Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 2, No. 1,
April 201
6, pp. 221 ~ 22
8
DOI: 10.115
9
1
/ijeecs.v2.i1.pp22
1-2
2
8
221
Re
cei
v
ed
Jan
uary 10, 201
6
;
Revi
sed Ma
rch 7, 2
016;
Acce
pted Ma
rch 2
2
, 2016
Satellite-Based Land Surface Temperature Estimation
of Bogor Municipality, Indonesia
Ema Kurnia*
1
, I Nengah Surati Ja
y
a
2
, Widiatma
ka
3
1
Departme
n
t of Master of Science in Informat
i
on T
e
chnol
og
y for Natural Re
sources Ma
na
geme
n
t, F
a
culty
of Mathematic
and N
a
tural Sc
ienc
e,
Bogor A
g
ricult
ural U
n
iv
ersit
y
2
Departme
n
t of F
o
rest Manag
ement, F
a
cult
y of
F
o
restry
, B
ogor Agr
i
cultur
al Univ
ersit
y
3
Departme
n
t of Soil Scie
nce a
nd La
nd R
e
sou
r
ces, F
a
cult
y
of
Agricultur
e, Bogor Agr
i
cultur
al Univ
ersit
y
*Corres
p
o
n
d
e
n
c
e author: em
a
k
urni
a@gm
ail.
com
1
, ins-ja
ya
@ipb.
ac.id
2
, w
i
diatmak
a
@
y
a
h
oo.com
3
A
b
st
r
a
ct
T
he e
a
rth
’
s
av
erag
e te
mp
erat
ure h
a
s b
e
e
n
a
big
issu
e o
n
th
e gl
ob
al w
a
r
m
i
ng. T
he w
a
r
m
i
ng of th
e
earth is
l
a
rge
l
y
the
r
e
sults of emissio
n
of
car
bon
di
oxid
e
an
d oth
e
r gr
ee
nh
ouse
gass
e
s (
G
HG) from
hu
ma
n
activities. As a
hinterl
a
n
d
of the Cap
i
tal Cit
y, in
the last tw
o decad
es, Bogor is a
l
so
getting w
a
r
m
er
in
comparis
on w
i
th the prev
io
us deca
des. T
h
is
pap
er pr
es
ents
how
the lan
d
surface te
mp
er
ature (LST
) ha
d
bee
n esti
mate
d usin
g Spl
i
t-W
i
ndow
(SW
)
alg
o
rith
m an
d
how
its spatial
distributi
on i
n
Bogor Mu
nici
p
a
lit
y
w
a
s compute
d
.
T
he s
pectral
r
adi
ance
of
La
n
d
sat-8 T
I
R
b
and
s 10
an
d 11
, th
e
em
i
ssi
vi
ty va
l
u
e
s
, a
n
d
wa
te
r
vapor us
ed as
the inp
u
t on S
W
Algorith
m
. T
he study reve
al
ed that the te
mperatur
e
w
i
thin
the built-u
p ar
ea,
have
w
a
rmer t
e
mper
ature t
h
a
n
the
i
r surr
ou
n
d
in
g ra
ng
ing
from 4
0
0
C to
45
0
C of 3,4
03.9
h
a
. T
he
use
of
SW
alg
o
rith
m is q
u
ite
r
e
li
ab
le a
nd accur
a
te
t
o
esti
mate
th
e
LST
d
e
riv
e
d
from La
nds
at-8
h
a
vin
g
a mean
devi
a
tion
of onl
y 2.7%, less than
stand
ard ac
ceptab
le of 10
%.
Ke
y
w
ords
: LST, SW
Algorithm
, TIR
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
No
w, the glob
al wa
rming
h
a
s b
een
an in
ternatio
n
a
l issue th
at attra
c
ting the i
n
ternational
attention. At the site lev
e
l, the land
surf
a
c
e
temp
eratu
r
e
(LST
) ha
s b
een
use
d
by ma
ny
resea
r
chers
as a
n
indi
cat
o
r of en
ergy
balan
ce
. Spe
c
ifically, the
LST ha
s bee
n used a
s
a
key
para
m
eter th
at de
scribe
s t
he la
nd
su
rfa
c
e
pro
c
e
s
se
s. Within th
e u
r
ban
a
r
ea
s, t
he LST
might
be
clo
s
ely relate
d to the urba
n heat isla
nd
(UHI
) whi
c
h
is mainly affected by the h
u
man a
c
tivities.
The cau
s
e
s
o
f
UHI mainly
come
s from the hum
an a
c
tivities that includ
e lifestyle
, that use fo
ssil
fuel in main
human
activities, e.g., liqu
i
d petrol
eum
gasse
s for
coo
k
in
g, ele
c
tricity for ro
o
m
heating an
d cooli
ng, gaso
line or die
s
el
for trans
port
a
tion, indu
stry etc. The waste from th
e
energy u
s
ag
e by eve
r
y
hou
seh
o
ld
was
also a
seco
nda
ry co
ntributo
r
to t
he h
eat. As the
popul
ation in
the
city also grow
s, thi
s
tend
s to
expand
the
are
a
an
d in
crea
se it
s ave
r
a
g
e
temperature.
Develo
pment
of b
u
ilt-up
a
r
ea
wi
th
the
less g
r
e
en
o
pen
area m
a
y cau
s
e
glo
b
a
l
temperature
cha
nge
s that
result in a chang
e of
cli
m
ate eleme
n
t
s, espe
cially
the increa
se
in
temperature.
The UHI extent may vary across a cit
y
,
dependin
g
on the spati
a
l distrib
u
tion
of the
gree
n hou
se
gasse
s (GHG
) emitter an
d
abso
r
b
e
r.
