c
); La
nd
sat
-
8 OL
I
im
age f
ro
m
2
01
7
(
d)
Fig
ure
3. N
DVI
-
base
d
e
xtract
s w
at
er
bo
dy fr
om
the Landsa
t
-
5
TM
im
age 1
98
9,199
7,
a
nd
2007
(a
-
c
);
Lan
ds
at
-
8 OLI im
age 2017 (
d)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Ext
ra
ct
io
n
of
Water
-
body
Ar
ea
fr
om H
i
gh
-
r
esolutio
n La
nd
sa
t I
m
ag
er
y (
B
. Cha
nd
r
abab
u N
aik)
4115
Fig
ure
4. N
D
MI
-
ba
sed
ex
t
ra
ct
s w
at
er
body
from
the Lands
at
-
5
TM
im
age
1989,1
997, an
d
2007
(a
-
c
);
Lan
ds
at
-
8 OLI
i
m
age 2017 (
d)
Fig
ure
5. N
D
WI
-
ba
sed
ex
tra
ct
s w
at
er
body
from
the Lands
at
-
5
TM
im
age
1989,1
997, an
d
2007
(a
-
c
);
Lan
ds
at
-
8 OLI im
age 2017 (
d)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4111
-
41
19
4116
Fig
ure
6. MN
D
WI
-
base
d
e
xt
racts wate
r bod
y from
the Landsat
-
5
TM i
m
a
ge 198
9,199
7,
and 20
07 (
a
-
c);
Lan
ds
at
-
8
OLI i
m
age 2017 (
d)
Fig
ure
7. Cha
nges i
n
the
area
of Rese
r
vo
i
r N
agarj
un
a
Sa
gar ge
ner
at
e
d usin
g (a)
M
NDWI
1989; (
b) MN
D
WI
and ND
VI
1989 a
nd 19
97
;
(c) M
NDWI,
N
D
VI
an
d NDWI
1989, 1
997,
a
nd 20
07
;
(
d)
M
NDWI,
NDWI
and
NDVI 1
989,
1997,
20
07 and
2017.
4.1.
Eva
lu
at
i
on
of
th
e c
ha
n
ge
s
The
Re
ser
voir
water
area
was
ext
racted
us
in
g
U
nsupe
rv
ise
d
cl
assif
ic
at
ion
an
d
s
pectral
wate
r
ind
e
xing
m
et
h
od
s
i
n
1989,
1997,
2007,
a
nd
2017.
The
s
ho
reli
ne
shru
nk
in
the
twe
nty
-
ei
gh
t
-
ye
ar
durati
on
of
i
m
ages.
H
ow
e
ver, the Rese
r
voir
has bee
n
in
a crit
ic
al
sit
u
at
ion
i
n
rece
nt y
ears
du
e
to
c
ha
ng
e
s in
s
urface
w
at
er
and
getti
ng
higher
m
ud
an
d
s
and.
Acc
ordin
g
to
the
N
D
VI,
ND
M
I,
N
D
WI,
an
d
MND
WI,
betwee
n
1989
a
nd
2007
there
we
r
e
li
ttle
bit
changes
in
Re
se
rvo
ir
water
area
,
a
nd
t
he
cha
nges
betwee
n
20
07
and
2017
the
r
e
wa
s
m
axi
m
u
m
chan
ge
s
in
Re
ser
voir
wate
r
area
as
show
n
in
F
ig
ure
7.
T
he
N
DVI,
ND
M
I,
NDWI,
an
d
M
N
D
WI
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Ext
ra
ct
io
n
of
Water
-
body
Ar
ea
fr
om H
i
gh
-
r
esolutio
n La
nd
sa
t I
m
ag
er
y (
B
. Cha
nd
r
abab
u N
aik)
4117
resu
lt
s
e
xpos
e
that
the
surfac
e
area
of
Re
se
rvoir
Nag
a
r
j
una
Saga
r
in
Ja
nua
ry
1989,
1997,
2007,
a
nd
20
17
wa
s
app
r
ox
im
at
ely
205
km
2
,
203
km
2
,
196
km
2
,
an
d
160
km
2
as
s
how
n
i
n
Table
3
.
