TELKOM
NIKA
, Vol.12, No
.2, June 20
14
, pp. 475~4
8
4
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v12i2.1982
475
Re
cei
v
ed
Jan
uary 8, 2014;
Re
vised Ma
rch 26, 2014; A
c
cepted Ma
y
11, 2014
Weed Control Decision Support System Based on
Precision Agriculture Approach
Rizky
Muly
a
Sampurno*
1
, Kudang
Bor
o
Seminar
2
, Yuli
Suharnoto
3
1
Information T
e
chn
o
lo
g
y
for Natura
l Reso
ur
ces Mana
gem
ent, Bogor Agr
i
cultura
l
Univ
er
sit
y
, BIOT
ROP
Camp
us, Jl. Ra
ya T
a
jur Km.
6, Bogor, Indo
nesi
a
, Ph./F
ax: +
62 251 3
560
77/38
14
16
2
Department o
f
Mechanic
a
l a
nd Bios
ystem
Engi
neer
in
g,
Bogor Agr
i
cultur
al Univ
ersit
y
, D
r
amag
a Camp
us,
Bogor, Ind
ones
ia, Ph./F
ax: +
62 251 8
6
2
3
0
2
6
3
Department o
f
Civil an
d Envi
ronme
n
tal En
gi
neer
ing,
Bo
gor
Agricultur
al U
n
iversit
y
, Dram
aga C
a
mpus,
Bogor, Ind
ones
ia, Ph./F
ax: +
62 251 8
6
2
7
2
2
5
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: rizk
y
m
s@
gm
ail.com
1
, ksemi
nar@a
pps.i
pb.
ac.id
2
,
suhar
noto@
ap
ps.ipb.ac.i
d
3
A
b
st
r
a
ct
Herbici
des h
a
v
e bee
n w
i
del
y used for w
eed contro
l in
mo
der
n agric
ul
ture. How
e
ver
the use o
f
herb
i
cid
e
s is p
o
tentia
ly intro
d
u
cin
g
ne
gativ
e impact to
the
e
n
viro
nment d
u
e
to excessiv
e
use of h
e
rbic
id
es.
Based
on
pr
eci
s
ion
agr
icultur
e
princ
i
pl
es, u
n
i
que
an
d
precis
e treat
me
nt of
herb
i
cid
e
su
pp
l
y
for a
partic
u
l
a
r
area
for cro
p
p
r
oducti
on
must be
perfor
m
ed.
T
he o
b
jectiv
e
o
f
this res
earch
i
s
to d
e
vel
o
p
a
decisi
o
n
su
ppo
rt
system (DSS) f
o
r sche
dul
lin
g
of w
eed spray
i
ng a
nd for se
le
ct
ing the
prop
e
r
no
z
z
l
e
s
i
z
e
of
the spray
e
rs that
introd
uce mi
ni
mu
m ne
gativ
e impact
to
th
e envir
on
me
n
t. T
he
ma
in s
e
t of
data r
equ
ire
d
for our
prop
os
ed
system i
n
cl
ude
s the set of 1
0
year
s w
eather
data ser
i
es ac
q
u
ire
d
fro
m
re
mote sens
in
g (N
OAA and TRM
M)
and
a set
of ve
getatio
n i
ndex
from MODIS E
V
I. T
he w
eat
her data
set is
u
t
ili
z
e
d
to d
e
ter
m
i
ne th
e p
l
anti
n
g
time
peri
od of
pad
dy crop a
n
d
to det
er
mi
ne
the prop
er si
ze of the spray
e
rs for w
eed sprayi
ng. Our D
SS
prototype
has
bee
n i
m
p
l
e
m
e
n
ted a
nd te
sted with real dat
a
set in Jon
ggo
l
d
i
stri
ct, We
st
Java, Indonesia.
The impl
e
m
ent
ation, testin
g result
s, and futu
re enh
anc
e
m
e
n
t of our system are
discuss
e
d
in this pa
per.
Ke
y
w
ords
: DSS, precision
agriculture, we
ed control, herbicides,
spray
drift, weather pattern, tem
p
oral
data
1. Introduc
tion
Wee
d
s a
r
e a
serio
u
s p
r
o
b
l
em for agri
c
ultural cro
p
. They rob ma
in cro
p
s of
sunlight,
water an
d n
u
t
rient cau
s
ing
pro
d
u
c
tion l
o
sse
s
b
o
th in
quantity an
d
quality. Lo
sses
due to
we
ed
were fo
r
wh
eat (9.8%), rice
(10.8%
), maize
(13
%
), so
rgh
u
m
(1
7.8%
), p
o
t
atoes
(4%
)
and
grou
ndn
ut (1
1.8%). Even, an un
cont
rol
l
ed weed
ca
n de
cre
a
se yield until 20
-80%. He
rbi
c
ide
s
are the do
min
ant tool use
d
for wee
d
co
ntrol in mod
e
rn
agri
c
ultu
re [1]-[4].
Although he
rbicid
e has p
o
s
itive benefit in killing
the target weed
s, it
potentialy becom
es
negative impact if som
e
remains in the air and
drift. Spray drift from
herbi
c
ide can
cause
crop
prote
c
tion
ch
emical
s to b
e
dep
osite
d
in
unde
sirable
a
r
ea
s [5]. It h
a
s
se
riou
s
co
n
s
eq
uen
ce
s
such
as da
mage t
o
sen
s
itive a
d
joining
crop
s, dama
ge
to
susce
p
tible
off-target a
r
e
a
s, environm
ental
contami
natio
n, illegal
he
rbicid
e resi
du
es, lo
we
r yiel
d re
sult
s, an
d he
alth ri
sks to
animal
s
and
peopl
e [6]-[10
]
.
Spray drift co
ntinue
s to be
a majo
r pro
b
lem in a
pply
i
ng he
rbi
c
ide
s
. Fa
ctors th
at cau
s
e
drift are un
suitable weat
her
conditio
n
s
and
sp
ray
e
r setup [11
]. Drift can happ
en du
e
to
unsuitable
weather. It potentially occurred eve
r
y
time whe
n
sp
ray
e
r turn
ed on.
The kn
owl
e
d
ge
of we
athe
r
condition
will
help fa
rme
r
a
nd d
e
ci
sio
n
make
r to
d
e
cide the
ap
pro
p
riate
technol
ogy
and method
f
o
r era
d
icating
we
ed,
pl
an, and effectiv
el
y execute
spray appli
c
atio
n
s
to
avoid
spray
drift and othe
r potential wa
ste.