Ta
hare
p
o
r
ted th
at heat islan
d
s ca
n develo
p
in ‘po
c
kets’
a
r
oun
d
single
building
s
and
temperature
differen
c
e
s
of 4 °
C
h
a
ve
been
re
po
rted
along
a
sin
g
l
e
st
reet [1]. It
is al
so m
enti
oned
that th
e
UHI i
s
affe
cted by
the
hei
ght an
d
spa
c
i
n
g
of building
s
and their o
r
i
entation rel
a
tive to t
he prevailing win
d
that rest
rict
airflow a
nd limit
cooli
ng. Urba
n with very hi
gh ratio
between buil
d
i
ng
height an
d st
reet width, a
s
well a
s
the v
e
ry
den
se settlement may ha
ve high wa
st
e heat from h
u
man a
c
tivities. In Bogo
r, whe
r
e only a
few
tall buildi
n
g
s
are availa
b
l
e, the d
o
mi
nant h
eat
contributo
r
s
might be
co
me fro
m
h
u
man
(metab
olic an
d no
n-m
e
tab
o
lic) a
c
tivities and
road
tra
ffic. Fan
and
Sailor
pointe
d
out
that ro
ad
traffic contrib
u
ted ab
out 3
2
% of heat e
m
issi
on
wh
ile
the hum
an
metaboli
c
he
at emission
is only
8% [2].
Freq
uently,
cities have
warmer lan
d
’
s av
e
r
ag
e t
e
mpe
r
ature t
han it
s
su
rroundi
ng
sub
u
rb
an a
n
d
ru
ral a
r
e
a
s
. The
ra
pid
develop
men
t
of built-up
are
a
s i
n
B
ogor,
su
ch
as
settlement
s, tall building
s
for hotel, commerce,
a
nd/or offices,
mainly alters the phy
sical
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 221 –
228
222
cha
r
a
c
teri
stics of
the l
and
surfa
c
e,
from
veget
ation
to
non
-veg
etated a
r
e
a
s.
Th
e repla
c
em
en
t of
gree
n veg
e
ta
ted surfa
c
e
s
with n
on-ve
g
e
tated a
nd n
on-p
o
rou
s
u
r
ban m
a
teri
als
with hig
h
h
eat
cap
a
city and l
o
w
sola
r refle
c
tivities, su
ch
as con
c
rete
masse
s
, asp
halt roa
d
s
an
d metal surfa
c
e
s
exhibit a high degree of therm
a
l inerti
a [3]. Thes
e area
s are ch
ara
c
teri
ze
d by a high level of
absorptio
n of
sol
a
r ra
diati
on, with
a
greater c
apa
city for the
r
mal
co
ndu
ctivity as
co
mpa
r
ed
to
natural su
rfaces
[4].
Within the bui
lt-up are
a
, the glass-walle
d building
s
m
a
y reflect the incomi
ng sho
r
t-wave
sola
r ra
diatio
n (sunlig
ht), and con
s
eq
u
ently the surf
ace, a
s
well
as the ai
r te
mperature i
n
the
surro
undi
ng
building
may increa
se. In the cities, it
is also quite
commo
n tha
t
emissi
on from
publi
c
traffic
as
well
as p
r
i
v
ate traffic m
a
y incr
ea
se
the G
H
G
rel
e
ase
d
to th
e a
t
mosph
e
re, then
eventually in
crea
se th
e surf
ace te
mpe
r
at
ure. Som
e
ga
sses th
at are
emitted into t
he atmo
sp
here
will act as a
gree
nho
use gas that is transparent
to sho
r
t-wave solar ra
diation
and absorb l
ong-
wave
radi
ation of the e
a
rt
h thus in
cre
a
s
ing
gl
ob
al warmin
g. Urba
n develo
p
me
nt can
rai
s
e
the
local te
mpe
r
ature
of the
city whe
r
e th
e rate
of tem
peratu
r
e
ri
se
is p
r
op
ortio
n
a
l to the
rate
of
urba
n develo
p
ment [5]. T
he increa
se
of populatio
n
also ca
used
an incre
a
se
of heat waste,
mainly emitted from no
n-meta
boli
c
activities
such as vehi
cl
es, pe
rso
nal
waste, en
e
r
gy
con
s
um
ption
and manuf
actures et
c. This may a
ffect local
climate chan
ge espe
ciall
y
air
temperature
dire
ctly and indire
ctly [6].
The temp
e
r
at
ure of the urban area
s might be effectively
manag
ed a
n
d
slightly mo
dified by increasi
ng the
e
x
tent of GHG
abso
r
b
e
r o
r
by redu
cin
g
the
sou
r
ce of heat gasses. The com
m
on
strategy
app
lied is by increa
sing the a
m
ounts of he
at
energy ab
sorbed an
d sto
r
ed in the veg
e
tation. V
ege
tation woul
d be a very effective way a
s
it
delivers seve
ral me
cha
n
isms of co
oling
simult
ane
ou
sly and it ha
d been recog
n
ize
d
as a v
e
ry
che
ap way to implement th
e cooli
ng stra
tegy.
To spatially
measure the spatial di
strib
u
ti
on of the heat emitter and heat ab
sorbe
r
, it
need
s to de
velop a techniqu
e to derive land
surface temp
eratu
r
e qui
ckly, con
s
iste
ntly,
accurately, comprehe
nsiv
ely and with
a rea
s
o
nabl
e
co
st. One te
chni
que that f
r
equ
ently app
lied
is by usi
ng t
he rem
o
te sensi
ng ap
pro
a
ch
es.
T
r
adit
i
onally, the land surfa
c
e
temperature
is
mappe
d by usin
g interp
o
l
ation techni
q
ue usin
g the
data record
ed by each
national weat
her
station. Inte
rpolation
met
hod
fo
r sp
arsely statione
d
net
work
h
ad b
een
a
focu
s of
ma
ny
resea
r
chers,
inclu
d
ing th
e
examinatio
n
of geo
statistics and det
ermini
stic ap
proa
ch
es.