T
he
res
ult
sho
w
t
hat
the
Re
servoir
surf
ace
water
body
cha
ng
es
ar
ou
nd
2.1
31
km
2
betwee
n
1989
to
1997,
6.4
94
km
2
between
1997
t
o
2007,
36.01
6
km
2
between
2007
t
o
2017.
T
he
ov
e
rall
cha
nges
in
surface
water
a
rou
nd
44.
641
km
2
bet
wee
n
1989
to
20
17
a
s sho
wn in T
ab
le
4
.
Table
3.
Per
for
m
ance
Eval
uation
o
f
T
he
Sate
ll
it
e
-
m
ult
iband Ind
e
xes
u
se
d
f
or Surfa
ce
Water
E
xtracti
on
Ind
ex
es
Year
Min
Thresh
o
ld
Max
Thresh
o
ld
Mean
Stan
d
ard
Dev
iatio
n
Peri
m
e
ter
(
k
m
)
Area
(k
m
2
)
NDVI
1989
-
0
.55
5
0
.71
6
0
.03
4
0
.13
2
2
5
8
.095
2
0
5
.574
1997
-
0
.29
0
0
.61
3
0
.06
8
0
.12
8
2
5
7
.113
2
0
3
.443
2007
-
0
.37
5
0
.67
2
0
.04
0
0
.15
5
2
5
2
.296
1
9
6
.949
2017
-
0
.58
2
0
.41
6
0
.11
9
0
.08
3
2
1
2
.021
1
6
1
.529
NDM
I
1989
0
.31
1
0
.74
5
0
.18
6
0
.14
5
2
6
4
.112
2
0
1
.555
1997
0
.30
4
0
.77
4
0
.28
2
0
.15
2
2
6
0
.436
2
0
1
.101
2007
0
.33
4
0
.77
8
0
.31
8
0
.18
2
2
5
4
.346
1
9
2
.115
2017
0
.32
3
0
.78
2
0
.43
3
0
.17
3
2
1
6
.170
1
6
0
.933
NDWI
1989
0
.60
7
0
.92
4
0
.44
6
0
.18
8
2
6
1
.423
2
0
4
.229
1997
0
.63
5
0
.91
8
0
.50
5
0
.18
7
2
5
9
.029
2
0
1
.630
2007
0
.61
3
0
.92
3
0
.47
5
0
.20
7
2
5
2
.034
1
9
5
.679
2017
0
.62
3
0
.98
7
0
.54
4
0
.19
8
2
1
1
.232
1
6
1
.411
MND
W
I
1989
0
.60
3
0
.93
6
0
.21
4
0
.25
6
2
6
3
.827
2
0
3
.784
1997
0
.60
1
0
.95
0
0
.25
0
0
.23
6
2
5
8
.978
2
0
1
.879
2007
0
.61
0
0
.94
8
0
.29
9
0
.26
9
2
5
2
.152
1
9
4
.783
2017
0
.61
7
0
.98
8
0
.40
7
0
.26
4
2
1
3
.627
1
6
1
.335
Table
4.
Stat
ist
ic
s
of the
Re
se
rvoir
Surface
Ar
ea
Cha
ng
e
s
Year
Su
rf
ace
wate
r
ar
ea
(
k
m
2
)
Su
rf
ace
wate
r
ar
ea
chan
g
e (
k
m
2
)
Total su
rf
ace
wat
e
r
are
a chan
g
es (k
m
2
)
1989
2
0
5
.574
-
2
.13
1
-
4
4
.64
1
1997
2
0
3
.443
-
6
.49
4
2007
1
9
6
.949
-
3
6
.01
6
2017
1
6
0
.933
Si
m
il
arly
,
The
NDVI,
N
DM
I,
N
D
WI,
an
d
MND
WI
res
ul
ts
expose
that
the
surface
pe
rim
e
te
r
of
Re
servoir
Na
ga
rju
na
Saga
r
in
Janu
a
ry
1989,
1997,
2007,
a
nd
2017
was
a
ppr
ox
im
at
ely
26
4
km
,
260
km
,
254
km
,
and
216
km
as
sh
own
i
n
Table
3
.