The p
r
og
re
ss of informatio
n techn
o
logy
has b
een
ap
plied wi
dely in agri
c
ultu
re
su
ch a
s
pre
c
isi
on a
g
ri
culture [12],[13]. A weed
co
ntrol me
th
od i
n
pre
c
i
s
ion
a
g
riculture u
s
ing multi-agen
ts
based
has
been
develo
ped in
[1]. The m
e
thod
wa
s a
sup
e
rviso
r
y
syst
em to d
e
termine
techn
o
logy
a
nd liq
uid a
ppl
icator capa
cit
y
and
co
nt
roll
ing a
gent
s. T
hat sy
stem h
a
s t
w
o fu
ncti
ons,
con
s
ultatio
n
functio
n
before sp
raying (o
ff farm
) and spray co
ntrolli
ng by multi-in
telligent age
n
t
s
(on farm)
which a
pplie
d
for gro
undn
ut farming.
De
cisi
on ma
king m
e
thod
was
co
nsid
ering
influen
ce
fa
ctors on wee
d
control
a
c
tivity
such
a
s
crop, we
ed, he
rbici
de, weat
her, ap
plication
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 475 – 48
4
476
time and sp
rayer tech
nolo
g
y. While ag
ents were fo
r:
image acq
u
i
s
ition, filterin
g, crop d
e
tection,
determi
nation
we
ed
den
sit
y
, and d
e
termination
he
rb
icide
do
sa
ge.
The
weathe
r co
ndition
s
b
o
th
spatially
and
tempo
r
ally
have n
o
t stu
d
ied m
o
re
in
previo
us re
sea
r
ch. Relat
i
onship
with
this
resea
r
ch i
s
th
e sy
stem
buil
t
in [1] ne
ed
s to be
imp
r
ov
ed in
kno
w
le
dge th
at rel
a
ted
with
weat
her
along with
sp
ray sched
ulin
g to make a deci
s
io
n t
hat environ
ment friendly. Spra
y sched
ule of
a
cro
p
studi
ed throu
gh veget
ation index
d
e
rived from M
O
DIS satellite
[14].
The sp
atial and tempo
r
a
l
variability weath
e
r con
d
itions a
r
e i
m
porta
nt so
urces fo
r
agri
c
ultu
ral a
c
tivities su
ch
spray appli
c
a
t
ion. Integrati
on meteo
r
olo
g
ical
satellite
with Num
e
ri
cal
Weath
e
r Pre
d
iction
(NWP
) p
r
od
uct i
s
promi
s
in
g in
find timely
weather vari
ab
les
as inp
u
t
for
deci
s
io
n making to resolve probl
em
s in spray appl
ication e
s
p
e
cially for area
which sp
arse
coverage of weather stations
[15
],[16]. However, the availability
of data in
real-time i
s
still
difficult to a
c
hieve.
The
Tro
p
ical
Ra
infall
Mea
s
u
r
ing Mi
ssi
on
(TRMM) dat
a is capa
ble
of
providin
g dail
y
rainfall. NWP produ
cts from
NCEP/NOAA such
as 2 m te
mp
eratu
r
e, wi
nd
, and
r
e
la
tive
hu
mid
i
ty (
R
H
)
a
r
e u
s
ed
as
o
t
he
r
inp
u
t. Moreover
Data f
r
om expe
rien
ce co
uld
be u
s
e
d
for
scheduli
n
g an
d b
e
com
e
de
ci
sion
suppo
rt for p
r
epari
ng to
ols and
ma
chin
ery bef
ore
spray
appli
c
ation
co
ndu
cted [14].
The o
b
je
ctive of this re
search i
s
to
develop a
d
e
ci
sion su
pp
ort
sy
stem (DSS)
fo
r
sched
ulling o
f
weed spraying and fo
r sele
cting the
prope
r no
zzle size of the sp
rayers that
introdu
ce min
i
mum negativ
e impact to the enviro
n
me
nt. To optimize spray sche
duling, we lo
ok
for suita
b
le weather
co
nditi
on du
ring
spray applicat
io
n time. Spray application time is ide
n
tified
throug
h crop
phen
ology which d
e
rived
by MODIS EVI.
2. Method
The re
se
arch
was
con
d
u
c
ted by usin
g remote sen
s
in
g approa
ch. It is useful to
give a
better u
nde
rstanding
abo
u
t
the earth’
s
phen
omen
o
n
[17],[18]. In this
s
t
udy, remote
s
ens
ing
use
d
to
stud
y index of v
egetation
an
d we
at
he
r
condition.
Cro
p
phe
nolo
g
y whi
c
h i
denti
f
ied
throug
h veg
e
t
ation index [
19] u
s
ed to
kno
w
sp
ray
appli
c
ation ti
me. Re
se
arch condu
cted
on
paddy plantat
ion. Remote
sen
s
in
g tech
nology ha
s capability to reco
gni
ze the
phases of pl
ant
gro
w
th th
rou
gh
study of v
egetation i
ndi
ce
s fro
m
pla
n
ting to h
a
rv
est [20]. Th
e
gro
w
th
pha
se of
the study focuse
d on the planting
ph
ase. Planting phase use
d
to estimate the
interval time of
pre
plantin
g t
o
po
st e
m
erg
ence.
Th
en, t
hat inte
rval
consi
dered
as
the time fo
r
spraying.
Wea
t
her
variable
s
duri
ng
spraying t
i
me are
studi
ed in
ord
e
r t
o
optimi
z
e th
e we
ed
co
ntrol and
minimi
ze
negative imp
a
ct to the environm
ent.
Figure 1. Gen
e
ral fram
ework of this study
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Weed Control
Deci
sion Support System
Bas
ed on Preci
s
ion .... (Ri
z
ky Mul
y
a Sam
purno)
477
2.1. Stud
y
Location
This stu
d
y is
an exami
nati
on of
crop
an
d w
eathe
r
co
ndition
s in
ri
ce field
whi
c
h
locate
d
in Jon
ggol, B
ogor,
We
st Java, Indone
si
a (upp
er left corne
r
: 6
o
25’
S 106
o
7’E; lower
right corner
6
o
36’S 10
7
o
8
’
E) an
d h
a
s an
are
a
of
135.6
5
km
2
(Fig
ure
2
)
.