T
h
is
method
will p
r
ovide very rough m
ap si
nce the
dist
a
n
ce
s bet
wee
n
climatol
ogy
stations
are
very
low. The different inte
rpol
ation method
can
p
r
ovide d
i
fferent accu
racy and d
e
viation.
No
w, the ava
ilability of re
motely se
nse
d
data
re
co
rd
ed u
s
ing
the
r
mal ba
nd
s, couplin
g
the available
middle
-
infra
r
e
d
, near-infra
red, as
well a
s
the visible
band
s,ha
sgiv
en a very go
od
pro
s
pe
ct. Th
ere i
s
no i
n
te
rpolatio
n met
hod
req
u
ire
d
in this ap
pro
a
ch. T
he te
m
peratu
r
e
s
we
re
derived
from
every
g
r
id of the
data. The
algo
ri
thm
used by
com
b
in
ing the
therm
a
l, nea
r-i
nfra
red
and re
d-b
a
n
d
of Landsat data, the pixel-ba
s
ed te
mperature mi
ght be deriv
ed. On the LST
estimation, t
he accuracy
of LST esti
mation is
m
a
inly affected by the surface
capabilit
y o
f
emitting radi
a
t
ion. In many algorithm
s, the LS
T e
s
timations a
r
e b
a
s
ed o
n
the a
s
sumption th
at
the grou
nd su
rface a
c
t
s
as
a blackb
ody (emissivity equals o
ne).
The Split-Win
dow
(SW) alg
o
rithm i
s
the
most
commo
nly used, giv
en that this al
gorithm
remove
s the
atmosp
he
ric
effect and obt
ains the LS
T
from the linea
r or no
nline
a
r combin
ation
of
the brig
htne
ss tempe
r
atu
r
es of two
ad
jace
nt cha
n
n
e
ls
cente
r
ed
at 11 and
12
μ
m. W
an [7
]
prop
osed a
n
e
w
refinem
en
t of the gen
eralize
d
SW
al
gorithm
add
e
d
with
a qu
ad
ratic te
rm of t
h
e
differen
c
e a
m
ong
st the brightn
e
ss te
mperatur
e
of the adja
c
en
t thermal inf
r
ared chan
n
e
ls.
Remote
se
n
s
ing i
s
a
practical way to acco
mplish
the monito
ri
ng an
d a
s
se
ssi
ng the
LST
becau
se
it re
pre
s
ent
s a re
latively
low-cost
a
n
d
ra
pid
metho
d
to
a
c
qui
re
up
-to-date info
rmat
ion
over a l
a
rg
e
geog
rap
h
ical
area. T
he
stu
d
ies
rel
a
ted t
o
the la
nd
su
rface
tempe
r
ature
s
can
al
so
be foun
d in
Da
sh [8] an
d
Akho
ond
zad
eh an
d Sara
djian [9] an
d
Liu an
d Zha
n
g
[10]. The L
S
T
has be
en
used a
s
a
key
para
m
eter in
the p
h
ysi
c
s
of land
surfa
c
e
processe
s. Other relat
ed
studie
s
o
n
la
nd surfa
c
e te
mperature
ca
n be fou
nd i
n
Lian
g
et al
.
[11]; Zhang
and
He [12].
In
orde
r to
cont
rol the u
r
ba
n
developm
ent
esp
e
ci
ally
as a p
r
event o
f
urban
heat
islan
d
in Bog
o
r
Munici
pality and also as
a predi
ction of t
he land
su
rfa
c
e temp
eratu
r
e, the main
obje
c
tive of the
study wa
s to estimate the
land su
rface
tem
peratu
r
e
(LST) ba
sed
on the Land
sat-8 O
L
I an
d
TIRS image
ri
es u
s
ing Split
-Win
d
o
w
(SW) algo
rithm in Bogor Mu
nici
pality.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Satellite-Based Lan
d Surf
ace Tem
p
e
r
a
t
ure Esti
m
a
tion of Bogor M
unici
pality…(Em
a
Kurnia)
223
2. Rese
arch
Metho
d
2.1 Stud
y
Area
Bogor M
unici
pality is loca
ted in a hinterla
n
d
area
of the Capit
a
l City Jakarta that
con
s
i
s
ted of
6 distri
cts
an
d 68 village
s,
with
area ext
ent of abo
ut 11.
694
ha. G
eographi
cally
, the
city is locate
d betwe
en 0
6
o
48'40
'' an
d
06
o
46'2
2'' E
a
st longitu
de
; and betwe
en 6
o
30'
53'' and
6
o
40'08
'' Sou
t
h latitude (Fi
gure 1
)
. The
city has
bee
n the main d
e
stinatio
n of many dome
s
tic
tourist
s
within
the Jab
odeta
bek, and
the resi
dent
ial
a
r
ea of
many
p
eople
who
are working
in t
he
Capital
City
Ja
karta.
Now, the city
ha
s bee
n
frequ
e
n
tly sele
cted
by Hi
s Excell
ency P
r
e
s
ide
n
t
Jo
ko
Wido
do
as hi
s
se
con
d
wo
rki
ng offi
ce. Thu
s
, Bo
gor m
uni
cipal
ity which i
s
n
o
w b
e
coming
the
cente
r
of vari
ous
activities
su
ch a
s
com
m
erce
, tou
r
ism, the re
side
ntial and
pre
s
ident pal
ace
ha
s
been
chosen
as a
study
site. Bogo
r
Munici
pa
lity
has a hi
gh
rate of p
opul
ation g
r
o
w
th
and
developm
ent.