T
he
resu
lt
s
how
th
at
the
Re
ser
voir
surface
pe
rim
et
er
changes
arou
nd
3.676
km
between
1989
to
1997,
6.0
90
km
between
1997
to
2007,
38.17
6
km
between
20
07
t
o
2017.
The
ov
e
rall
ch
ang
e
s
in
s
urfa
ce
per
im
et
er
arou
nd
47.
942
km
between
1989
t
o
20
17
a
s
show
n
i
n
Ta
ble
5
.
The
re
su
lt
sho
w
that
de
creas
ing
s
urface
w
at
er
area
in
R
eservoir
from
the
pe
rio
d
19
89
t
o
20
17,
a
nd
by
ob
s
er
ving the result
ing
a
naly
ses o
f
m
axi
m
u
m
ch
ang
es in t
he
surface
water bod
y, as w
el
l as sur
face p
e
r
i
m
et
er,
occurre
d
in
th
e
per
io
d
2007
to
2017.
T
he
Re
ser
voir
Na
garj
un
a
Sa
ga
r
lost
near
ly
ar
ound
one
-
f
ourth
of
it
s
su
r
face
area
c
om
par
ed
to
19
89.
T
he
m
axi
m
um
su
rf
ace
wate
r
area
c
hanges
are
obse
rv
e
d
i
n
the
wester
n
r
egio
n
and w
e
ste
rn
-
nort
h parts
of th
e
Re
ser
vo
ir
.
Table
5.
Stat
ist
ic
s o
f
the Rese
rvoir
Surface
P
erim
e
te
r
Chan
ges
Year
Su
rf
ace
p
eri
m
e
ter
(k
m
)
Su
rf
ace
p
eri
m
e
ter
ch
an
g
e (
k
m
)
Total su
rf
ace
p
eri
m
e
te
r
ch
an
g
e (
k
m
)
1989
2
6
4
.112
-
3
.67
6
-
4
7
.94
2
1997
2
6
0
.436
-
6
.09
0
2007
2
5
4
.346
-
3
8
.17
6
2017
2
1
6
.170
4.2.
Accur
acy asse
ssment
anal
yse
s
The
pr
ob
a
bili
ty
(
%)
that t
he c
la
ssifie
r
has
l
abeled a
n
im
ag
e p
ixel i
nto
t
he
g
r
ound tr
uth
Cl
ass. I
t i
s the
pro
bab
il
it
y
of
a
ref
ere
nce
pixe
l
being
c
orrec
tl
y
cl
assifi
ed.
Pr
od
ucer’s
acc
ur
acy
:
It
re
pr
es
ent
pix
el
s
that
belo
ng
to
the
tr
uth
cl
a
ss
but
fail
to
be
cl
assifi
ed
int
o
the
pro
per
cl
ass
(
om
issi
on
error
=
num
ber
of
c
orrectl
y
cl
assifi
ed
pix
el
s
in
each
cat
egory
pe
r
to
ta
l
nu
m
ber
of
c
la
ssifie
d
pi
xels
in
that
cat
eg
or
y
(co
lum
n
total
))
,
Use
r’
s
accu
racy:
Re
li
abili
t
y,
prob
a
bili
ty
a
pix
el
cl
ass
on
t
he
m
ap
rep
r
e
sents
th
e
cat
e
gory
on
the
gro
und
(c
omm
issi
on
error
=
num
ber
of
c
orrectl
y
cl
assifi
ed
pix
el
s
in
each
cat
e
gory
per
total
nu
m
ber
of
cl
a
ssifie
d
pix
el
s
in
that
cat
egory
(row
total
))
,
O
ver
a
ll
accuracy:
T
he
overall
accuracy=
T
otal
nu
m
ber
of
co
rrec
tl
y
c
la
ssifie
d
pix
e
l
s
(d
ia
g
onal
)
pe
r
total
num
ber
of
r
efe
ren
ce
pi
xe
ls,
Ka
ppa
c
oe
ff
ic
ie
nt
(K
hat
):
It
m
easur
e
of
agr
eem
ent
bet
ween
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4111
-
41
19
4118
the
cl
assifi
cat
ion
m
ap
an
d
th
e
ref
e
ren
ce
dat
a.
a
discrete
m
ulti
var
ia
te
te
ch
nique
of
us
e
in
accuracy
asses
sm
ent.