About 64.3
% of Jo
ngg
o
l
is
agri
c
ultu
ral a
r
ea, with lan
d
use
as follo
ws: pad
dy field, mixed gard
ens, a
nd pla
n
tations. Pad
d
y
field covers a
bout 5
1
.3
km
2
or
37.8 %
of the total
a
r
ea. A
c
cordi
n
g [21], Jong
g
o
l is the l
a
rg
est
prod
ucer
of rice eve
r
y ye
ar in Bo
gor,
so it is
o
ften
referre
d
to a
s
the
cent
ral
of rice in Bo
gor
distri
ct. We
were
not st
udie
d
wh
ole
are
a
of Jon
ggol.
We dete
r
mine
d
several p
add
y rice
fields fo
r
sampl
e
s
which
have area about >
5
00 m
2
. It was
rel
a
ted
with the
highe
st
spati
a
l re
sol
u
tion
of
each satellite
data. These
fields we
re
pre
s
ente
d
by several pixe
ls of MODIS image while all
pixels con
s
id
ered a
s
on
e g
r
id cove
ra
ge
of NOAA and
TRMM data.
Figure 2. Location of study
, Jongg
ol
dist
rict, We
st Jav
a
, Indone
sia.
2.2. Satellite Data
2.2.1. MODIS EVI
The
MO
DIS prod
uct use
d
in
this study is
t
he Ve
geta
t
ion Indices
(VI) Com
p
o
s
ite 16
-day
Global 50
0 m SIN Grid V005 or MO
D13A1 p
r
od
u
c
t, which p
r
o
v
ided the ne
eded vegetat
ion
phen
ology d
a
ta. In ad
dition, the
pro
d
u
ct h
ad
al
re
ady be
en
systemati
c
ally
corre
c
ted
for the
effects
of ga
seo
u
s an
d a
e
ro
sol
scatte
ring.
T
he M
O
DIS EVI is embe
dde
d i
n
the
MO
D1
3A1
prod
uct. T
he
MODIS L
and
Di
sci
pline
Group
(MO
D
LA
ND 201
0) de
veloped th
e
EVI for use
with
MODIS data followin
g
this
equatio
n:
(1)
whe
r
e,
ρ
∗
nir
,
ρ
∗
red
and
ρ
∗
blue
are the re
mote se
nsi
n
g
reflecta
nces in the nea
r-i
nfrared, re
d
and
blue, respe
c
tively,
L
is
a
soil adj
ustme
n
t
factor an
d
C
1
and
C
2
de
scribe t
he
use o
f
the blu
e
b
a
n
d
in co
rrectio
n
of the re
d b
a
nd for
atmo
spher
i
c
aeroso
l
scatterin
g
. T
he coefficie
n
ts,
C
1
,
C
2
and
L
,
are em
piri
call
y determine
d
as 6.
0, 7.5 and 1.0, re
sp
ectively.
G
is a gain factor set to 2.5. The
EVI data were develope
d in the above form (Equ
at
ion (1)) in ord
e
r
to optimize
the vegetation
sign
al with i
m
prove
d
se
n
s
itivity in high bioma
s
s re
gion
s. The E
V
I also mini
mize
s atmo
spheri
c
influen
ce
s
wi
th the ‘
aerosol resi
stan
ce’
term
wh
ich use
s
th
e blu
e
ba
nd
to correct ae
ro
so
ls
influen
ce in b
and re
d [22].
In this stu
d
y we u
s
e
d
the
MODIS EVI d
a
ta set
s
which we
re a
c
q
u
ired from
Ja
nu
ary 201
0
to Decembe
r
2012 a
nd cap
t
ured 6
9
time seri
es
with the interval ti
me 16 day
s. The stu
d
y area is
covered by o
n
ly one M
O
DIS tile: h28v09. MODIS
E
V
I data we
re
extracted
fro
m
the MO
DIS VI
prod
uct (MO
D
13A1
)
u
s
in
g the MODI
S Reproje
c
tion Tool
(US
G
S LP DAAC 200
9b
) an
d the
L
A
N
DUS
E
For
e
st
Set
t
lem
ent
Mi
xe
d G
a
r
d
en
Plan
tati
ons
Ric
e
fie
l
d
W
a
te
r
b
ody
Bar
e
land
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 475 – 48
4
478
sele
cted
out
put form
at wa
s G
eoTIF
F and
c
oord
i
nate system
wa
s
g
eog
raphi
c coo
r
di
nate
system
s on d
a
tum Wo
rld
Geod
etic System of 1984.
2.2.2. NOAA
NCEP/NCAR Reanaly
s
is
1
Weath
e
r
dat
a obtain
ed from National
Oce
ani
c & A
t
mosph
e
ri
c A
d
minist
ration
(NOAA)
that issu
ed b
y
The Missio
n of the ESRL Physical Scien
c
e
s
Divi
sion (PSD).
The NCEP/NCAR
Rea
nalysi
s
1
proje
c
t is usin
g a state-of-th
e
-a
rt analysi
s
/fore
cast system t
o
perform data
assimilation
u
s
ing
pa
st d
a
ta from
19
48
to the
pr
esent. It ha
s temp
oral resolution
of 4-time
s dai
ly,
daily and mo
nthly values for 194
8/01/0
1
to present
whi
c
h ha
s gri
d
global of spatial re
soluti
on.
Weath
e
r data
used
in thi
s
study
we
re
a
s
follo
ws
:
ai
r temperature,
relative humi
d
ity,
u-wi
nd a
nd
v-wind. E
a
ch
variable
ha
s
near the
su
rf
ace
(.
sig 9
95
level) d
a
taset on
a 2.5
ᵒ
×
2.5
ᵒ
grid in daily
resolution. T
he pro
d
u
c
t (.sig 995 level
)
, air tem
perat
ure, rel
a
tive humidity and
wind are ab
ove
surfa
c
e
exa
c
tly 2 m, 2
m and
10
m, re
spe
c
tive
ly. For thi
s
study
we
u
s
ed
four ki
n
d
s
of
NCEP/NCAR
Re
analy
s
is 1
of
d
a
ta set
s
whi
c
h we
re
acquired
fro
m
Janua
ry 2
003 to
Dece
mber
2012 a
nd coll
ected 3
650 ti
me se
rie
s
for
each paramet
er with daily i
n
terval time.