Figure 1. Study area resea
r
ch
2.2
The Suppor
ting Data
The main da
ta used are Digital Satellite Landsat-8
OLI (Operational Lan
d Image
r)
image
rie
s
pat
h 11
2-row 65
; acq
u
ire
d
o
n
13
th
Septem
ber
201
4. Th
e ba
nds u
s
ed
parti
cula
rly red
band
(ba
nd 4
)
, nea
r infrare
d
ban
d (band
5), an
d The
r
mal Infra
r
ed
Senso
r
(TIRS
)
, namely b
a
n
d
10 and 11. The data wa
s capt
ure
d
at approxim
at
ely 10:00:37
a.m. local time. Land
sat
-
8
provide
s
m
e
tadata of th
e
band
s
su
ch a
s
the
r
mal
co
nstant a
nd
re
scaling fa
cto
r
value that u
s
ed
for cal
c
ulatin
g the LST. Other p
r
ima
r
y data
used
to accom
p
li
sh the stu
d
y
are su
rface
temperature,
land cove
r condition
man
ually
me
asured a
n
d
ob
se
rved at
5 Poi
n
ts. Althoug
h t
h
e
Land
sat-8 O
L
I recorded
o
n
13
th
Septe
m
ber
201
4, while th
e gr
o
und me
asure
m
ent wa
s d
o
ne in
March 2
015,
the differe
nce
betwe
en the
temperat
ure
in March
201
5 and
in Sept
embe
r 20
14 i
n
the day time is not
sig
n
ificantly differen
t. Th
e land cover
a
nd land
use
within at the
measurement
points did
n
’t cha
nge d
r
a
s
tically.
2.3
Soft
w
a
re, Ha
rd
w
a
re, an
d Tools Us
ed
The spatial
a
nalysi
s
was mainly
pe
rformed
u
s
in
g ArcMap
9.3
whi
l
e the d
a
ta p
r
oce
s
sing
of Land
sat imageri
e
s
were pro
c
e
s
sed u
s
ing ERD
AS i
m
agine 9.1.
The processi
ng platform was
a pe
rsonal
compute
r
with
printin
g
devi
c
e
s
.
For gro
und m
e
a
s
ure
m
ents, the t
ools
used
were
Therm
o
mete
r, GPS, and Camera.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 221 –
228
224
2.4 Split-Windo
w
(S
W
)
Algo
rithm
The the
o
ry
unde
rlying th
e techniqu
e
of SW
i
s
t
hat the
radi
ance attenu
ation for
atmosp
he
ric absorption
is p
r
op
ort
i
onal to
th
e ra
dian
ce
differen
c
e
of simulta
neou
s
measurement
s at two different wavele
ngths
[13]. The SW tech
nique u
s
e
s
two TIR ba
n
d
s
typically locat
ed in the at
mosp
he
ric
wi
ndo
w betwee
n
10.30 a
nd
12.50
μ
m. F
u
rthe
rmo
r
e, the
followin
g
inp
u
t
requi
red
by
the SW al
gori
t
hm is
b
r
ightn
e
ss temp
erat
ure, me
an a
n
d
differe
nce i
n
land surfa
c
e
emissivity (LSE) and wate
r vapor.
The
la
nd su
rface
te
mpe
r
at
ure wa
s cal
c
ulated
by
co
n
v
erting
the Di
gital
Nu
mbe
r
(DN) of
the two
the
r
mal ba
nd
s
(b
and
10, 1
1
) i
n
to top
of at
mosp
he
ric ra
dian
ce val
ue,
and
then
int
o
the
brightn
e
ss te
mperature.
T
herefo
r
e,
co
rrection
of the
spe
c
tral
em
i
s
sivity has to
be in
clud
ed.
LSE
wa
s de
rived f
r
om
NDVI thresh
old meth
o
d
by usi
ng
th
e OLI ba
nd
s
4, 5, and th
e
emissivity values
of TIR band
s
10 and 1
1
. In physical Atmosp
heri
c
, t
he
moistu
re cont
ent of the earth atmosp
here
is on
e of the
most imp
o
rt
ant paramete
r
s; it is
h
a
rd
to rep
r
e
s
ent
water va
po
r becau
se of
its
spa
c
e
-
time variation [14].
Wate
r vapor
conte
n
t wa
s the avera
ge o
f
moisture fo
r Bogor a
r
ea.
To
obtain the
wa
ter vapo
r
con
t
ent in Septe
m
ber
201
4
wa
s to
multiply the moi
s
ture
by the ratio
of
water vap
o
r
conte
n
t to th
e total stand
ard atmo
sph
e
ric p
r
ofile
s for the tropi
cal area [15]. The
brightn
e
ss te
mperature,
mean
and
di
fference
i
n
L
SEand
water vapo
r
conte
n
t we
re
u
s
e
d
to
cal
c
ulate the
LST. The formula is in eq
uation-1 [16]:
LST = BT
10
+ C
1
(BT
10
-BT
11
) + C
2
(BT
10
-BT
11
)
2
+ C
0
+ (C
3
+C
4
w) (
1
-
ε
) + (C
5
+C
6
w)
∆
ε
(1)
whe
r
e:
LST is Lan
d surfa
c
e tem
p
eratu
r
e (
0
Kelv
in); C
0
to C
6
are Split-win
d
o
w Coefficie
n
t
values
[17]; BT
10
and BT
11
are B
r
ightne
ss tem
peratu
r
e
s
of
band 1
0
an
d
band 1
1
(in
0
K);
ε
is
mean LSE of
TIR band
s; w is atmo
sp
h
e
ric
wate
r vapor
content;
and
∆
ε
is diff
eren
ce in
LSE.
2.5 Ground
Me
a
s
uremen
ts
For validatin
g
the land su
rf
ace temp
erature
de
rived from the mode
l in equation
1, the
authors mad
e
g
r
ou
nd m
e
asu
r
em
ents.
Grou
nd
-ba
s
e
d
temp
eratu
r
e me
asure
m
ents
we
re
taken in
13 days d
u
ri
n
g
March 20
1
5
startin
g
fro
m
2
nd
March to 28
th
March.
