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.
10
,
No.
3
,
June
2020
,
pp.
3208
~
3215
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
10
i
3
.
p
p3
208
-
32
15
3208
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Estimati
on of TRMM
rai
nfall f
or l
andslid
e occ
u
rren
ce
s
based
on rainf
all thres
hold ana
lysis
No
r
ais
yah Taj
udin
1
,
No
r
suz
il
a
Y
a’ac
ob
2
,
D
arm
awaty M
ohd
Ali
3
,
N
or A
iz
am
Ad
n
an
4
1,2,3
W
ire
le
ss
Co
m
m
unic
at
ion
T
e
chnol
og
y
(W
iCo
T),
Fa
cult
y
of
Elec
tr
ic
a
l Engi
ne
er
ing,
Univer
siti
Te
kno
logi
MA
RA,
Ma
lay
s
ia
4
Facul
t
y
of
Arch
it
e
ct
ure
,
Pl
anni
n
g
and
Surve
y
ing, Unive
rsit
i
T
ekn
ologi
MA
RA,
M
al
a
y
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Sep
26
,
2019
Re
vised
Dec
2
1
,
2019
Accepte
d
Ja
n 8
, 2020
La
ndslid
e
c
an
b
e
tri
gg
ere
d
b
y
in
t
ense
or
pro
longed
rai
nf
al
l
.
Pre
ci
p
it
ation
d
at
a
obta
in
ed
from
ground
-
base
d
obse
rva
ti
on
is ve
r
y
a
cc
ura
te
and
com
m
onl
y
used
to
do
ana
l
y
sis
an
d
la
ndslide
pr
edicti
on
.
How
eve
r,
thi
s
appr
oac
h
is
costly
with
it
s
own
li
m
itat
i
on
due
to
l
ac
k
of
density
of
gr
ound
stat
ion
,
espe
cially
in
m
ounta
in
area.
As
an
al
ternat
iv
e,
sat
ellite
der
iv
ed
rai
nf
al
l
techn
ique
s
have
bec
om
e
m
ore
fa
vora
ble
to
ov
ercom
e
the
se
l
imitations.
Moreov
er,
the
sa
te
l
lit
e
de
rive
d
r
ai
nf
al
l
esti
m
at
ion
n
eeds
to
be
va
li
da
te
d
on
i
ts
accu
racy
a
nd
its
ca
pab
il
i
t
y
to
p
red
ict
la
ndslid
e
which
pr
esum
ably
tri
gg
ere
d
b
y
ra
infall.
Thi
s
paper
pre
s
ent
s
the
inve
st
ig
at
ion
of
using
t
he
TRMM
-
3B42V7
dat
a
in
compari
son
to
the
availa
b
le
r
ai
n
-
gaug
e
da
t
a
in
Ulu
Kela
n
g,
Sela
ngor
.
The
m
onthly
a
ver
age
rai
nf
all,
cumulat
iv
e
r
ainfal
l
and
r
ai
nfa
ll
thr
eshold
ana
l
y
sis
from
19
98
to
2011
is
co
m
par
ed
using
qu
ant
itati
v
e
st
at
ist
i
ca
l
criter
ia
(Pea
rson
cor
r
el
a
ti
on,
bia
s
,
root
m
ea
n
square
err
or,
m
ea
n
diff
erent
and
m
ea
n
)
.
The
resul
ts
fro
m
ana
l
y
s
is
show
ed
tha
t
the
r
e
is
a
significan
t
and
strong
positi
ve
cor
r
elat
ion
bet
wee
n
th
e
TRMM
3B42V7
and
rai
n
gauge
data.
The
thre
shold
d
eri
va
ti
ve
from
the
sate
llit
e
produc
ts
is
lower
tha
n
the
rai
n
gauge
m
ea
sur
ement.
The
find
ing
s
indi
c
a
te
d
th
at
t
he
proposed
m
e
t
hod
ca
n
be
appl
i
ed
using
T
RMM
sate
ll
ite
e
stim
at
es
produc
t
s
to
der
ive
r
ai
nf
al
l
thr
eshold
for
the poss
ibl
e
l
andsli
de
occ
u
rre
nce
.
Ke
yw
or
d
s
:
Lan
ds
li
de
Ra
in
ga
uges
Ra
infall
thre
shold
Sate
ll
it
e rainf
al
l est
i
m
at
es
TRM
M
Copyright
©
202
0
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
:
Nors
uzila
Y
a
’a
cob,
W
i
reless C
omm
un
ic
at
ion
Te
chnolo
gy
(
W
i
CoT)
,
Faculty
of Elec
tric
al
Engineer
ing
,
Un
i
ver
sit
i Te
knol
og
i M
ARA
,
40450 S
hah A
l
a
m
, S
el
ango
r,
Ma
la
ysi
a.
Em
a
il
:
no
rs
uzila
@salam
.u
itm
.
edu.m
y
1.
INTROD
U
CTION
Ra
infall
is
kn
own
as
one
of
the
m
ai
n
facto
r
w
hich
co
ntribute
to
la
ndsli
de
occ
urre
nce,
t
hu
s
ca
us
in
g
a
har
m
fu
l
eve
nt
to
pe
op
le
,
so
ci
et
ie
s
and
econom
ie
s
wo
rld
wide
un
der
cl
i
m
a
te
change
co
ndit
ion
s
[1
-
3]
.
Re
la
ti
on
sh
i
ps
betwee
n
rai
nf
a
ll
and
la
ndsli
de
s
are
fr
e
qu
e
nt
ly
co
m
plex
due
to
i
nconsist
ent
an
d
inc
omplet
e
la
nd
sli
de
recor
ds
,
a
nd
t
he
ava
il
abili
ty
of
rain
fall
data
[4
]
.
In
Ma
la
ysi
a,
the
com
par
ison
of
la
nd
sli
de
tri
ggeri
ng
factor
s
hows
r
ai
nf
al
l
ha
s
a
m
ajor
c
ontrib
ution
to
the
e
ve
nt,
w
hich
is
a
bo
ut
58%
with
a
n
a
ver
a
ge
of
2550
m
m
of
rainfal
l
per
ye
ar
[5
]
.
C
urren
tl
y,
the
pr
e
dicti
on
of
la
ndsli
de
trigg
e
re
d
by
rain
fall
is
base
d
on
exa
m
ining
the
e
m
pirical
relat
ion
sh
i
p
bet
ween
r
ai
nf
al
l
char
act
erist
ic
s
a
nd
past
la
nd
sli
de
occ
urren
ce
[6
]
.
Th
e
pr
e
de
fine
d
thres
ho
l
ds
we
r
e
set
to
def
in
e
the
best
separ
at
ors
for
trigg
e
rin
g
an
d
non
-
trig
ge
rin
g
r
ai
nf
al
l
conditi
on
to
fore
cast
a
n
ea
rly
warnin
g
t
oo
l
s [3,
4] a
nd lan
ds
li
de hazar
d a
ssessm
ent.