2.2.3. TRMM
3B4
2
The rainfall product
from
T
R
MM satellite
is co
m
b
inati
on of the Precipitation
Radar
(PR),
TRMM
Micro
w
ave Ima
g
e
(TMI), a
nd V
i
sible
an
d Inf
r
ared S
c
an
n
e
r
(VIRS) [2
3]. TRMM
3
B
42
daily data i
s
t
he data
level
3 the results
of dat
a p
r
o
c
e
ssi
ng 1B
01, 2
A
12, 3B31, 3
A
44 an
d Glo
bal
pre
c
ipitation i
ndex (GPI).
The final 3B
42 prec
i
p
itation (in mm/h
r
) estimate
s
have a 3-ho
urly
temporal re
solution and
a
0.25
ᵒ
x 0.25
ᵒ
spatial
re
so
lution. Spatia
l cove
ra
ge
e
x
tends f
r
om
50
degree
s sout
h to 50 deg
rees n
o
rth lat
i
tude. T
he d
a
ily accumul
a
ted (b
egin
n
ing at 00Z a
n
d
endin
g
at 2
1
Z
; unit: mm)
rainfall
pro
d
u
c
t is de
rived
from thi
s
3
-
h
ourly p
r
od
uct
.
The d
a
ta a
r
e
store
d
in flat
binary. In this study we
used th
is p
r
o
d
u
c
t whi
c
h
we
re acquired from Ja
nua
ry 2003
to Decembe
r
2012 a
nd coll
ected 3
650 ti
me se
rie
s
wit
h
daily interval time.
2.3. Data Pro
cessing
MODIS EVI data obtain
ed in Geo
T
IFF format. EVI was extracte
d u
s
ing
MODIS
conversi
on toolkit or MODIS reproj
ection tool that provided by
NASA. MODIS has system
atically
corre
c
ted
but
not geo
metri
c
ally corre
c
te
d so th
at
ne
cessary
re
ctified man
ually. The rectification
wa
s d
one
u
s
e the
co
rrect
ed b
e
a
c
h ve
ctor [24
]. Wh
ile the
we
ath
e
r d
a
ta o
b
tai
ned i
n
n
e
tCDF
format and
g
eometri
cally
corre
c
ted. Cli
m
ate dat
a o
perato
r
(cd
o
) used to m
a
nipulate n
e
tCDF
data format. For exampl
e, it uses to comput
e wi
nd
spee
d and
dire
ction whi
c
h de
rived from
northe
r
n
and
so
uthern
wind. The
next
step
wa
s
l
a
yer stackin
g
. The
M
O
DIS and weath
e
r data
were
seq
uent
ially stacke
d
to pro
d
u
c
e th
e time-se
r
ie
s data
set a
n
d
then
clipp
e
d
to cove
r
stu
d
y
area
com
p
o
s
i
t
e perio
d. Th
e sta
cke
d dat
a we
re eval
u
a
ted to get te
mporal pattern from the ti
me
s
e
ries
data [19].
2.4. Data
An
aly
s
is
Several point
s
of study area were
ta
ke
n
that
rep
r
e
s
ented l
o
catio
n
of p
addy fi
eld. The
EVI of these points were
analyze
d
time-seri
e
s
every 16 day
durin
g thre
e
years. Wea
t
her
con
d
ition of
a
pixel weathe
r dat
a
where
points l
o
cate
d wa
s
co
nsi
d
ered
a
s
weat
her
co
ndition
of
all point of
st
udy area b
e
cause
it
ha
s coarse sp
atial resoluti
on.
Weather conditi
on was analy
z
ed
time-serie
s d
a
ily during 1
0
years ob
se
rv
ation.
2.5. Estimation of We
ed
Con
t
rol Time
Normally, we
ed control i
n
pre
c
isi
on a
g
ri
cultur
e is
pe
rforme
d twi
c
e,
i.e. pre-planti
ng an
d
post-eme
r
ge
nce [1]. The
s
e time co
ul
d be estimat
ed wh
en the
planting time kno
w
n. Ri
ce
phen
ology from plantin
g
to harve
sting
rep
r
e
s
ente
d
trough EVI,
the plantin
g
time used
as
referen
c
e
to
estimate
time
for we
ed co
ntrol. Da
ily weather conditi
on
d
u
rin
g
pre
-
plantin
g
to
p
o
st
emergen
ce i
n
terval
wa
s a
nalyze
d
. Pre
-
planti
ng
esti
mated a
mon
t
h before pla
n
ting mo
nth
and
post-eme
r
ge
nce
e
s
timate
d a m
onth aft
e
r pl
anting
m
onth. The
n
th
e interval
fro
m
pre-h
a
rve
s
t to
post-eme
r
ge
nce
con
s
id
ered as
wee
d
control time.
2.6. Dev
e
lop
m
ent Ap
plication to
Dete
rmine Nozzl
e
Spra
y
e
r
This
appli
c
ati
on
wa
s d
e
vel
oped
with
ob
jective to
det
ermin
e
n
o
zzl
e
si
ze
for
sp
rayer. It
expecte
d cou
l
d minimi
ze d
r
ift on we
ed
control. Thi
s
simple
appli
c
ation could b
e
com
b
ine
d
with
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Weed Control
Deci
sion Support System
Bas
ed on Preci
s
ion .... (Ri
z
ky Mul
y
a Sam
purno)
479
agent of we
e
d
control system [1] to improve p
r
e
c
isi
on of spray appli
c
ation.
The rul
e
s (IF
–
T
H
EN)
to
dec
id
e
w
h
ic
h
no
zz
le
s
p
r
a
ye
r w
e
r
e
ac
qu
ired from
previous
re
se
arch
es [1, 2
5
] wh
ich
interp
reted i
n
to deci
s
io
n tree. Thi
s
a
p
p
licatio
n was desi
gne
d a
nd devel
ope
d usi
ng
syst
em
developm
ent
life cycle (SDLC) [26]. SDLC i
s
the
traditional
methodol
ogy
used to de
velop,
maintain, and
repla
c
e information syste
m
. The diffe
rent phases of
the SDLC are: investigation
,
analysi
s
, de
si
gn, impleme
n
t
ation and ma
intenan
ce.