The mea
s
u
r
ement day
s were
sele
cted ran
domly, at the following sp
ecific date: 2
nd
,4
th
,7
th
,9
th
,11
th
,14
th
,16
th
,18
th
,21
st
,23
rd
,25
th
,
28
th
, and 30
th
. The locati
ons
of mea
s
urem
ent were sel
e
cte
d
p
u
rpo
s
ively at
five different
locatio
n
s
by
con
s
id
erin
g t
he cha
r
a
c
teri
stics of
la
nd
cover to be
repre
s
e
n
ted
such
as built-up,
urba
n fo
re
st, rice
field, h
o
u
s
ing
an
d ma
nufactu
red
a
r
ea. The
temp
eratu
r
e m
e
a
s
urem
ents we
re
done bet
wee
n
10.00 – 12
.00 am local
time, which i
s
the clo
s
e
s
t time to the
Land
sat-8 image
captu
r
ed.
A total of five th
ermom
e
ters
at mea
s
u
r
em
ent poi
nts
we
re m
ounte
d
a
t
1.5 m
heig
h
t
in
an ope
n sp
ace that prote
c
ted from solar
radiatio
n.
2.6 Data
An
aly
s
is
To
kno
w
t
he
con
s
i
s
te
ncy a
n
d
th
e relation
shi
p
bet
wee
n
these
gro
und-ba
sed
temperature
s
and the ave
r
age
of LSTs estimation,
then the Pea
r
son’
s correlat
ion co
efficien
ts
were de
rived
.
Besides, th
e deviation b
e
twee
n the land’
s su
rface
temperatu
r
e
estimate (L
ST)
and the a
c
tua
l
temperatu
r
e
were
cal
c
ulat
ed by usin
g mean deviati
on (M
D) a
s
in
equation
-
2:
MD
∑
100%
(2)
3. Results a
nd Analy
s
is
3.1
Land Surfac
e Tempera
t
u
r
e Distrib
u
ti
on
The in
put p
a
ram
e
ters to
derive
the
LST of
SW
algorith
m
in
clude
s the
bri
ghtne
ss
temperature
of the two adj
ace
n
t
band
s
of the TIRS, mean a
nd dif
f
er
en
ce of e
m
issivity whi
c
h is
an FV
C
can
be e
s
timated
from th
e re
d an
d ne
ar-i
nfrared
refle
c
tance
of the
OLI ba
nd
s, a
nd
water vap
o
r
conte
n
t. LST output port
r
a
y
ed that it
va
ried fro
m
301
to 322 Kelvin and converted
into Celsi
u
s b
y
subt
ra
cting
273.15.
The
spatial di
strib
u
t
ion of LS
T
wi
thin
the
study
area i
s
sh
own
in Figu
re 2.
Acro
ss the e
n
tire
study area, LST
valu
es in
crea
sed
from the o
u
tskirt
s toward
s
the
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Satellite-Based Lan
d Surf
ace Tem
p
e
r
a
t
ure Esti
m
a
tion of Bogor M
unici
pality…(Em
a
Kurnia)
225
inner u
r
b
an area
s, whi
c
h
range
d from
27
0
C to 50
0
C, with a mean of 37
0
C and a stan
d
a
rd
deviation of
3.67
0
C. The rise of surf
ace temperature
will affect to the increasi
ng of
air
temperature
esp
e
ci
ally in the urba
n areas (e
.g., [18]). The LST pattern was f
ound to be n
on-
symmetri
c
al
but rather
co
nce
n
tric,
with
high
-tem
p
e
rature
zone
s
clustered
towa
rds the
cente
r
of
the study a
r
e
a
. The a
r
ea
s
with the lo
we
st veget
ation levels were
correspon
ded to
the
land co
ve
r
types of built-up are
a
with t
he value of L
S
T rang
ed from 40
0
C to 5
0
0
C. Conve
r
sely, high values
of NDVI in
dicating the p
r
e
s
ence of
g
r
een
vegetation which mainly
o
c
cu
rs, at the
south
e
rn
pa
rt of
the stu
d
y a
r
e
a
. The
corre
s
pondi
ng l
and
cover
cla
s
ses are fa
rmlan
d
and
g
r
a
s
s a
r
ea
with the
L
S
T
rang
ed from
27
0
C to 39
0
C. Some patch
es of high NDVI were
also noticea
ble
within the ce
ntral
regio
n
of the
study area
and
corre
s
p
o
nded to
th
e
urba
n forest
area. T
he p
e
r
ce
ntage
area
according to the tempe
r
atu
r
e interval
s was sho
w
n in
Table 1.
Figure 2. LST map
Table 1. LST
percenta
ge a
r
ea
Class
Temper
ature
Interval (
0
C)
Area (H
a)
Proportion
(%)
1
27 - 30
131.4
1.12
2
30 - 35
2,907.0
24.86
3
35 - 40
5,116.0
43.75
4
40 - 45
3,403.9
29.11
5
45 - 50
136.3
1.17
Total 11,694.6
100.00
The high
est LST
of
m
o
re than
4
5
0
C is
spread
in
the
No
rth
and
Center pa
rts of
Bogo
r
with an a
r
ea
of 136.26 Ha. The small
e
st
area of 13
1.4 Ha is va
rie
d
from 27
0
C t
o
30
0
C. LST
with
the large
s
t a
r
ea of
5,116.0
5
Ha i
s
va
rie
d
from
35
0
C t
o
40
0
C. It ha
s b
een
largel
y demon
strated
that cities
wi
th variable
land
scape
s a
nd clim
at
es can exhibit
tempe
r
atures several
de
gree
s
highe
r tha
n
t
heir
ru
ral
su
rroun
ding
s. T
he a
r
ea
with
high
LST i
s
an
are
a
that
den
se
with
the
settlement
s and roa
d
s (b
uilt-up
)
.