Pr
eci
pitat
ion
da
ta
can
be
obta
ined
i
n
three
ways
in
the
ur
ban
a
rea
wh
ic
h
are
gro
und
obser
vatio
ns
,
weathe
r
ra
da
r
ob
s
er
vations,
a
nd
sat
el
li
te
m
on
it
or
i
ng
te
ch
niq
ue
s.
Gro
und
ob
s
er
vations
pro
vid
e
the
m
ost
d
irect
and
acc
urat
e
m
easur
em
ents
at
the
gauge
locat
ions,
but
it
is
diff
ic
ult
to
ob
ta
in
l
ong
-
te
rm
pr
eci
pitat
ion
data
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
Esti
ma
ti
on
of
TRMM r
ain
f
all f
or
l
ands
li
de o
ccurre
nces
bas
e
d on r
ain
fall
t
hr
es
ho
l
d
…
(
N
or
aisy
ah T
ajud
in
)
3209
because
of
t
he
geog
raphical
r
est
rict
ion
of
th
e
sta
ti
on
l
ocati
on
an
d
t
he
s
hortage
of
m
et
eo
ro
l
og
ic
al
ob
se
r
vatio
n
equ
i
pm
ent.
Ma
intai
ning
the
s
e
s
ta
ti
on
s
are
al
so
no
t
cost
e
ff
ect
ive.
Wh
il
e
preci
pitat
ion
of
ra
dar
ob
se
r
vations
hav
e
la
rg
e
uncertai
nties
relat
ed
to
er
ror
in
el
ect
ronic
sign
al
under
c
ha
ll
eng
in
g
ope
r
at
ing
en
vir
on
m
ents.
In co
ntrast,
satel
li
te
r
e
m
ote sen
sin
g
ca
n pro
vi
de
obse
rv
at
i
ons of
globa
l p
r
e
ci
pitat
ion
a
nd
cl
ouds
[7
-
10
]
.
Sate
ll
it
e
der
iv
ed
rai
nf
al
l
pro
duct
s
are
t
he
m
os
t
acce
pted
al
te
r
nat
ive
sou
rce
to
overc
om
e
the
lim
it
ation
s
of
gro
und
te
ch
niques.
Re
cent
ly
,
sat
el
li
te
info
rm
at
ion
has
be
com
e
avail
able
at
high
s
patia
l
and
tem
po
ral
res
ol
ution
a
nd
ov
e
r
la
rg
e
area.
Howe
ver,
al
l
sat
el
li
te
pr
eci
pitat
ion
produ
ct
s
prov
i
de
in
direct
est
i
m
at
es
of
pr
eci
pitat
ion
wh
ic
h
are
subj
ect
e
d
to
bi
as
and
stoc
ha
sti
c
err
ors,
dep
e
nd
highl
y
on
the
hy
dro
-
cl
im
at
ic
char
act
eris
ti
cs
of
a
re
gi
on.
The
refore
,
s
at
el
li
te
-
der
ived
preci
pitat
ion
pro
du
ct
s
nee
d
to
be
assessed
by
co
m
par
ing
the
da
t
a
with
groun
d
obser
vatio
ns
[7
,
11,
12]
.
The
pe
rfor
m
ance
of
rai
nf
al
l
produc
t
var
ia
ti
on
var
ie
s
by
da
ta
sou
rces
a
nd
retri
evin
g
al
gorit
hm
s
m
od
e.
Th
e
rain
fall
pro
du
ct
perf
or
m
ance
f
or
the
sam
e
ty
pe
of
data
al
so
s
hows
dif
fer
e
nc
es
in
dif
fer
e
nt
re
gions
a
nd
s
easo
ns
.
These
factors
in
dicat
e
that
the
pe
rfor
m
ance
of
rain
fall
sat
el
li
te
pr
oducts
a
re
af
f
ect
ed
by
the
seaso
n,
to
po
gr
a
phy,
locat
ion
a
nd
hydro
-
cl
i
m
at
ic
char
act
e
risti
cs o
f
t
he
stu
dy ar
ea. Th
e
refo
re,
t
he
reli
abili
ty
of the
sat
el
li
te
ra
infall
prod
uct
needs
to b
e v
al
idate
d and
co
m
par
e
d t
o
the g
rou
nd
-
base
d
m
easur
em
ents in
a sp
eci
fic area an
d
tem
po
ral scal
es
b
efore
it
can
be
used
in
a
ny s
ub
se
que
nt appli
cat
ion.
Nu
m
erous
stu
dies
on
the
ev
al
uation
of
sat
el
li
te
pr
eci
pitat
ion
pro
duct
s
with
gro
und
m
easur
em
ents
hav
e
bee
n
co
nducte
d
at
the
r
egio
nal
scal
e
[
4,
13,
14]
.
T
he
auth
or
s
i
n
[
15]
hav
e
vali
dated
that
TRM
M
sat
el
li
te
rainf
al
l
est
i
m
ates
can
be
us
e
d
to
determ
ine
cum
ulate
d
est
i
m
at
ed
rainf
al
l
–
eve
nt
durati
on
(E
D)
th
res
hold
for
the
f
or
eca
st
of
the
possi
ble
la
nd
sli
de
occurr
ence
in
the
Um
br
ia
reg
i
on,
Ce
ntral
Ital
y.
A
si
m
il
ar
stud
y
ha
s
bee
n
cond
ucted,
whil
e
the
pa
pe
r
i
n
[16]
ha
s
validat
ed
the
t
hr
es
ho
l
d
a
nd
te
ste
d
the
ef
fecti
ve
ness
of
the
di
f
fer
e
nt
thres
ho
l
d
de
finiti
on
m
et
ho
ds
with
ind
e
pe
ndent
la
ndsli
de
inform
at
ion
.
The
res
ult
sh
ows
the
G
PM
sat
el
li
te
rainf
al
l p
r
oduc
ts can b
e a
pp
li
ed
to
der
ive
rai
nf
al
l t
hr
e
shold
for
the
po
ssi
bl
e land
sli
de occ
urren
ce
.
I
n
Ma
la
ysi
a
,
stud
ie
s
on the
evaluati
on
of s
at
el
li
te
p
recipit
at
ion
pro
du
c
ts
are
ver
y l
im
i
ted
.
Most
of
t
he
stu
dies
co
nducte
d
are
base
d
on
t
he
com
par
iso
n
of
m
ulti
ple
tem
po
ral
(d
ai
ly
,
m
on
thly
and
annual)
rain
fal
l
betwee
n
sat
el
li
te
pr
od
ucts
a
nd
gro
und
obs
erv
at
io
n
[8,
17,
18]
.
T
he
ai
m
of
this
st
ud
y
was
t
o
evaluate
the
pote
ntial
us
e
d
of
TRM
M
-
3B
42V
7
data
in
com
par
ison
t
o
the
rain
-
gauge
data
base
d
on
rainf
al
l
thres
ho
l
d
anal
ysi
s
fo
r
the
predict
ion
of
la
ndsli
de
occ
urre
nce
in
Ul
u
Ke
la
ng
,
Sela
ngor
.