Some rule
s o
f
decisi
on-ma
king b
a
sed o
n
weath
e
r co
ndition
s we
re
[1]:
1)
Rule
s for pa
rameter of Wi
nd Speed
(km/hr)
-
If WS < 2, then do not sp
ra
y
-
If
2
≤
WS
≤
3, then sp
rayin
g
with air a
ssi
sts, with a me
dium droplet
size
-
If
4
≤
WS
≤
6, then use a fi
ne dro
p
let si
ze
-
If
7
≤
WS
≤
10, then use a
medium d
r
opl
et size
-
If
11
≤
WS
≤
14, then use a coa
r
se drop
let size
-
If
15
≤
WS
≤
20, then sp
ra
ying with air a
ssi
st
s, with
coarse d
r
opl
et size
-
If WS > 20, then do not spray
2)
Rule
s for pa
rameter of Air
Tempe
r
atu
r
e
(
˚
C)
-
If T < 15, then spraying wit
h
dropl
ets fin
e
-
If
15
≤
T < 20
, then spraying with dro
p
let
medium
-
If
20
≤
T < 25
, then spraying with co
arse
droplet
s
-
If T > 25, then spraying wit
h
air
assi
sts,
with co
arse d
r
oplet
3)
Rules
for parameter of Relative Humidit
y
(%)
-
If RH < 40, then sp
raying
with air a
ssi
st
s, with co
arse
droplet si
ze
-
If
40
≤
RH <
60, then sp
ra
ying with med
i
um dro
p
let
-
If
60
≤
RH <
80, then sp
ra
ying with fine dropl
ets
-
If RH > 80, then sp
raying
with air a
ssi
st
s, with mediu
m
dropl
et size
3. Results a
nd Discu
ssi
on
Wee
d
control
method
wa
s develop
ed b
y
utilizat
ion o
f
weathe
r dat
a that obtain
ed fro
m
remote
se
nsi
ng. It becam
e e
sse
ntial f
o
r a
r
ea
wh
ich sparse
coverag
e
of
m
e
teorol
ogy
station
s
and re
quires
a long time serie
s
data. Ti
me se
rie
s
dat
a in previou
s
years u
s
e
d
to optimize spray
sched
uling b
y
lookin
g
for suitabl
e
weat
her co
ndition and
to pre
p
a
r
e
the sprayin
g
ma
chin
ery and
equipm
ent for weed
control
that minimize the negativ
e impact of h
e
rbi
c
ide
s
to the enviro
n
me
nt.
Figure 3. Utilization of weat
her data for
weed control
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 475 – 48
4
480
3.1. Spra
y
Application Ti
me
In pre
c
isi
on
farming,
wee
d
co
ntrol i
s
done t
w
o t
i
mes i.e. p
r
e-pla
n
ting a
nd po
st-
emergen
ce.
To kno
w
the
s
e time
s, time for pl
anti
ng is i
dentifi
ed first. Paddy planting t
i
me
identified through multi-te
mporal analy
s
is of
veget
ation index. Figure 4 sh
owe
d
pattern
of
enha
nced ve
getation i
nde
x (EVI) of
pad
dy field in
stu
d
y area
whi
c
h fluctu
ated
d
u
ring
three ye
ars
data o
b
servat
ion. EVI clea
rly sho
w
s the
annu
al v
eget
ation
g
r
o
w
th cycle, rep
r
e
s
enting
inte
nsi
v
e
cro
ppin
g
with
multiple harv
e
sts. Th
e EVI pattern
of pa
ddy has a
n
al
most symm
etrical
bell sha
pe
[14]. The vegetative growt
h
stage corre
l
ated with
the
incre
a
si
ng EVI value until it reache
s the
maximum. Th
e maximum
of EVI value sho
w
s ve
ry g
r
een
col
o
r fro
m
paddy le
afs. It happe
ns in
headi
ng/pani
cle sta
ge [20]
.
Figure 4
sh
o
w
that the
cro
pping
cycl
e o
f
paddy
starte
d abo
ut from
April to
July,
Octob
e
r
to Febru
a
ry, resp
ectively for cycl
e 1 an
d cycle 2.
T
he time for pla
n
ting co
nsi
d
e
r
e
d
as time
wh
ere
cro
ppin
g
cycl
e sta
r
ted. T
he pla
n
ting t
i
mes
we
re A
p
ril a
nd O
c
t
ober.
The
s
e
times
we
re
not
gene
ral time
for pla
n
ting,
but the
s
e tim
e
s ju
st fo
r
th
e point
s of
st
udy area [1
4]. In real
co
nd
ition
the planting
time may earlier o
r
later
than thes
e p
o
ints. As me
ntioned in p
r
evious
se
ctio
n,
planting
pha
se is u
s
e
d
to
estimate th
e
time interval
of pre
-
pla
n
tin
g
to po
st-em
e
rge
n
ce. The
n
,
that interval i
s
con
s
ide
r
e
d
as th
e time
for spraying.
Here the
spraying time
a
pproxim
ated
one
month b
e
fore
and
after th
ese
month
s
i
.
e. spraying t
i
me are Ma
rch to May a
n
d
Septemb
e
r to
Novemb
er for cycle 1 an
d cycle 2 re
spe
c
tifely.
Figure 4. We
ed co
ntrol tim
e
whi
c
h e
s
timated from pla
n
ting time through veg
e
tation index
3.2. Wea
t
her
Patter
n
Figure 5 sho
w
s
weath
e
r p
a
ttern in Jo
ng
gol duri
ng ten
years. Thi
s
informatio
n is
use
d
to
facilitate farm
er o
r
de
cisi
o
n
maker to
know
the we
ather con
d
ition
before apply
i
ng
he
rbi
c
ide
to
prote
c
t the crop. By this in
formation, p
r
oblem
s
on
weed
control such
as d
r
ift a
nd ru
n-off ca
n be
minimized by
pre
pari
ng m
a
chi
nery
and
sp
raye
r ea
rli
e
r b
e
fore
spraying time. E
v
ery pa
ramet
e
r
have o
w
n
ch
ara
c
teri
stic
a
nd ge
ne
rally i
n
sa
me flu
c
tu
ated patte
rn f
o
rm. Dry an
d
wet
sea
s
o
n
are
clea
rly see
n
in weath
e
r pat
tern (Fi
gure 5
)
whi
c
h affect
ed by monso
on.
Gene
rally, du
ring te
n yea
r
s rainfall i
s
hi
gh in ye
ar-en
d
to ea
rly ye
ar
while l
o
w i
n
in mid
-
year. Win
d
speed i
s
fluctu
ates. Win
d
is
high in
yea
r-e
nd to Ja
nua
ry every year a
bout more th
an
10 km/s. Fo
r ten years, m
i
nimum an
d maximum
temperature
s
a
r
e 23.5°
C an
d 30°
C. Rela
tive
humidity de
creased
whe
n
air temp
eratu
r
e in
crea
se
. It is abo
ut 65
– 95%, hig
h
in year-en
d
to
early year, a
nd lowers in
middle year. West
e
r
n wi
nd of sea
s
o
nal mon
s
oo
n
during p
e
rio
d
De
cemb
er to
Jan
uary take
s abu
nda
nt water vapo
r so
rainfall in that
time tends to
high.