LST
and l
and
co
ver value
s
were
com
pute
d
to un
derst
and
further h
o
w
LST intera
ct with land co
ver par
a
m
ete
r
s. The tabul
ated LST an
d land cove
r as
sho
w
n in Ta
b
l
e 2.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 221 –
228
226
Table 2. Tab
u
lated LST a
nd land
cover
No Land
Cover
LST(
0
C)
Min Mean
Max
1
Urban
Forest
27 34.5 42
2 Waterbod
y
28
36
44
3 Farmland
27
36
45
4
Grass
29 36.5 45
5 Built-up
29
39
49
The hig
h
e
s
t mean LST
wa
s foun
d i
n
bu
ilt-u
p area havin
g a
temperature
of 39
0
C,
followe
d by the grass of 3
6
.5
0
C, farmla
nd and
wate
rbody of 36
0
C, and the lowest tempe
r
at
ure
detecte
d in th
e urb
an forest of 34.5
0
C. This impli
e
s th
at urba
n dev
elopme
n
t ha
s brou
ght up L
S
T
by repl
aci
ng
natural
veget
ati
on with
a
non-evapo
rati
ng
a
nd non
-tran
s
pi
ring
su
rface
such
a
s
stone, metal,
and con
c
rete
[19] [20]
.
3.2
Relatio
n
ship
bet
w
e
e
n
LS
T and Air Te
mperatu
r
e
The co
rrelati
on between
LSTs an
d air tem
peratu
r
e
in the canop
y layer was
gene
rally
high, d
ue to
the tran
sfer
of therm
a
l e
n
e
r
gy emi
tted from the
surfa
c
e to
the
atm
o
sp
here [21]
[3].
The ai
r tem
p
eratu
r
e
ha
s a
po
sitive co
rrelation
wi
th t
he LST
at a
day-time
of d
a
y 1 a
nd d
a
y 11
whe
r
e th
e ai
r temperature
norm
a
lly fluct
uate le
ss
tha
n
LST a
c
ross a given
area
duri
ng the
d
a
y
[22]. In gen
eral, the
clo
s
e
relation
ship
b
e
twee
n LST
and th
e ai
r te
mperature
ha
s b
een
sho
w
n to
be valid [23] [24].
Pearson’
s co
rrel
a
tion coef
ficients a
naly
s
is
(
r)
sh
ows moderately high a
s
soci
ation with
the value of 0.78 with the
significant correlat
ion of
> 0.01 (1%)
and me
an de
viation of 2.7%.
S
o
me re
se
ar
ch al
so
sho
w
s t
he si
gnif
i
c
a
nt
relat
i
on
shi
p
duri
ng nig
h
t
-time mea
s
u
r
eme
n
t [25] [26].
Although
the
r
e
are
differences bet
we
en LST
an
d
air tempe
r
ature,
a m
o
derate
to
hi
gh
percenta
ge
o
f
the ai
r temp
eratu
r
e
ca
n b
e
e
s
timat
ed f
r
om the
LST
a
s
in
dicated
b
y
coeffici
ents
o
f
determi
nation
(R
2
). T
he val
ue R
2
obtain
ed was i
n
the
ran
ge of 0.6
1
(Fig
ure 3),
this me
an
s th
at
the variatio
n
of
in situ
te
m
peratu
r
e
ca
n
be explai
ned
by the LS
T in
to 61%. In thi
s
ca
se, the
st
udy
area
wa
s lim
ited only for
Bogor M
uni
ci
pality, with
the a
s
sumpti
on that the
ecol
ogi
cal fa
ctor
outsid
e
the st
udy area
wa
s not con
s
ide
r
ed.
Figure 3. Rel
a
tionship bet
wee
n
LST an
d air tempe
r
a
t
ure
By all mea
n
s,
we
shoul
d n
o
t forg
et that
t
he obtai
ned
relation
ship
is
based
on th
e
data of
only two
day
s, ho
weve
r
complex, me
a
s
ureme
n
t ca
mpaign
s. In
the futu
re,
wh
en u
s
in
g d
a
ta of
more m
e
a
s
u
r
ements
on d
a
ys with
simi
lar envi
r
onm
ent
al conditio
n
s to that of
the investig
ated
days, the re
sult could b
e
refined.
y
=
0.6829x
+
16.835
R²
=
0.6119
30
35
40
45
50
20.0
25.0
30.0
35.0
40.0
45.0
LST
(
0
C)
Air
Temperature
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Satellite-Based Lan
d Surf
ace Tem
p
e
r
a
t
ure Esti
m
a
tion of Bogor M
unici
pality…(Em
a
Kurnia)
227
4. Conclusio
n
Land
sat-8 TI
RS sen
s
or i
s
cap
able
of reco
rdin
g the
radia
n
t heat
data on th
e
earth'
s
surfa
c
e
in the
thermal i
n
fra
r
ed
spe
c
trum.
The radi
a
n
ce heat info
rm
ation in the th
ermal
sp
ectru
m
is strongly inf
l
uen
ced by th
e su
rface temperature a
n
d
the obje
c
t
emissivity. As describ
e by the
Stefan-Boltzmann l
a
w, th
e total amo
u
n
t of emitted
energy is
directly propo
rtional to th
e fo
urth
power of the
obje
c
t’s temp
eratu
r
e. The
split-win
d
o
w
algorith
m
wa
s u
s
ed to
det
ermin
e
the L
S
T of
an area. The
land u
s
e cha
nged in Bo
go
r Muni
cipality
from the non
-built area int
o
built are
a
h
a
s
affected to th
e increa
sing
of LST. The land surf
a
c
e t
e
mpe
r
ature i
n
Bogor M
u
n
i
cipality ca
n
be
estimated from the therm
a
l sensors of satellite
data exhibit by r =
0.