The
eval
uation
was
cond
ucted
by
usi
ng
five
sta
ti
sti
cal
ind
ic
at
or
s
to
com
par
e
the
aver
a
ge
m
on
thly
rainf
al
l,
c
um
ula
ti
ve
30
-
da
y
and
3
-
day
rain
fall
base
d
on
14
histor
ic
al
la
nd
sli
de
eve
nts
an
d
r
ai
nf
al
l
thres
ho
l
d
for
both
der
i
vations
from
T
RM
M
3B42V7
an
d
R
ai
n
ga
uge
data
series.
2.
RESEA
R
CH MET
HO
D
2.1.
St
ud
y are
a
This
st
ud
y
wa
s
co
nducte
d
i
n
Ulu
Kelan
g,
Sela
ng
or,
lo
c
at
ed
at
the
l
at
it
ud
e
of
3°1
2’3
0
’
N
a
nd
longit
ud
e
of
101°
45’
28’
’
E
with
a
dista
nc
e
of
5
km
fr
om
Ku
al
a
Lumpur
ci
ty
cente
r.
Ulu
Kela
ng,
S
el
ango
r
was
kn
own
as
the
m
os
t
su
scepti
ble
re
gion
of
la
ndsli
des
in
Ma
la
ysi
a.
Com
m
on
ly
,
the
cl
i
m
a
te
of
this
ar
ea
is
ho
t
an
d
hum
id
as
it
is
in
the
trop
ic
al
m
on
so
on
re
gion
an
d
pro
nounce
d
wet
or
rai
ny
seaso
ns
sta
rted
from
Februa
ry
un
ti
l
Ma
y
and
Sept
e
m
ber
unti
l
Dece
m
ber
each
ye
ar.
Acc
ordin
g
to
the
Ma
la
ysi
an
Me
te
orolo
gical
Dep
a
rtm
ent
(
MM
D)
,
the
ac
cum
ulate
d
m
o
nth
ly
rain
fall
f
or
Ul
u
Kelan
g
is
between
58 an
d
420
m
m
p
er
m
on
th
,
with
an
a
ve
rage
of
200
rainy
days
pe
r
ye
ar
and
2440
m
m
aver
a
ge
a
nnual
rainf
al
l.
T
he
Ulu
Kelan
g
ar
ea
has
su
f
fer
e
d seve
ra
l fata
l l
andsl
id
es cases
due to
r
ai
nfal
l eve
nts
[19
]
.
2.2
.
Va
li
d
at
i
on
pr
ocess
T
her
e
are
tw
o
ind
e
pende
nt
da
ta
sources
of
r
ai
nf
al
l
in
form
a
ti
on
from
19
98
to
2011
wer
e
us
e
d
in
this
stud
y.
The
firs
t
is
a
daily
rainf
al
l
m
easur
e
m
ent
ob
ta
ine
d
by
2
rain
ga
uges,
nam
el
y
th
e
Em
pan
gan
Gen
ti
ng
Kelan
g
Stat
ion
and
JPS
Am
pan
g
sta
ti
on
loc
at
ed
in
Ulu
K
e
la
ng
A
rea,
Sel
ango
r.
The
sec
ond
sou
rce
of
r
ai
nf
al
l
inf
or
m
at
ion
is
a
sat
el
l
it
e
r
ai
nf
al
l
est
i
m
ate
product,
w
hich
is
TRM
M
-
3B42
V7
w
it
h
a
reso
luti
on
of
0.25O
x
0.25O,
pro
vi
ded
by
NA
S
A
TRM
M.
The
us
e
d
of
TRM
M
-
3B
42V7
data
in
this
stud
y
is
base
d
on
the
adequate
r
el
ia
bili
ty
of
this
pr
od
uct
to
est
i
m
at
e
the
m
on
t
hly
and
an
nu
al
preci
pitat
io
n
ove
r
the
trop
ic
a
l
reg
i
on
[
20,
21]
.
Accor
ding
to
the
J.
Pask
a
at
el
,
a
m
od
erate
correla
ti
on
be
tween
TRM
M
-
3B42V7
an
d
gro
un
d
ob
s
er
vation
w
as
fou
nd
f
or
da
il
y
pr
eci
pitat
ion
i
n
Pe
nin
s
ul
ar
Ma
la
ysi
a
[22].
The
validat
ion
was
c
ondu
ct
ed
by
com
par
ing
th
r
ee
analy
ses,
w
hich
are
the
a
ver
a
ge
m
on
thl
y
rain
fall
,
cu
m
ula
ti
ve
30
-
da
y
an
d
3
-
day
rainf
al
l
base
d
on
14
hi
storical
la
nd
sli
de
eve
nts
an
d
rainf
al
l
thre
sho
ld
analy
sis
in
Ulu
Kela
ng,
S
el
angor.
The
a
v
era
ge
m
on
thly
rainf
a
ll
was
analy
zed
ove
r
14
ye
ar
s
from
19
98
t
o
20
11.
T
he
pe
rio
d
was
sel
ect
ed
base
d
on
the
14
histor
ic
al
la
ndsli
de
eve
nts
wh
ic
h
occ
urre
d
from
1998
to
2011.
Th
e
ne
xt
c
om
par
iso
n
was
bas
ed
on
the
cum
ulati
ve
30
-
day
an
d
3
-
day
rainf
a
ll
par
am
e
te
r
wh
ic
h
in
vo
l
ve
d
in
the
co
nst
ru
ct
io
n
of
r
ai
nf
al
l
thres
ho
l
d
for
Ulu
Kelan
g,
S
el
angor.
Fi
nally
,
the
rain
fall
thres
hold
(E
3
-
E3
0
diag
ram
)
de
rive
d
f
r
om
both
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.
10
, No
.
3
,
J
une
2020
:
32
08
-
3215
3210
TRM
M
-
3B42
V7
a
nd
Ra
in
gauge
was
co
m
par
ed
to
ide
ntify
the
reli
abili
ty
of
rainfal
l
threshold
ba
sed
on
TRM
M
-
3
B4
2V7
pro
duct
f
or
t
he
pre
dicti
on
of
la
ndsli
de
oc
cu
rr
e
nce.
Figure
1
s
hows
the
fl
owc
har
t
of
validat
io
n proc
ess.
2.3
.
Va
li
d
at
i
on
s
t
ati
stics
The
Me
a
n
(M
),
Me
an
Dif
fe
ren
ce
(MD
),
Roo
t
Me
a
n
S
qu
a
re
E
rror
(
RM
SE),
Bi
as
and
Pears
on
correla
ti
on
c
oe
ff
ic
ie
nt
(R
)
w
e
re
us
e
d
t
o
co
m
par
e
the
dif
f
eren
ces
an
d
c
orrelat
ion
bet
w
een
TRM
M
-
3B
42V7
and
rain
ga
uge
s.