Farme
r
or de
cisi
on
maker ca
n u
s
e
p
a
st weath
e
r dat
a to find
o
u
t
the optim
al ti
me for
sched
uling,
prep
ari
ng m
a
chi
nery an
d
sprayer. O
p
timal wee
k
for sp
rayin
g
determi
ne
from
interval time
for
sp
raying
b
o
th in
crop
cy
cle
1 a
nd
cro
p
cy
cle
2. According
to [1]
,
ideal
co
nditi
on
of wind for
spraying is
3.2 –
9.6
km/h and wi
nd m
o
re than
9.6 km/h can cause drift. Somet
i
me
sprayer
nee
d
air a
s
sist te
chnolo
g
y by a
dding
pre
s
su
re to help
he
rbicid
e drop to
target
wee
d
. In
[1], do not
spray
whe
n
th
e rel
a
tive hu
midity less th
an 4
0
% an
d
temperature
s
above
25°
C in
orde
r to
redu
ce
drift
cau
s
e
d
by tem
pera
t
ure i
n
version
s
o
r
evapo
rati
on, al
so i
n
cre
a
se
s th
e ta
rg
et
depo
sition a
n
d
cove
rag
e
. In study area
the tem
perature
can b
e
hi
gher
and lo
wer than 2
5
°
C
but
humidity more than 65%.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Weed Control
Deci
sion Support System
Bas
ed on Preci
s
ion .... (Ri
z
ky Mul
y
a Sam
purno)
481
Figure 5. We
ather patte
rn
of study area
and its ut
ilization for ch
aracterizi
n
g
we
ather conditio
n
durin
g sp
ray
duratio
n
The optimum
wee
k
for sp
raying ca
n be
determi
n
ed by
using crite
r
ia
that
can minimize
spray drift [1]
.
For exam
pl
e (Fig
ure 6
)
, wee
d
c
ontrol can
be
co
nd
ucted i
n
p
r
e-planting
or
after
planting a
nd
post-eme
r
ge
nce. Sp
ray a
pplication
pe
rformed after planting
u
n
til
post-eme
r
ge
nce
(April
– M
a
y). That time
in
cluded
to
critical pe
ri
od
of
weed
com
petition. It is
abo
ut 40
days afte
r
planting.
Spraying
he
rbicid
e for pa
d
d
y is
app
rop
r
i
a
te pe
rform
e
d when
ever
a
fter plantin
g. I
t
is d
u
e
to in ge
neral
time before p
l
anting
wee
d
control
i
s
con
ducte
d by pl
owin
g. Re
ma
ining bi
oma
s
s,
soil a
nd wee
d
are
mixed. Wee
d
on the
top soil m
o
ved to lower l
a
yer of soil. Wee
d
is
hard
to
gro
w
even weed will b
e
died be
cau
s
e t
hey can
not
continue the p
hotosynth
esi
s
process.
(a)
(b)
Figure 6. We
ather conditio
n
durin
g we
e
d
control
interval in both pa
ddy planting
cycle 1 p
e
rio
d
March – May
(a) a
nd pa
dd
y planting cycle 2 perio
d Septembe
r – Novembe
r
(b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 475 – 48
4
482
3.3. Application for
Minimizing Spray
Dr
ift
To find the
ideal
con
d
itio
n for
sp
rayin
g
in ide
a
l tim
e
and
co
nditi
on of e
a
ch
p
a
ram
e
ter
based on ref
e
rence
and previous study
is
still hard
t
o
achieve.
Therefore,
software to determ
ine
nozzle spray
e
r size
consider
to weather cond
ition will hel
p ov
ercome thi
s
limitation. The
prototype inte
rface
of appli
c
ation
to sele
ct the nozzle
sprayer b
a
se
d on we
ather
con
d
ition
s
wh
ile
the kn
owl
edg
e is rep
r
e
s
en
ted troug
h d
e
ci
sion tree
(Figure 7.a).
Rainfall i
s
th
e first p
a
ram
e
ter
whi
c
h de
cide
s do spray or do not spray, becau
se
sp
ray appli
c
atio
n will not con
ducte
d in rai
n
y
day and
he
rb
icide
parti
cle
s
will run off al
ong
with
rain water. Win
d
become
s
se
cond
p
a
ramet
e
r,
followin
g
by tempe
r
ature a
nd humidity. Weath
e
r p
a
rameters can
be inputted
manually o
r
can
use
weath
e
r
data take
n fro
m
the pas
t da
ta which stored in datab
ase.
For exampl
e, this application applie
d usin
g weath
e
r
data of Jo
nggol. Weed
contro
l
perfo
rmed
o
n
no
rmal
weat
her condition
, this
appli
c
a
t
ion will
reco
mmend
the
use
of m
ediu
m
sized no
zzle
(Figu
r
e 7.b
)
. While the
co
ndition in
whi
c
h the large
wind
spe
e
d
s
(15
-
20
km/h)
the
system
will
re
comm
end
the
use fine
n
o
zzle
si
ze. Air
assist te
ch
nol
ogy u
s
e to
a
v
oid d
r
ift due
to
wind
spee
d a
nd tu
rbul
ence
of
wind
direction. Wh
en
a
pplied
an
extreme
we
ather co
ndition
s, f
o
r
example wi
n
d
spee
d > 20
km/h, rainy, wind di
re
ct
io
n that is rapid
l
y changin
g
, the air humi
d
ity is
very low a
n
d
very hi
gh te
mperature
s
, t
he
syst
em
re
comm
end
do
not
sp
ray
(F
igure
7.c).Hi
g
h
wind
speed and wind
di
rection
change
will make a
big drift.
It
occur spray
turbul
ence whi
c
h
resulted pa
rti
c
le
s of sp
ra
ying drop to
non-ta
rget
wee
d
effectively. In
[1], tempe
r
ature
and
humidity clo
s
ely related
to pa
rticle
evaporat
ion. T
he very lo
w of humidity
and ve
ry h
i
gh
temperature will
make spraying
pa
rticles evaporate f
a
st
er in the
air befor
e reachi
ng
the crop,
and the pa
rticle
s that have rea
c
he
d the plant can
n
o
t work effect
ively becau
se it will evaporate
fas
t
er.