78, with
si
gnificant
correl
ation
0.01 of th
e ai
r
temperature
and m
ean
de
viation of 2.7
.In the fram
e
of this stu
d
y, a
data coll
ection in different seasons
coul
d also be
a new direction, whi
c
h
can provide a possi
bility
to examine the spe
c
ific
sea
s
on
al
feature
s
and e
nabl
e their co
mpa
r
i
s
on.
Ackn
o
w
l
e
dg
ments
Special
than
ks to th
e la
b
o
rato
ryof Ge
ophysi
cs a
n
d
Meteo
r
olo
g
y of Bog
o
r Ag
ricultu
r
al
University for the equi
pme
n
t provide
d
a
nd Nil
s
Nölke
from Gotting
en University for su
gge
stio
ns,
corre
c
tion
s a
nd comm
ent
s for improvement of t
he original ma
n
u
script. It helps a lot for the
revisio
n
of this pap
er.
Referen
ces
[1]
T
aha HG, Akbari H, Sa
ilor
D
,
Ritschard
R.
Caus
es an
d
Effect of Heat
Islands: T
he
Sensitiv
ity o
f
Urba
n Microc
li
mates
to Surfa
c
e Para
meters
and A
n
thro
po
g
enic
Heat
. L
a
w
r
ence B
e
rke
l
e
y
Lab
orator
y.
Rep
o
rt No. 298
64. La
w
r
enc
e Berkel
e
y
, D
a
vi
s Califor
nia, U
SA. 1990.
[2]
Fan H, Sa
ilor
DJ. Mod
e
li
n
g
the Imp
a
cts of
Anthro
po
g
enic
He
ating
on th
e Urb
a
n
Clim
ate o
f
Phil
ade
lp
hia:
A Compar
iso
n
of Impl
ement
ations i
n
T
w
o
Pbl Sch
e
mes.
Atmosp
heric Enviro
n
m
ent.
200
5; 39: 73–
8
4
.
[3]
Arrau CP
a
n
d
Pe
na MA.
T
he Urba
n
Heat Is
l
and
(
UHI) Effect. 201
0. Avai
la
bl
e o
n
li
ne
at
http://
w
w
w
.
ur
b
anh
eatisl
a
n
d
s.com (Accesse
d Janu
ar
y
3
rd
,
201
6).
[4]
Rose LA, Dev
adas MD. Ana
l
y
s
is of and S
u
rf
ace T
e
mperature an
d La
n
d
Use/ Lan
d Cover T
y
p
e
s
Using
Remot
e
Sensi
ng Imag
e
r
y
– A C
a
se i
n
Chen
na
i Cit
y
,
India. Sev
enth
In
ternatio
nal Confer
en
c
e
on Urb
an Cl
im
ate, Yokoh
a
ma
, Japan. 20
09.
[5]
F
u
kui Y. A stu
d
y
on s
u
rface
temperatur
e p
a
ttern
s in
the
T
o
ky
o
metrop
o
litan
area
usi
n
g aster
data
.
Geoscience Journal
. 200
3; 7: 343-
346.
[6]
Mas’at A. 200
8
.
T
he impact of devel
opme
n
t on clim
ate vari
ations i
n
Jakart
a.
BMKG Bulletin
. 2008; 4.
[7]
W
an Z
.
Ne
w
refi
neme
n
ts an
d v
a
li
d
a
tion
of th
e co
llecti
on-
6 MODIS l
and-s
u
rfa
c
e
temperatur
e/e
m
issivit
y
prod
u
c
t.
Rem
o
te Sens. Environ
. 2
0
14; 140: 3
6–4
5
.
[8]
Dash, P. La
nd surfac
e temper
ature a
nd emis
s
i
vit
y
retrieval fro
m
satellite m
easur
ement
s
.
Dissertati
on. Institut fur Meteorolo
g
ie
u
nd Kl
i
m
aforschu
ng. ISSN 094
7-8
6
2
0
. 2005.
[9]
Akhoo
ndz
ade
h
M, Saradjian
MR. Comparis
on of
Lan
d Su
rface T
e
mperature Mapp
ing
Using Mo
di
s
and Aster Ima
ges in S
e
mi-A
rid Area. R
e
m
o
te Sens
ing
D
i
visio
n
, Surve
y
ing a
nd Ge
om
atics Dept.,
F
a
cult
y
of Engin
eeri
ng, Un
i
v
ersi
t
y
of T
ehran, T
ehran,
Iran. T
he Internatio
nal Arch
i
v
es of th
e
Photogrammet
r
y
,
Remote Sensing
and Spatial Information
Sciences. Vol. XXX
VII. Part B8. Beijin
g
200
8
[10]
Liu
H, Z
han
g S
hen
gla
ng Z
.
La
nd Surfac
e T
e
mperatur
e Retr
ieval
from
the
Medi
um Res
o
l
u
tion S
pectra
l
Imager MERSI
T
hermal Data.
T
E
LKOMNIKA Indo
nesi
an J
o
urna
l of El
ectri
c
al En
gi
neer
in
g
. 20
14; 1
2
:
728
7-72
98
[11]
Lia
ng S, Li X, W
ang J. Advanc
ed Re
mote
Sensi
n
g
:
T
e
rrestrial Informatio
n
Extraction an
d
Appl
icatio
ns; Elsevier Sci
enc
e
:
Amsterdam, T
he Netherlan
d
s. 2012.
[12]
Z
hang Z
,
He
G. Generatio
n
of Lan
dsat su
rface temper
ature
pr
oduct for
Chi
na, 20
00
–
201
0.
Int. J.
Rem
o
te Sens
. 201
3; 34: 736
9
–73
75.
[13]
Sobri
no JA,
Li
Z
L
, Stoll MP, B
e
cker F
.
Multi-
c
han
nel
a
nd m
u
lti-a
ngl
e a
l
g
o
ri
thm for estimat
i
ng s
ea
an
d
lan
d
surfac
e te
mperatur
e
w
i
t
h
AT
SR data.
In
ternatio
nal
Jo
u
r
nal
of R
e
mote
Sens
ing
.