The
point
-
to
-
pi
xel
assessm
ent
for
com
pari
ng
the
pi
xels
and
the
c
orres
pondin
g
rain
ga
ug
e
s
was
us
e
d.
The
Me
an
is
use
d
to
de
fine
t
he
central
te
nde
ncy
of
rain
fall
data
set
f
or
both
rain
gauge
s
an
d
TRM
M
-
3B42
V7,
w
hile
MD
is
the
m
ean
diff
ere
nce
betw
een
the
tw
o
da
ta
set
s.
RM
SE
m
easur
es
the
m
ean
error
betwee
n
rain
gau
ges
a
nd
TRM
M
-
3B
42V
7
rai
nf
al
l,
i
n
m
i
ll
i
m
et
er
(m
m
).
The
Bi
as
is
a
m
easur
e
m
ent
of
how
t
he
a
verage
sat
el
li
te
r
ai
nf
al
l
est
im
a
t
es
the
m
agn
it
ud
e
as
c
om
par
ed
t
o
t
he
groun
d
-
base
d
ra
infall
ob
s
er
vation,
w
her
e
is
the
pe
r
fect
sco
re
is
1.
A
bias
value
above
or
be
lo
w
1
i
ndic
at
es
an
a
ggre
gate
s
at
el
li
t
e
rainf
al
l
ove
re
stim
ation
or
unde
restim
at
ion
of
the
gr
ound
-
base
d
preci
pitat
ion
a
m
ou
nts
resp
e
ct
ively
.
The
Pear
son
c
orrelat
ion
is
use
d
to
sho
w
the
cor
relat
io
n
be
tween
s
patia
l
distrib
ution
of
the
TRM
M
-
3B
42V
7
and
rain
ga
ug
e
s,
wit
h
t
he
va
lue
ra
nges
betw
een
0
–
1.
0
i
ndic
at
es
no
co
rrel
at
ion
w
hile
the
val
ue
of
1
s
how
s
a
perfect
co
rrel
at
ion
.
T
he
e
qu
at
io
ns
f
or
t
he
sta
ti
sti
cal
m
et
rics
that
wer
e
us
e
d
f
or
validat
io
n
ar
e
show
n
in Ta
ble 1.
Figure
1.
Flo
w
char
t
of v
al
i
dation p
ro
ces
s
Table
1.
List
of the
stat
ist
ic
al
m
et
rics e
m
plo
ye
d
to
validat
e
the p
e
rfo
rm
ance of TRMM
-
3B
42V7 [2
3]
Statistical
Metr
i
cs
Equ
atio
n
Perf
ect Score
Mean (
M)
=
1
∑
=
1
-
Mean Dif
f
erent (
M
D)
=
1
∑
(
−
)
=
1
0
BIAS
=
∑
∑
1
Ro
o
t M
ean
Sq
u
are
E
r
ror (
R
MSE
)
=
√
1
(
−
)
2
0
Pearso
n
Co
rr
elatio
n
(
R)
=
∑
(
−
̅
)
(
−
̅
)
=
1
√
∑
(
−
̅
)
2
=
1
√
∑
(
−
̅
)
2
=
1
1
W
her
e
Xi
is t
h
e repre
sent
at
ion
of
rai
n
gaug
ed
r
ai
n
fal
l
,
at
a ti
m
e i,
Yi
is t
h
e
r
epr
ese
nta
ti
on
of TRM
M
-
3B42V7 a
t a
t
ime
i
.
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
Esti
ma
ti
on
of
TRMM r
ain
f
all f
or
l
ands
li
de o
ccurre
nces
bas
e
d on r
ain
fall
t
hr
es
ho
l
d
…
(
N
or
aisy
ah T
ajud
in
)
3211
3.
RESU
LT
S
A
ND AN
ALYSIS
3.1
.
Co
m
pa
ri
s
on
of
av
er
age m
onthly
ra
in
fa
ll
Figure
2
sho
ws
the
avera
ge
m
on
thly
rainf
al
l
for
the
case
-
stu
dy
of
ov
er
14
-
ye
ar
s
per
io
d,
by
com
par
ing
the
TRM
M
-
3B
42V7
an
d
ga
ug
e
data.
It
al
so
shows
the
distri
buti
on
of
rain
fa
ll
data
an
d
la
ndsli
de
even
ts
occurre
nces
in
19
98
to
2011.
T
he
com
par
ison
be
tween
tw
o
rai
nf
al
l
data
re
ve
al
s
that
the
TRM
M
rainf
al
l
pr
oduc
t
has
overe
stim
at
ed
rain
fall
di
stribu
ti
on
in
Ja
nu
a
ry,
Fe
brua
r
y,
and
Decem
ber
.
I
n
the
m
ean
tim
e
,
betwee
n
Ma
rc
h
to
N
ov
em
be
r,
the
TRM
M
pro
du
ct
sho
ws
an
unde
resti
m
at
e
rainf
al
l
distribu
ti
on.
H
oweve
r,
the
high
est
a
m
ou
nt
of
TR
MM
rainf
al
l
pro
du
ct
is
rec
orde
d
duri
ng
this
per
i
od.
E
ven
t
hough
th
ere
is
a
diff
e
re
nce
in
the
rain
fall
value
of
TRM
M
pro
du
ct
,
bu
t
it
the
ave
rag
e
m
on
t
hly
rainf
al
l
der
i
ved
by
rain
gaug
e
rainf
al
l
data
s
hows
sim
i
la
r
t
rend.
F
ro
m
the
analy
sis,
TR
MM
-
3B42
V7
has
dem
on
stra
te
d
a
sign
ific
a
nt
and
strong
posit
ive
correla
ti
on
to
the
rai
n
ga
uge
rainf
al
l
data
wi
th
R
eq
uiva
le
nt
to
0.8
8.
The
RM
SE
betwe
e
n
rai
n
-
gauge
a
nd
TR
MM
data
ha
s
lowe
r
value,
wh
ic
h
is
12.
97
m
m
.
m
on
th
-
1
that
is
cl
os
e
t
o
the
rain
-
ga
uge
data.
T
his
fin
ding
i
s
in
agr
eem
ent
with
the
stu
dy
in
[8,
24]
wh
ic
h
re
porte
d
the
go
od
pe
rfor
m
ance
of
TRM
M
pro
du
ct
s.
How
ever,
the
TRM
M
3B4
2V7
has
unde
resti
m
at
ed
the
rain
-
ga
uge
pr
eci
pitat
ion
with
a
BIAS
of
-
0.0
985.
The
m
axi
m
u
m
m
o
nth
ly
aver
a
ge
rainf
al
l
f
or
r
ai
n
-
gauge
data
is
38
2.8
75m
m
in
No
vem
ber
a
nd
321.7
475
m
m
for
TRM
M
product
was
rec
orde
d
in
A
pri
l.
Me
an
wh
il
e,
t
he
m
ini
m
u
m
values
wer
e
13
3.5
m
m
(F
e
bruar
y)
a
nd
14
1.867
5
m
m
(Ju
ly
)
res
pe
ct
ively
.