(a)
(b)
(c
)
Figure 7. Sprayer droplet selectio
n ba
se
d on we
ather
con
d
ition. Knowle
dge represe
n
tation (a
),
appli
c
ation in
norm
a
l we
ath
e
r co
ndition
(b) appl
i
c
atio
n
in extreme weather
con
d
ition (c)
4. Conclusio
n
s & Future
Direc
t
ions
The
DSS for wee
d
a
ppli
c
ations
ha
s b
een d
e
velop
ed an
d te
ste
d
with
a real
data
set
a
c
qu
ir
ed
fr
om r
e
mo
te
sen
s
ing
d
e
v
ic
es
. T
h
e
prop
ose
d
sy
stem
can
ge
nerate optimal
sp
ray
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Weed Control
Deci
sion Support System
Bas
ed on Preci
s
ion .... (Ri
z
ky Mul
y
a Sam
purno)
483
sched
uling
a
nd recomme
nd the
p
r
ope
r
size of
no
zzle
s u
s
e
d
for sp
ray
appli
c
ation o
n
p
a
d
d
y
cro
p
s b
a
se
d
on th
e
wea
t
her
co
nditio
n
, and
thu
s
intro
d
u
c
ing
minimal
sp
ra
y drift an
d
bad
environ
menta
l
impact.
In our
cu
rren
t implemente
d
prototype
data
a
c
qui
siti
on from
rem
o
te se
nsi
ng
device
s
su
ch
as
NO
AA, TRMM a
nd MO
DIS i
s
ca
rri
ed o
u
t
usin
g
sepa
ra
te appli
c
ation
interfa
c
e
s
. T
he
future e
nha
n
c
eme
n
t is to
build o
ne
app
lication i
n
terf
ace
to a
c
qui
re data
set fro
m
these
sen
s
ing
device
s
, allo
wing th
e e
a
s
y and
faste
r
a
c
qui
sition
of re
quired
data
sets f
r
o
m
more dive
rse
sou
r
ces
of re
mote sensi
n
g
device
s
in
a
n
integr
ated
manne
r. Mo
reover, du
e to
the variabilit
y of
weath
e
r
cha
r
acters
and pl
anting time p
e
riod
s in
different g
eog
ra
phical area
s, the use of f
u
zzy
inferen
c
e
sy
stem for im
pro
v
ing the time
sched
uling
of we
ed
sp
rayi
ng i
s
hig
h
ly recom
m
en
ded
for
the future sy
stem develop
ment.
Referen
ces
[1]
Sola
hud
in
M. Devel
opm
ent
of
w
e
ed
Contr
o
l Met
hod
in
Precisio
n F
a
r
m
ing B
a
se
d o
n
Multi
Ag
en
t
Comp
uting.
Ph
D Thesis
. Bog
o
r: Graduate S
c
hoo
l IPB. 201
3.
[2]
Evans LT
.
Crop Evoluti
on, Ad
aptatio
n an
d Yi
eld
. UK. Camb
ridg
e Univ
ersit
y
Press. 19
96.
[3]
Utami S. Kem
e
limp
a
h
an J
e
n
i
s Gulma
T
anaman
W
o
rte
l
pad
a
Sistem
Pertani
an
Org
anik.
BIOMA
.
200
4; 6: 54-58.
[4]
Harker KN, JT
O’Don
o
van.
W
eed C
ontrol,
W
eed Ma
nag
ement, an
d Int
egrate
d
W
e
e
d
Mana
geme
n
t.
W
eed T
e
chn
o
l
ogy
. 201
3; 27: 1–1
1.
[5]
Nu
y
t
t
ens D, M
De Sch
a
mp
hel
eire, K B
aete
n
s
, E Brusselm
a
n
, D D
e
ke
yser,
P Verb
ov. Drif
t from F
i
eld
Crop S
p
ra
ye
rs
Usin
g a
n
Inte
grated
Appr
oa
ch: Res
u
lts of
a F
i
ve
‐
Y
ear Stud
y
.
T
r
ansa
c
tions of
T
h
e
ASABE
. 2011;
54(2): 40
3-4
0
8
.
[6]
Red
d
y
K
N
, W Ding, RM Z
abloto
w
i
cz, SJ T
homson,
Y
Hua
ng, LJ Krutz. Biolog
ical
respons
es t
o
gl
yp
hosat
e drif
t from aeri
a
l
a
pplic
atio
n i
n
n
on-g
l
yph
o
sate-
r
esistant c
o
rn.
Pest Man
age
ment Sci
ence
.
201
0; 66(1
0
): 1148-
115
4.
[7]
Marrs RHA, J F
r
ost. A microcosm ap
proac
h to the d
e
te
cti
on of the
effects of herbici
de
spra
y drift
i
n
pla
n
t communit
i
es.
Journ
a
l of Enviro
n
m
enta
l
Mana
ge
me
nt.
199
7; 50: 369-
388.
[8]
Kjaer C, M Strandb
erg, M Erlands
en. Metsulfaro
n spra
y
drift red
u
ces fruit
y
i
el
d of Ha
w
t
o
r
n
(Cratae
gus mo
nog
yn
a L.).
Sci
ence of the T
o
t
a
l Envir
o
n
m
e
n
t.
2006; 35
6(1-3
)
: 228-23
4.
[9]
Otto S, L Lazzaro, A Finizio,
G Z
anin.
Estimating ecotoxicological e
ffects of pesticide spray
drift on
nontar
get arthr
opo
ds i
n
fiel
d h
edg
ero
w
s.
E
n
v
i
ron
m
e
n
tal Tox
i
colo
gy a
nd
Ch
emistry.
200
9; 28(4):
8
53-
863.
[10]
Butler EMC, AG Lane, CM O'Sulliv
an, PCH
Miller, CR Gl
a
ss. By
st
and
er
expos
ure to pe
sticide spr
a
y
drift: Ne
w
data for model
developm
ent
and v
a
lidation.
Biosyste
m
s Engi
neer
in
g
. 2
010; 1
07(
3):
162
‐
1
68.
[11]
Wo
lf T
M
.