K
hat
>
0.8
0
r
ep
re
sent str
ong ag
r
ee
m
ent and
g
ood ac
cu
racy.
K
hat
=0.
40
-
0.8
0
is
m
idd
le
,
K
hat
<
0.4
0
is
poor.
Kap
pa
C
oeffi
cie
nt
=
(
TS
X
TC
S
)
−
∑
(
C
O
L
.
TOTA
L
X
R
O
W
.
TOTA
L
)
TS
2
−
∑
(
C
O
L
.
TOTAL
X
R
O
W
.
TOTA
L
)
X100
(1)
Wh
e
re,
TS=T
ot
al
n
um
ber
of
Sam
ples, TCS=
Total
num
ber o
f
Co
rr
ect
Sa
m
ples.
Stat
ist
ic
al
par
a
m
et
ers
of
the
a
ccur
acy
assess
m
ent
resu
lt
show
that
MN
D
WI
ac
hieve
d
1.5
51
km
2
has
the
abso
l
ute
error,
96.06%
has
us
e
r’
s
acc
ur
acy
,
93.
18%
has
produce
r
’s
accuracy
,
96.
43%
has
ov
erall
accuracy,
a
n
d
0.890
6
is
ka
pp
a
c
oe
ff
ic
ie
nt
s
as
sh
ow
n
in
Table
6
.
T
houg
h,
the
i
nd
e
x
m
et
ho
ds
,
in
cl
ud
i
ng
MND
WI,
NDWI,
and
NDV
I
pro
vid
e
d
good
acc
ur
aci
es
as
com
par
ed
with
ND
M
I
m
et
hod.
I
ns
te
ad
of
al
l
ind
e
xed
m
et
ho
ds,
M
NDWI
pro
vid
es
be
tt
er
re
su
lt
s
t
han
oth
e
r
m
et
ho
ds
(
NDWI,
N
D
VI
,
a
nd
N
DM
I)
f
or
detect
ing
an
d
extracti
ng
s
urf
ace
water
bo
dy
in
the
Re
se
r
vo
i
r.
Her
e
,
the
Re
ser
vo
ir
pr
ovides
a
lot
of
ben
e
fits
for
a
gr
ic
ultur
a
l
(fo
od
cr
ops),
in
du
st
rial
,
do
m
est
ic
pu
r
pos
e,
a
nd
hum
an
existi
ng
in
it
s
su
r
rou
nd
i
ng.
T
hu
s
,
it
require
d
c
orre
ct
ing
m
easur
e
m
ents
to
a
vo
i
d
any
obsta
cl
es
in
t
he
Re
se
rvo
ir
t
o
get
it
s
ori
gin
al
c
onditi
on
s
f
or
resto
rin
g
the
Reserv
oir
.
Table
6.
Stat
ist
ic
al
Param
et
ers
of Accu
racy A
ssessm
ent o
f
C
hanges
in Su
rfac
e
W
at
er
Bo
dy
Metho
d
Ch
an
g
ed
water
b
o
d
y
(
k
m
2
)
Ab
so
lu
te
Er
ror
(
k
m
2
)
Prod
u
cer's
accurac
y
(
%
)
User's
accurac
y
(
%
)
Ov
erall
accurac
y
(
%
)
Kap
p
a
co
ef
f
icien
t
Ref
erence
4
4
.00
0
0
.00
0
100
100
100
1
NDVI
4
4
.04
5
0
.04
5
8
9
.57
9
1
.50
9
2
.86
0
.80
9
6
NDWI
4
2
.81
8
1
.18
2
9
2
.31
8
8
.91
9
4
.64
0
.81
0
0
MND
W
I
4
2
.44
9
1
.55
1
9
3
.18
9
6
.06
9
6
.43
0
.89
0
6
NDM
I
4
0
.62
2
3
.37
8
8
7
.59
8
8
.96
9
1
.67
0
.76
2
5
5.