This
analy
sis
al
so
sh
ows
t
hat
the
hi
gh
e
st
nu
m
ber
of
la
nd
sli
de
e
ven
t
s
occurre
d
w
he
n
there
are
hi
ghest
a
m
ou
nt
of
rainf
al
l
durin
g
the
rainy
season
o
bvio
us
ly
in
May
and
N
ovem
ber
.
T
her
e
fore,
th
e
possibil
it
y
of
rain
fall
as
a
tr
igg
e
rin
g
factor
f
or
la
ndsli
de
oc
currence
s
in
wet
or
rainy seas
on is
higher
in Ulu
Kelan
g,
Sela
ng
or.
Figure
2.
Mo
nth
ly
av
e
rag
e
r
ai
nf
al
l
(
1998
-
20
11) for
rain ga
ug
e
, TRMM
-
3B
42V7 a
nd
the num
ber
of
l
andsl
ide e
ven
ts
3.2
.
Co
m
pa
ri
s
on o
f
cum
u
la
tive 3
0
-
d
ay r
ainfall
an
d
3
-
d
ay ra
i
nf
all
The
nex
t
a
naly
sis
was
ref
e
rr
i
ng
t
o
the
pa
ra
m
et
er
inv
ol
ve
d
in
the
der
i
vation
of
the
rain
f
al
l
threshold
m
od
el
f
or lan
dsl
ide o
cc
urre
nc
es in
Ulu Kela
ng Selan
gor. T
he
c
um
ulati
ve
30
-
day an
d 3
-
da
y rainfall
were use
d
to
com
par
e
the
rainf
al
l
intensit
y
between
rain
ga
ug
e
a
nd
TRM
M
-
4B
42V
7
duri
ng
l
andsl
ide
occ
urren
ces
.
The
c
um
ulati
ve
rain
fall
d
at
a i
s b
ase
d o
n
14
histor
ic
al
lan
dsl
ides that
occur
red
betw
een
19
98 to 2
011.
Fig
ur
e
3
sh
ows
the
rain
fall
intensit
y
f
or
c
um
ulati
ve
30
-
day
an
d
c
um
ula
ti
ve
3
-
day
rain
f
or
ga
ug
e
an
d
T
RMM
-
3B
42V7.
Both
histo
gr
a
m
s
in
Figure
3
(a)
a
nd
(
b)
s
how
that
TRM
M
-
3B42
V7
da
ta
has
unde
rest
i
m
at
ed
as
com
par
e
d
t
o
the
rai
n
ga
ug
e
rain
fall
inten
s
it
y.
The
dif
fere
nces
i
n
per
ce
ntage
of
rain
gauge
to
TRM
M
-
3B42
V7
is
63%
(Cum
ulati
ve
30
-
day)
a
nd
45
%
(Cum
ulati
ve
3
-
day)
res
pe
ct
ively
.
The
m
axim
u
m
cu
m
ulati
ve
30
-
da
y
r
ai
nf
al
l
intensit
y
for
T
RM
M
-
3B42V7
is
387.3
6
m
m
,
wh
il
e
rain
gauge
is
75
9.5
m
m
;
and
f
or
cum
ulati
ve
3
-
day
are
94.89
m
m
fo
r
TRM
M
and
196
m
m
fo
r
the
rain
gauge.
O
ver
al
l,
these
a
naly
ses
ind
ic
at
e
that
the
rain
gau
ge
rainf
al
l
has
e
xc
eeded t
he TR
MM
-
3B42
V7 ra
infall
.
Each
TMPA
-
3B
42V7
produc
t
was
plo
tt
ed
a
gainst
ga
ug
e
da
ta
in
scat
te
r
pl
ots
base
d
on
cum
ulati
ve
30
-
day
an
d
3
-
da
y
rainfal
l
duri
ng
la
ndsli
des
oc
currence
s f
r
om
19
98
to
20
11
at
U
lu
Kela
ng
Sela
ngor,
a
s sh
ow
n
in
Fi
gu
re
4.
T
his
a
naly
sis
is
carried
ou
t
t
o
analy
ze
the
c
orrelat
ion
betwe
en
TRM
M
-
3B
42V
7
a
nd
rai
n
ga
ug
e
.
Fr
om
the
anal
ysi
s,
bo
t
h
TR
MM
-
3B42
V7
cum
ulati
ve
30
-
day
a
nd
3
-
da
y
rainf
al
l
pres
ented
a
si
gn
i
ficant
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.
10
, No
.
3
,
J
une
2020
:
32
08
-
3215
3212
strong
posit
ive
co
rr
el
at
io
n
to
the
rain
ga
ug
e
rain
fall
data.
The
co
rrel
at
ion
co
ef
fici
ent
of
TRM
M
-
3B
42V
7
ver
s
us
rain
ga
ug
e
is
0.6
6
f
or
cum
ulati
ve
30
-
day
rain
fall
and
0.6
3
for
c
um
ula
ti
ve
3
-
day
rain
fall
.
By
re
ferrin
g
to
RSM
E
valu
e
fo
r
bo
t
h
cum
ulati
ve
30
-
day
and
3
-
day
rain
f
al
l,
it
sh
ows
th
at
TRM
M
-
3B42V7
data
is
cl
ose
r
to
rain
ga
ug
e
dat
a
set
.
T
he
RM
SE
val
ue
is
5.81
m
m
fo
r
cu
m
ula
ti
ve
30
-
da
y
rainf
al
l
wh
il
e
for
c
um
ulativ
e
3
-
day
rainf
al
l
is 5.31m
m
.
The
res
ult
of
t
he
sta
ti
sti
cal
m
et
rics
fo
r
TRM
M
-
3B42
V7
v
e
rsus
rai
n
gauge
e
ve
nt
rai
nf
al
l
is
sh
ow
n
in
Ta
ble 2
.
3.3
.
Co
m
pa
ri
s
o
n
of r
ainfall
t
hre
sho
ld
analy
sis f
or
TR
M
M
-
3B
42V7 and r
ai
n gauge r
ainfall
dat
a
To
def
i
ne
the
rainf
al
l
that
t
rigg
e
re
d
the
la
ndsli
de,
it
requi
res
the
ide
ntific
at
ion
of
rain
f
al
l
tim
e
or
conditi
on
a
nd
oth
er
relat
ed
inform
at
ion
of
the
la
ndsli
de
occ
urre
nce
s.
Ba
se
d
on
the
rain
fall
an
al
ysi
s
,
the
rain
fall
con
diti
ons
that
ha
d
trig
ger
e
d
th
e
la
nd
sli
des
in
Ulu
Kela
ng
a
re
ei
ther
sho
rt
and
lo
ng
durat
ion
or
prolo
nged
co
nt
inuous
per
io
d
of
rain
fall
.