Meeting v
a
ria
b
le
ap
plicati
on
go
als
w
i
th new
app
l
i
catio
n
tech
nol
ogy
. Agric
u
ltur
e an
d Agr
i
-
F
ood Ca
nad
a, Saskatoo
n
Res
earch C
entre. Paper of
Grain
Researc
h
and
Develo
pme
n
t Corp
oratio
n
.
200
6.
[12]
Cook S, Braml
y
R.
Precisi
o
n
Agriculture: U
s
ing Pa
ddock
Information to Make Crop
pi
n
g
Systems
Internatio
na
lly
Competitiv
e.
Proceed
in
g. Emergi
ng T
e
chno
log
i
es i
n
Agricult
ure: F
r
om ide
a
s to
ado
ptio
n. Bure
au of Reso
urce
Sciences C
onf
er
enc
e, 25 – 2
6
Jul
y
20
00, A
u
strali
a. 200
0.
[13]
Auern
hammer
H. Precisi
on F
a
rming-T
he E
n
vironme
n
tal
C
hall
e
n
ge.
C
o
mputers a
n
d
El
ectronics
in
Agricult
ure
. 20
01; 30: 31-
43.
[14]
Sampur
no RM
, Kudang BS,
Yuli S.
Char
acteri
z
i
n
g
T
e
mpora
l
Dyna
mic
of W
eather Varia
b
il
ity to
Supp
ort Decisi
on Makin
g
on
W
eed Co
ntrol
.
Proceed
ing. 2
nd
Internatio
nal
Confere
n
ce o
n
Adaptiv
e
and Intel
l
i
gent
Agroi
ndustr
y (ICAIA), 16 – 17
September 2
0
13, Bogor, Ind
ones
ia. 20
13; 61-7
1
.
[15]
R
o
eb
e
l
i
n
g R
A
, Va
n Pu
tte
n E, Ge
no
ve
se
G, R
o
se
ma
A. Ap
p
l
i
c
a
t
i
o
n o
f
me
te
osa
t
d
e
r
i
v
ed
meteoro
l
og
ical
informatio
n
for crop
y
i
el
ds pred
ictions i
n
Europ
e
.
Intern
ation
a
l Jo
urna
l
of Remot
e
Sensi
n
g
. 20
04;
25: 538
9-54
01
.
[16]
De Wit A, Bar
u
th B, Boogaard H,
van
Di
ep
en
K, van
Kraa
lin
gen
D, Mic
a
le
F
,
te Roll
er J,
Supit I, va
n
den W
i
j
n
g
aart
R. Using E
R
A-INT
E
RIM
for regi
on
al cr
op
yie
l
d forec
a
sting i
n
Eur
ope.
C
lim
a
t
e
Research.
20
1
0
; 44: 41-5
3
.
[17]
Muna
di K,
Far
dhi
A. Sate
llit
e Imag
er
y
an
d In-situ
D
a
ta
Overla
y
Ap
pr
oach
for Fis
h
er
y Zon
a
tio
n
.
T
E
LKOMNIKA T
e
leco
mmunic
a
tion C
o
mputi
n
g Electron
ics a
nd Co
ntrol
. 20
10; 8(3): 21
7-2
24.
[18]
Manu
hutu R
L
, Rian
a
DH. S
ebu
ah Al
gorit
ma Poho
n Ke
putusa
n
untuk
Klasifikas
i Cit
ra Satelit Aq
u
a
MODIS.
T
E
LKOMNIKA T
e
lecommunic
a
tio
n
Co
mp
uting El
e
c
tronics an
d C
ontrol
. 20
10; 8(
2): 131-1
36.
[19]
Sakamoto T
,
Yokoza
w
a
M,
T
o
ritani H, Shib
a
y
ama M, Ishitsuka N, O
hno H. A crop ph
en
ol
o
g
y
detectio
n
meth
od us
ing
time-
s
eries MODIS
data.
Re
mote Sensi
ng of
En
viron
m
e
n
t
. 2005; 96 (3-4):
366-
374.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 475 – 48
4
484
[20]
Sari DK, Ismullah IH, Wid
y
o N, Sulasdi, Ha
rto AB.
Detecting Rice Ph
en
o
l
og
y in Pa
dd
y
F
i
elds
w
i
t
h
Compl
e
x
Crop
pin
g
Patter
n
U
s
ing
T
i
me Series MODIS Dat
a
, A Case st
u
d
y
of N
o
rthern
Part of W
e
st
Java – Ind
ones
ia.
Journ
a
l Sci
ence IT
B
. 2010
; 42(2):91-1
06.
[21]
BPS-Statistics of Indon
esia. B
ogor R
ege
nc
y
In Figur
es 2
0
0
8
.
Indonesi
a
: Ba
dan Pus
a
t Statistik. 2008.
[22]
Setia
w
a
n
Y, K
uni
hiko
Y, W
i
l
l
i
a
m DP.
Char
a
c
terizin
g
T
e
mporal
Veg
e
tatio
n
D
y
n
a
mics
of
Lan
d Us
e
in
Regi
on
al Sca
l
e
of Java Island,
Indones
ia.
Jou
r
nal of La
nd
us
e Scienc
e.
201
1; 1:1-30.
[23]
Huffman GJ, Adler
RF, Bolvin DT
,
Gu G, Nelken EJ
, Bo
w
m
an KP, Hong Y
,
Stocker EF, Wolf DB. T
he
T
R
MM Multisatellit
e Preci
p
i
t
ation An
al
ysi
s
(T
M
PA): Quasi-Gl
oba
l M
u
lti
y
e
a
r, Com
b
in
ed-Se
nsor
Precipit
ation E
s
timates at F
i
ne Scals.
Journ
a
l of Hydro
m
eteoro
l
ogy.
2
007
; 8:38-55.
[24]
Rusdi
y
a
t
moko
A, Z
uba
id
ah
A.
Ana
lisis
Spektral
Data
Mod
i
s u
n
tuk
Pe
ma
ntau
an
Huta
n/La
han.
Pertemua
n Ilmi
ah T
ahuna
n MAPIN XIV.
SDA. Suraba
ya. 2
005: 1-9.
[25]
T
epper G.
W
eather Esse
ntia
l
for Pestici
de
Appl
icatio
n
. A
u
strali
a. Grain
Rese
arch &
Devel
opm
ent
Corp
oratio
n. 2012.
[26]
O’Brein JA.
Ma
nag
e
m
ent Infor
m
ati
on Syste
m
. Arizona USA.
McGra
w
Hi
ll. 1
999.
Evaluation Warning : The document was created with Spire.PDF for Python.