CONCL
US
I
O
N
The
m
ai
n
aim
of
t
his
stu
dy
is
to
detect
the
sp
at
iotem
po
ral
c
hanges
of
s
urface
wa
te
r
area
i
n
Nag
a
r
j
una
Sa
ga
r
Re
ser
vo
i
r
f
r
om
the
per
io
d
of
1989
to
20
17,
by
th
e
sat
el
li
te
i
m
age
interp
retat
ion
a
nd
GIS.
Using
sat
el
li
te
i
m
ages
to
extract
inform
at
i
on
reg
a
r
ding
Re
servoir
water
area
cha
ng
e
is
faster
an
d
m
or
e
accurate
tha
n
ot
her
obser
vatio
n
m
e
tho
ds.
Se
ver
al
in
dex
m
e
thods
(
NDVI,
NDWI,
MN
D
WI,
and
ND
M
I)
we
re
us
e
d
f
or
detec
ti
ng
a
nd
e
xtra
ct
ing
s
urface
water
a
rea.
T
he
res
ult
s
hows
that
decr
easi
ng
s
urface
wate
r
ar
ea
m
axi
m
u
m
chan
ge
s
occurre
d
about
44.64
1
km
2
in
the
per
i
od
1989
t
o
2017
.
Es
pecial
ly
,
from
20
07
to
20
17
the
Re
servoir
l
os
t
it
s
su
r
face
water
area
ab
out
36.
016
km
2
.
If
s
uch
dec
reasin
g
te
nd
e
ncy
fo
ll
ows
c
on
ti
nuously
th
e
Re
servoir
will
lose
it
s
m
axi
mu
m
su
rf
a
ce
wa
te
r
area
in
near
fu
tu
re.
The
st
at
ist
ic
al
par
am
et
ers
of
the
ac
cur
acy
assessm
ent
res
ult
sho
w
th
at
MND
WI
pr
ov
i
des
bette
r
res
ul
ts
as
com
par
e
d
t
o
oth
e
r
in
de
x
m
et
ho
ds
(
NDVI,
NDWI,
an
d
N
DMI)
s
o
that
the
MND
WI
m
et
hod
has
go
od
eff
ic
ie
nt
in
de
te
ct
ing
and
e
xtrac
ti
ng
s
urface
water
body
in
Na
ga
r
j
una
Saga
r
Re
servoir
.
T
he
f
uture
sc
op
e
of
these
m
et
ho
ds
co
uld
be
use
fu
l
for
detect
ing
an
d
extracti
ng
(
dec
reasin
g
or
i
ncrea
sing)
op
e
n
water
s
urface
area
in
the
w
orl
d
with
dif
fere
nt
ba
nd
wav
e
le
ng
t
hs
and d
i
ff
e
ren
t s
at
el
li
te
d
at
a.
ACKN
OWLE
DGE
MENTS
We
w
ou
l
d
li
ke
acknow
le
dge
to
the
Ce
ntr
e
of
Excell
e
nc
e
(CoE)
for
ut
il
iz
at
ion
of
re
so
urces
a
nd
avail
abili
ty
of
so
ft
war
e
t
oo
ls
.
The
gr
at
e
fu
l
t
hanks
t
o
the
U
SG
S
portal
f
or
dow
nlo
a
ding
Lan
ds
at
data
with
a
fr
ee
of c
os
t.
REFERE
NCE
S
[1]
S.
Kali
r
aj,
e
t
al
.
,
“
Applicati
on
o
f
remote
s
e
nsing
in
d
et
e
ct
ion
of
fore
st
cov
er
changes
using
geo
-
stat
is
tical
ch
ang
e
det
e
ct
ion
m
at
ri
c
es
-
A
ca
se
stud
y
of
deva
n
ampatti
rese
rve
fore
st,
t
a
m
il
nadu,
Ind
ia
,”
Nat.
Environ.
P
oll
uti.
Te
chnol.
,
vol.
11
,
pp
.
261
-
269
,
2012
.
[2]
V.
Markogia
nn
i,
et
a
l.
,
“
La
nd
-
us
e
and
v
egeta
t
ion
c
hang
e
de
te
c
ti
o
n
in
pl
asti
r
a
ar
ti
f
ic
i
al
l
ake
catc
h
m
ent
(Gree
c
e)
b
y
using r
e
m
ote
-
se
nsing a
nd
GIS
technique
s
,”
In
t. J.