T
he
refor
e
,
the
rai
nf
al
l
th
res
ho
l
d
that
inc
orp
or
a
te
s
cum
ulati
ve
3
-
day
ver
s
us
30
-
day
rainf
al
l
for
f
ourteen
sel
ect
ed
la
nd
sli
de
s
even
t
was
deri
ved
f
or
the
be
st
rainf
al
l
thr
esh
old
conditi
on
of
U
lu
Kela
ng,
Sel
ango
r.
T
his
m
et
hod
is
a
ppr
opriat
e
f
or
the
r
ai
nf
al
l
co
ndit
ion
that
had
tri
gg
e
re
d
la
nd
sli
de
in
t
he
area
[25]
.
T
he
pl
otted
rain
fall
threshold
(E
3
–
E
30
)
dia
gr
am
fo
r
histo
rical
rainf
al
l
that
has
resu
lt
ed
i
n
la
ndsli
de
us
i
ng
TRM
M
-
3B42V7
pro
duct
and
rain
ga
uge
rain
fall
data
is
sh
own
in
Fig
ur
e
5(a)
a
nd
(b)
res
pecti
vel
y.
Both
rai
nf
al
l
thres
ho
l
d
li
ne
s
wer
e
ge
ner
at
ed
f
ro
m
the
co
rr
el
at
ion
bet
we
en
the
3
-
day
(
E
3
)
an
d
30
-
day
(E
30
)
a
ntecede
nt
pr
eci
pitat
ion
i
nd
e
x base
d on the
oc
currence
of m
inor a
nd m
ajo
r l
and
sli
de
ev
e
nt
s.
(a)
(b)
Figure
3. Total
r
ai
nfal
l (a
)
c
um
ula
ti
ve
30
-
da
y rainfall
a
nd (b) cum
ulati
ve
3
-
day rai
nf
al
l
f
or r
ai
n ga
uge a
nd
TRM
M base
d on 14 l
a
ndsli
de
ev
e
nts
in
u
l
u k
el
ang
,
S
el
an
go
r
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
Esti
ma
ti
on
of
TRMM r
ain
f
all f
or
l
ands
li
de o
ccurre
nces
bas
e
d on r
ain
fall
t
hr
es
ho
l
d
…
(
N
or
aisy
ah T
ajud
in
)
3213
(a)
(b)
Figure
4. Scat
te
r plot
betwee
n rain
ga
ug
e
and TRM
M
3B42V7
base
d o
n 14
la
ndsli
de
e
ve
nts in
U
l
u Kela
ng,
Sela
ngor,
(a) c
um
ulati
ve
30
-
da
y rainfall
(b) c
um
ulati
ve
3
-
da
y rainfall
Table
2.
Res
ults o
f
sta
ti
sti
cal
m
et
rics
fo
r
TR
MM
-
3B42
V7
vs
rain ga
ug
e
e
ven
t
rain
fall
C
u
m
u
l
a
t
i
v
e
R
a
i
n
f
a
l
l
3
0
Da
y
3
Da
y
MD
1
4
9
.
7
8
5
1
.
6
7
B
I
A
S
0
.
3
6
8
7
0
.
5
4
4
4
R
M
SE
5
.
8
1
5
.
3
1
R
0
.
6
6
5
1
0
.
6
3
2
6
(a)
(b)
Figure
5. Plott
ing o
f
E
3
-
E3
0 d
ia
gr
am
fo
r
h
ist
or
ic
al
r
ai
nf
al
l t
hat h
a
s r
e
su
lt
e
d
in
lan
ds
li
de
usi
ng
(a)
TRM
M
-
3B
42V7
rai
nf
al
l
d
at
a a
nd (b)
ra
in g
a
uge
rainfa
ll
d
at
a
To
c
om
par
e
th
e
rain
fall
thr
es
ho
l
d
diag
ram
betwee
n
TRM
M
-
3B42
V7
a
nd
rain
ga
ug
e
r
ai
nf
al
l
data
,
bo
t
h
gr
a
phs
a
r
e
plo
tt
e
d
in
th
e
diag
r
am
as
in
Fi
gure
6.
Th
e
fig
ur
e
s
s
how
,
the
rain
fall
thres
hold
e
quat
ion
f
or
TRM
M
-
3B42
V7
is
lo
we
r
t
han
the
rai
n
ga
ug
e
rai
nf
al
l
t
hr
es
hold
eq
uat
ion
base
d
on
the
m
ajo
r
an
d
m
ino
r
la
nd
sli
de
t
hr
e
s
ho
l
d
e
qu
at
io
n’
s
analy
sis.
T
he
eq
uation
f
or
both
rainfal
l
thres
ho
l
d
is
ta
bu
la
te
d
in
Ta
ble
3.
By
ref
er
rin
g
t
o
the
c
om
par
ison
of
r
ai
nf
al
l
thres
hold
of
TRM
M
-
3B42
V7
an
d
rain
gauge,
it
sho
ws
that
the un
der
est
im
at
ion
of the “
gr
ound
-
base
d” ra
infall
d
at
a is
observe
d
[16]
.
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.
10
, No
.
3
,
J
une
2020
:
32
08
-
3215
3214
Figure
6. Com
bin
at
io
n of
rai
nf
al
l t
hres
ho
l
d deri
ved b
y
rain
g
a
ug
e
an
d TR
MM
-
3B42
V7
Table
3.
T
he
e
qu
at
io
n o
f
rai
nfal
l t
hr
es
hold
f
or m
ajo
r
a
nd m
inor lan
ds
li
de
occurre
nces
der
i
ved f
ro
m
r
ai
n
ga
ug
es,
TR
MM
-
3B42
V7
Rain
f
all Pr
o
d
u
ct
Major
Land
slid
e T
h
resh
o
ld
Mino
r
Land
slid
e T
h
resh
o
ld
Rain
Gaug
es
3
=
167
.
39
−
0
.
21
30
3
=
108
.
73
−
0
.
21
30
TRMM
-
3
B4
2
V7
3
=
140
.
016
−
0
.
448
30
3
=
104
.
56
−
0
.
448
30
4.
CONCL
US
I
O
N
In
this
st
udy,
the
sta
ti
sti
cal
m
et
ho
d
an
d
r
ai
nf
al
l
th
resho
ld
a
naly
sis
is
us
e
d
to
c
om
par
e
TRM
M
3B42V7
a
nd
r
ai
n
ga
ug
e
pr
ec
ipit
at
ion
data
s
et
s
on
da
il
y
scal
e.
The
analy
s
is
is
based
on
cum
ulati
ve
3
-
da
y
and
30
-
day
rain
fall
of
hist
or
ic
al
la
nd
sli
de
whic
h
is
the
p
a
r
a
m
et
er
fo
r
rainf
al
l
thre
shold
.
The
resu
lt
s
hows
,
the
RM
SE
of
aver
a
ge
m
on
th
ly
rainf
al
l
fo
r
TRM
M
3B4
2V7
is
12.
97
m
m
/
m
on
th
wh
ic
h
is
cl
os
e
to
t
he
rain
-
gauge
data.