19
96
; 17: 2
0
8
9
-
211
4.
[14]
Xi
n W
,
Xi
ao
b
o
D, She
ngl
a
n
Z
.
Retrievin
g
Atmosph
e
ric
Precipit
abl
e
W
a
ter Vapor
Using Artific
i
al
Neur
al Net
w
or
k
Appr
oac
h.
T
E
LKOMNIKA Indo
nesi
a
n
Jo
u
r
nal
of E
l
ectric
al E
ngi
ne
erin
g
. 201
3; 1
1
:
717
4-71
81.
[15]
Qin Z
,
Karn
iel
i
A, Berl
iner
P.
A mon
o
-
w
i
ndo
w
al
gor
ithm for
retrievi
ng
la
nd
surface
temp
e
r
ature fro
m
Lan
dsat T
M
da
ta an
d its a
p
p
l
i
c
ation
t
o
the Is
rael-E
g
y
pt b
o
r
der re
gi
on.
Int. J. Rem
o
te Sensing
. 20
01;
22(1
8
): 371
9–3
746.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 221 –
228
228
[16]
Jimen
e
z-Mun
o
z
JC, Sobr
in
o
JA, Skokov
ic
D,
Mattar C,
Cristob
a
l J.
Lan
d surfac
e
temperatur
e
retrieval m
e
tho
d
s from Lan
ds
at-8
thermal
in
frared sens
or
data.
IEEE Geosci. Rem
o
te
Sens. Lett
.
201
4; 11: 184
0
–18
43.
[17]
Skokovic D, So
brin
o JA, Jime
nez-Mu
noz JC,
So
ria G, Julie
n Y, Mattar C, and Cr
istob
a
l J
.
Calibr
a
tio
n
and
Va
lid
atio
n
of La
nd
Surfac
e T
e
mperatur
e
for
L
a
n
d
sat
8
– T
I
RS Sensor
. Lan
d
pro
duct
Val
i
dati
o
n
and Evo
l
uti
on, ESA/ESRIN Frascati (Ital
y
). 2
014: 6-9.
[18]
Che
ng KS, Su
YF
, Kuo FT
, H
ung W
C
, Ch
ia
ng JL
. Assessi
ng the
effect of land c
o
ver ch
ang
es on
a
i
r
temperatur
e us
ing rem
o
te sen
s
ing im
ages
– a pil
o
t stud
y
in
norther
n T
a
i
w
a
n
.
Lan
dscap
e
Urba
n
Pla
n
.
200
8; 86: 85–
9
6
.
[19]
Lo
CP, Quattro
c
hi D,
Luv
all
J.
App
licati
o
n
of
hig
h
-reso
lutio
n
therma
l i
n
frare
d
rem
o
te se
nsi
ng
an
d GIS
to assess the u
r
ban h
eat isl
a
n
d
effect.
Internation
a
l Jo
urna
l of Remote Se
n
s
ing
. 19
97; 18:
287
–3
04.
[20]
W
eng Q. A re
mote sens
in
g-
GIS evalu
a
tion
of ur
b
an
e
x
p
a
n
sio
n
a
nd
its i
m
pact on
surfa
c
e temper
ature
in the Z
huj
ia
ng
Delta, Chi
na.
Internati
o
n
a
l Jo
urna
l of Re
mot
e
Sensi
n
g
. 20
0
1
; 22:19
99–
20
14.
[21]
Nicho
l JE. A
GIS-Based Ap
proac
h to M
i
croclim
at
e Mo
nit
o
rin
g
i
n
Si
nga
pore
’
s Hi
gh-
Ri
se Ho
usi
n
g
Estates.
Photo
g
ra
mmetric En
gin
eeri
ng a
nd
Re
mote Se
nsi
n
g
. 199
4; 60: 1
225-
123
2.
[22]
EPA. 2009. Heat Islan
d
Comp
end
ium:
Ur
ban H
eat
Island Basi
cs. Availab
l
e
onli
ne at
:
http://
w
w
w
.
e
p
a
.
gov/heat-is
l
a
n
d
s/heat-is
l
an
d-
Co
mp
end
ium/
Urba
nHe
a
tIsla
ndBas
ics.pdf
(Accesse
d
Janu
ar
y
3
rd
,
20
16)
[23]
Mostovo
y
GV,
King
RL,
Red
d
y
K
R
, Kaka
ni
V
G,
F
ilippov
a M
G. Statistical e
s
timation
of d
a
i
l
y
ma
xim
u
m
and m
i
nim
u
m
air temp
eratur
es from MODIS
LST
data ov
er the state
of
Mississi
ppi.
G
I
Sci. Re
mot
e
Sens
. 200
6; 43
: 78–11
0.
[24]
Prihodko
L, Go
w
a
rd SN. Esti
matio
n
of a
i
r temper
ature fro
m
re
motel
y
s
e
nsed s
u
rface
o
b
servati
ons
.
Rem
o
te Sens. Environ
. 199
7; 60: 335
–3
46.
[25]
Ung
e
r J, Gal T
,
Rakonczai
J, Mucsi L, Sz
atmari J, T
obak J, Leeu
w
e
n
B, and F
i
al
a
K. 2009. Ai
r
temperatur
e ve
rsus surface
te
mperatur
e i
n
u
r
ban
env
ironm
ent. T
he seven
t
h intern
atio
nal
confere
n
c
e
on urb
an clim
a
t
e. Yokoham
a, Japa
n.
[26]
Li Z, Guo X, Di
xo
n P, He Y. Applic
ab
ilit
y of L
and Surfac
e
Temper
ature (L
ST
) estimates from AVHRR
satellit
e imag
e compos
ites in
norther
n Ca
na
da.
Prairi
e Per
s
pectives
. 20
0
8
; 11: 119.
Evaluation Warning : The document was created with Spire.PDF for Python.