Re
mot
e
S
ens
.
,
v
ol.
34
,
pp
.
1265
-
1281
,
2013
.
[3]
M.
Volpi,
et
al.
,
“
Flooding
e
xt
ent
cartogr
ap
h
y
with
La
nds
at
TM
imager
y
and
reg
ula
ri
zed
Kerne
l
Fis
her
’s
discri
m
ina
nt
an
a
l
y
sis
,”
Comput
.
Geosci
.
,
vol
.
57
,
pp.
24
-
31
,
2013
.
[4]
B.
Brisco,
et
al
.,
“
Sar
pola
ri
m
et
ric
cha
ng
e
d
et
e
c
ti
on
for
floode
d
vege
ta
t
ion
,”
Int
.
J.
Digit.
Earth
,
vol.
6,
pp.
103
-
114
,
2013
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Ext
ra
ct
io
n
of
Water
-
body
Ar
ea
fr
om H
i
gh
-
r
esolutio
n La
nd
sa
t I
m
ag
er
y (
B
. Cha
nd
r
abab
u N
aik)
4119
[5]
Y.
O.
Oum
a
an
d
R.
Ta
teishi,
“
A
wate
r
inde
x
for
rap
id
m
appi
ng
of
shore
-
li
ne
cha
nges
of
five
Ea
st
Afric
an
Rif
t
Vall
e
y
l
ake
s:
an
empiric
a
la
na
l
y
s
i
s
using
La
ndsat
TM
and
ET
M+
dat
a
,
”
Int
.
J.
R
e
mote
Sens
,
vol.
27
,
pp.
3153
-
3181
,
2006.
[6]
S.
Lu,
et
al
.
,
“
W
at
er
bod
y
m
apping
m
e
thod
with
HJ
-
1A/B
sate
ll
i
t
e
imager
y
,”
Int
.
J.
Appl
.
Earth
Obs
.
Geoinf.
,
vol
.
13,
pp
.
428
-
434
,
2011.
[7]
P.
J.
J.
Desme
t
and
G.
Go
ver
s,
“
A
GIS
proc
edur
e
for
aut
om
at
i
ca
l
l
y
c
al
cu
la
t
ing
the
US
LE
LS
fac
t
or
on
topogra
ph
ical
l
y
complex
l
and
sca
pe
un
it
s
,”
J.
S
oil
Wa
te
r Conse
rv
.
,
vol.
51,
pp.
427
-
433
,
1996
.
[8]
W
.
Zhou
and
B.
W
u,
“
Assess
m
ent
of
soil
ero
sion
and
sediment
de
li
ver
y
ra
ti
o
usin
g
remote
sensing
and
GIS
:
A c
ase
stud
y
o
f
upstre
a
m
cha
obai
h
e
r
iv
er
c
atchm
ent
,
no
rth
Chin
a
,”
Int
.
J
.
Sed
iment R
es
.
,
vol.
23
,
pp
.
167
-
173
,
2008
.
[9]
Z.
Du,
et
al.
,
“
Esti
m
at
ing
surfac
e
wate
r
area
changes
using
ti
m
e
-
serie
s
La
n
dsat
d
at
a
in
the
qingji
ang
rive
r
basin,
China
,”
J. A
ppl
.
Re
mote
Sens
.
,
v
ol.
6,
2012
.
[10]
F.
Sun,
e
t
al
.
,
“
Com
par
ison
and
i
m
prove
m
ent
of
m
et
hods
for
id
e
nti
f
y
ing
wa
te
rbo
die
s
in
r
emotely
sensed
imager
y
,”
Int.
J. Remote
Se
ns
.
,
vol
.
33
,
pp
.
6854
-
6875
,
201
2
.
[11]
W
at
er
Bo
d
y
Ext
ra
ction
from
Multi
-
Source
Sat
ellit
e
Im
ages
.
Avail
ab
le
onli
ne:
h
ttp://
c
it
es
ee
rx
.
ist.ps
u.
edu/
vi
ewdoc
/downloa
d
?
do
i=10.1.
1
.
301.