Th
e
TRM
M
3B42V7
has
overe
stim
at
ed
rainf
a
ll
distribu
ti
on
i
n
Ja
nu
a
ry,
Februa
ry,
a
nd
Dec
e
m
be
r
whi
c
h
are
at
the
end
of
N
or
t
he
ast
Mon
soo
n
seaso
n.
W
hile
the
RM
SE
of
c
um
ulati
ve
30
-
da
y
rainf
al
l
and
3
-
day
rainf
al
l
s
how
that
TRM
M
4B4
2V
7
pro
duct
cl
os
er
with
rai
n
ga
uge
data
s
et
s
about
5.8
1
m
m
(Cu
m
ulativ
e
30
-
day)
a
nd
5.3
1
m
m
(Cu
m
ulati
ve
3
-
day)
.
O
n
the
oth
e
r
ha
n
d,
the
c
orrel
at
ion
(R
)
bet
ween
rain
-
ga
uge
an
d
TRM
M
4B42
V7
prese
nted
a
sig
nificant
a
nd
str
ong
posi
ti
ve
co
rr
el
at
io
n,
t
hat
is
0.6
6
for
c
um
ulati
ve
30
-
day
rainf
al
l
a
nd
0.63
f
or
c
um
ulativ
e
3
-
day
rai
nfal
l
resp
ect
ively
.
The
com
par
ison
of
rain
fall
thres
hold
(E
3
–
E
30)
diag
ram
fo
r
histor
ic
al
rainfal
l
that
has
resul
te
d
in
la
nd
sli
de
show
s
that
the
rainf
al
l
threshold
e
qu
at
i
on
f
or
TRM
M
-
3B42
V7
is
l
ow
e
r
th
an
the
rain
gauge
rai
nf
al
l
thre
sh
ol
d
e
qu
at
io
n
base
d
on
m
ajor
an
d
m
ino
r
la
nd
sli
de
thres
ho
l
d
eq
ua
ti
on
analy
sis.
Ov
e
rall
,
the
r
esults
show
t
ha
t
the
com
par
ison
of
the
pa
ram
et
ers
of
ra
infall
thres
ho
l
ds
obt
ai
ned
from
th
e
rain
ga
uge
and
TRM
M
3B
42V7
us
i
ng
the
sta
ti
sti
cal
m
et
ho
ds,
re
ve
al
that
the
thres
ho
l
d
de
rive
d
from
th
e
sat
el
lite
pr
od
ucts
has
un
derest
i
m
at
ed
the
rainf
al
l
distrib
ut
ion
with
re
sp
e
ct
to
gro
und
-
base
d
ob
s
er
vations.
These
res
ults
are
in
a
gree
acc
orda
nce
th
e
fin
ding
in
[
4,
9,
16]
in
wh
i
c
h
the
un
der
est
im
at
ion
of
the
“
gro
und
-
ba
sed”
r
ai
nf
al
l
data
ob
serv
at
io
n.
Fin
a
ll
y,
the
pro
pos
ed
m
et
ho
d
w
hi
ch
is
us
in
g
t
he
TRM
M 3B4
2V7
prod
ucts
can
b
e
u
s
ed
to
obtai
n rai
nf
al
l t
hres
ho
l
d for
po
s
sible l
a
nd
sli
de.
ACKN
OWLE
DGE
MENTS
The
auth
ors
w
ou
l
d
li
ke
to
thank
Fac
ulty
of
Ele
ct
rical
En
gin
ee
rin
g,
U
niv
ersit
i
Tek
no
l
og
i
MAR
A
(U
iTM
)
f
or
th
ei
r
val
uab
le
s
upport.
T
his
r
esearch
is
par
t
ly
fund
e
d
by
t
he
M
al
ay
sia
n
Gove
rn
m
ent
thr
ou
gh
UiTM
un
der
600
-
IRMI/
DAN
A5
/
3/BEST
ARI
(
122/
2018).
We
ar
e
grat
ef
ul
to
NASA
T
MPA
f
or
provi
ding
the TRM
M p
r
oduct
ve
rsion
7 for
3B42(7
) da
ta
.
REFERE
NCE
S
[1]
E.
Mons
ie
urs,
O.
Dewitt
e
,
and
A.
Dem
ouli
n
,
“
A
su
sce
pti
bilit
y
-
bas
ed
rai
nf
al
l
thre
shold
appr
oac
h
for
la
ndsl
i
de
occ
urre
n
ce,”
Nat
ural
Haz
ards
and
Earth
S
yste
m S
ci
en
ce
s
,
vol. 19
,
no.
4
,
pp
.
775
–
7
89,
Apr
.
2019.
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
Esti
ma
ti
on
of
TRMM r
ain
f
all f
or
l
ands
li
de o
ccurre
nces
bas
e
d on r
ain
fall
t
hr
es
ho
l
d
…
(
N
or
aisy
ah T
ajud
in
)
3215
[2]
F.
Marra
et
a
l
.
,
“
Im
pac
t
of
ra
inf
al
l
spa
tial
aggr
e
gat
ion
on
th
e
id
ent
ifica
ti
on
of
d
ebr
is
flo
w
o
cc
ur
ren
ce
thre
sholds
,
”
Hydrology
and
Earth
Syst
em
Sc
i
enc
es
,
vol
.
21
,
n
o.
9
,
pp
.
4525
–
4
532,
2017
.
[3]
J.
C.
Robbins,
“
A
proba
bil
isti
c
appr
oac
h
for
as
sess
ing
la
ndslide
-
tri
gg
eri
ng
event
rai
nfa
ll
in
Papua
New
Guinea
,
using T
RMM
sa
te
ll
it
e
pre
ci
p
it
a
tion e
stim
ates,
”
J
ournal
of
Hydrol
ogy
,
vo
l. 541, part
A,
pp.
296
–
30
9,
Oct
.
2016
.
[4]
A.
Mondini,
D.
Kirschba
um
,
M.
Ross
i,
F.
Guzz
e
tt
i,
and
D.
Vali
g
i,
“
Com
par
ison
of
Sate
l
li
t
e
Ra
in
fal
l
Esti
m
at
es
an
d
Rai
n
Gaug
e
Me
a
surem
ent
s in
I
ta
l
y
,
and
Im
pac
t
on
La
ndsl
ide Mode
li
ng,
”
Cl
imate
,
v
ol.
5
,
no
.
4
,
pp
.
9
0,
2017
.
[5]
D.
Kaz
m
i,
S.
Qa
sim
,
I.
S.,
Har
ah
ap,
S.
B
aha
rom
,
M.
Im
ran
,
and
S
.
Moin,
“
A
Study
on
the
Contr
ib
uti
ng
Fac
tors
of
Maj
or
L
andsli
de
s in
Mal
a
y
s
ia,”
Civ
il E
ng
ine
en
ering
Journal
,
vo
l. 2, no. 12, pp. 66
9
–
678,
Jan
.
201
7.
[6]
Y.
Hong,
R.
Ald
er,
and
G
.