8033
&re
p=r
ep1&
t
y
p
e=
pdf
.
[12]
H.
Xu,
“
Modifi
ca
t
ion
of
norm
a
li
sed
diff
ere
n
ce
wate
r
ind
ex
(N
D
W
I)
to
enha
nc
e
open
wa
te
r
f
e
at
ure
s
in
remote
l
y
sensed
imager
y
,”
Int
.
J. Remote
Sens
.
,
vol.
27,
p
p.
3025
-
3033
,
2
006
.
[13]
W
at
er
Bod
y
E
xtra
c
ti
on
And
Change
Detect
i
on
Based
on
Multi
-
Te
m
pora
l
SA
R
I
m
a
ges.
A
vai
la
bl
e
online:
htt
p://adsabs.
h
ar
var
d.
edu
/a
bs/20
09SP
IE.
7
498E.
.
96Z
.
[14]
H.
Zhou,
et
al.
,
“
La
ndsca
pe
and
wate
r
qualit
y
ch
ange
detec
t
ion
in
urba
n
wetl
and:
A
post
-
cl
assificat
ion
compari
so
n
m
et
hod
with
IK
ON
O
S da
ta
,”
Pr
oce
dia
Environ
.
Sci
.
,
vo
l.
10,
pp.
1726
-
1731
,
201
1
.
[15]
Z.
Ta
ng,
e
t
al
.
,
“
Ext
ra
ct
ion
of
wa
te
r
bod
y
b
ase
d
o
n
La
ndsat
TM5
i
m
age
r
y
-
A
ca
se
s
tud
y
in
the
Yang
tz
e
rive
r
,”
Adv.
Inf.
Comm
.
Tech
nol
.
,
vol
.
393
,
pp
.
416
-
420
,
2013
.
[16]
W
.
Li,
et
a
l.
,
“
A
compari
son
o
f
la
nd
surfac
e
wate
r
m
appi
ng
using
the
norm
a
li
z
e
d
diff
e
ren
ce
wate
r
inde
x
from
TM,
E
TM+ and A
LI
,”
Re
mot
e
S
e
ns
.
,
vol
.
5,
pp
.
5
530
-
5549
,
2013
.
[17]
S.
K.
McFe
et
ers
,
“
Us
ing
the
nor
m
al
iz
ed
diff
ere
n
ce
wa
t
er
inde
x
(
ND
W
I)
withi
n
a
geogr
aph
ic
info
rm
at
ion
s
y
st
em
t
o
det
e
ct
sw
imm
ing
pools
for
m
osquito aba
t
ement:
a
pra
ct
i
cal appro
ac
h
,”
Re
mot
e
Se
ns
.
,
vol
.
5,
pp
.
3
544
-
3561
,
2013
.
[18]
J.
W
.
Rouse,
et
al.
,
“
Monitor
ing
Vege
ta
t
ion
S
y
st
ems
in
the
Grea
t
Plai
ns
with
ERT
S
(Ea
rth
Resou
rce
s
Te
chno
log
y
Sate
llite)
,”
Pro
c
ee
dings
of
Thir
d
Earth
Re
sour
ces
Technol
ogy
Sate
llite
S
ymposium,
Gr
ee
nbel
t
,
ON,
Canada
,
v
ol
.
SP
-
351,
pp.
309
-
317
,
1973
.
[19]
E.
H.
W
il
son
an
d
S.
A.
Sad
er,
“
Dete
c
ti
on
of
for
est
har
v
est
t
y
p
e
using
m
ult
iple
dat
es
of
La
ndsa
t
TM
imager
y
,”
Re
mote
Sens.
En
vi
ron
.
,
vol
.
80
,
p
p.
385
-
396
,
200
2
.
[20]
S.
K.
McFee
te
r
s,
“
The
use
of
t
he
norm
al
iz
ed
d
iffe
ren
ce
wat
er
inde
x
(ND
W
I)
i
n
the
de
li
ne
atio
n
of
open
wate
r
fea
tur
es
,”
Int
.
J.
Re
mote
Sens
.
,
v
ol.
17
,
pp
.
1425
-
1432
,
1996
.
Evaluation Warning : The document was created with Spire.PDF for Python.