Huffm
an,
“
Eva
luatio
n
of
the
pot
ent
i
a
l
of
NA
SA
m
ult
i
-
satelli
t
e
pre
ci
pi
ta
ti
on
anal
y
s
is
i
n
globa
l
la
ndsl
ide
haz
ard
assess
m
ent
,
”
Geophy
si
cal R
ese
arch
Let
te
r
s
,
vol. 33, no. 22
,
Nov
.
2006.
[7]
M.
Ren
et
al
.
,
“
As
sessment
of
S
at
e
ll
ite
-
Deri
v
ed
Prec
ipi
t
ation
Products
for
t
he
Be
i
ji
ng
Regi
on
,
”
in
Re
mote
S
ensing
,
vol.
10
,
no
.
12
,
p
p
.
1914
,
Nov
.
20
18.
[8]
M.
L.
T
an,
A
.
L.
Ibra
him,
Z
.
Dua
n,
A.
P.
Cra
cknell,
and
V.
Chap
l
ot
,
“
Eva
lu
at
ion
o
f
six
high
-
resolu
ti
on
satellite
and
ground
-
base
d
pr
ec
ip
it
a
ti
on
produ
ct
s ove
r
Mal
a
y
s
i
a,
”
in
Re
mot
e
S
e
nsing,
vol
.
7
,
no
.
2,
pp.
1504
–
152
8,
Jan
.
2015
.
[9]
M.
T.
Brun
et
t
i,
M.
Meli
l
lo,
S.
P
eru
ccacci,
L.
Ci
aba
t
ta
,
and
L
.
B
roc
ca,
“
How
far
are
we
from
t
h
e
use
of
sat
el
l
ite
rai
nfa
ll
produ
ct
s
in la
ndslid
e
for
e
ca
sting
?
,
”
R
emot
e
Sensing
of
Environ
ment
,
vo
l. 2
10,
pp
.
65
–
75
,
J
un
.
2018
.
[10]
T.
G.
Gebre
m
ica
el
e
t
al
.
,
“
Com
p
ari
son
and
v
al
id
at
ion
of
ei
gh
t
sat
el
lite
r
ai
nf
al
l
pro
duct
s
over
the
ru
gged
topogra
p
h
y
of
Te
k
eze
-
Atba
ra
Basin
a
t
dif
fer
ent
spati
al
a
nd
te
m
pora
l
sc
al
es,
”
Hydrology
and
Earth
S
yste
m
Sci
en
ces
Discuss
ions
,
pp.
1
–
31,
Aug
.
2017
.
[11]
Q.
Jiang
e
t
al
.
,
“
Acc
ura
c
y
ev
al
ua
ti
on
of
two
h
igh
-
resolut
ion
sa
tell
it
e
-
b
ase
d
ra
infa
l
l
produc
ts:
TRMM
3B42V7
and
CMO
RP
H i
n
Shanghai,”
in
W
ater
,
vol. 10, no. 1,
pp.
40
,
Jan
.
201
8.
[12]
P.
S.
Kati
rai
e
-
B
orouje
rd
y
,
A.
Akbari
As
anj
an,
K.
li
n
Hs
u,
and
S.
Soroo
shian,
“
Inte
rco
m
par
iso
n
of
PERS
I
ANN
-
CDR
and
TRMM
-
3B42V7
pre
ci
pitati
on
esti
m
at
es
at
m
onthl
y
and
dai
l
y
ti
m
e
sca
le
s,
”
At
mos
pheric
Re
search
,
vol.
193
,
pp
.
36
–
49,
Sep
.
2017
.
[13]
F.
Lo
Conti
,
K
.
L.
Hs
u,
L
.
V.
Noto,
and
S.
Soroos
hia
n
,
“
Evaluat
ion
and
co
m
par
ison
of
sate
ll
i
te
pr
ec
ip
it
a
tion
esti
m
at
es
with
re
fer
ence
to
a
loc
a
l
are
a
in
the
Med
it
err
ane
an
Sea
,
”
in
At
mos
pheric
Re
search
,
vol
.
138,
pp.
189
–
204,
Mar
.
2014
.
[14]
D.
A.
Hughes,
“
Com
par
ison
of
satelli
t
e
r
ai
nfa
ll
d
at
a
wi
th
observa
ti
ons
fr
o
m
gauging
stat
ion
n
et
wo
rks,”
Journal
o
f
Hydrology
,
vol.
327,
no
.
3
–
4
,
pp
.
399
–
410,
Aug
.
2006.
[15]
F.
Guzz
e
tt
i
,
M.
Ross
i,
S.
Lu
ciani,
D.
Kirschb
aum,
and
A
.
C.
Mondini,
“
TRM
M
sate
llite
ra
inf
al
l
esti
m
ates
for
la
ndslide
ea
rl
y
warni
ng
in
Ita
l
y:
pre
li
m
ina
r
y
re
sults,”
in
Proceedi
ngs
of
SPIE
-
The
Inte
rnational
Soci
ety
for
Optic
al
Engi
n
eer
ing
,
vol
.
8523
,
p
p
.
85230D,
Oct
.
2012
.
[16]
M.
Ross
i
et
a
l
.
,
“
Stat
isti
c
al
appr
oac
hes
for
the
d
efi
nition
of
la
nd
slide
r
ai
nfa
ll
thr
esholds
and
th
eir
unce
r
ta
in
t
y
usi
ng
rai
n
g
auge a
nd
satelli
t
e
d
ata,”
Ge
omor
phology
,
v
ol.
285
,
pp
.
16
–
2
7,
Ma
y
2017.
[17]
M.R.
Mahm
ud,
M.
Hashim
,
and
M.N.
M.
Reb
a,
“
How
eff
ective
is
the
new
g
ene
r
ation
of
GP
M
sat
e
ll
ite
pr
ec
ip
itati
o
n
in
cha
r
ac
t
erizin
g
the
rai
n
fall
v
ari
ab
il
i
t
y
ov
er
Malay
s
ia
?
,
”
in
Asia
-
Pacific
Jou
rnal
of
the
A
tmospheric
Scienc
es
,
vol.
53
,
no
.
3
,
pp
.
375
–
384
,
Aug
.
2017.
[18]
M.
I.
Nad
zri
and
M.
Hashim
,
“
Vali
d
at
ion
of
satel
li
te
pre
c
ipi
t
ation
using
TRMM
re
ce
nt
produ
ct
s,
”
34th
Asian
Conf.
Re
mote
Sens.
20
13,
AC
RS
2013
,
vol.
3
,
pp
.
1992
–
1999,
Jan
.
2013
.
[19]
S.
Jeong,
A.
Ka
s
sim
,
M.
Hong,
and
N.
Saadatkh
ah,
“
Sus
ce
pti
bi
lit
y
assess
m
ent
s
o
f
la
ndslide
s
in
H
ulu
Kela
ng
are
a
using a
g
eogr
ap
hic
informati
on
s
y
stem
-
base
d
pr
edi
c
ti
on
m
odel,”
in
Susta
inabi
l
it
y
,
vol
.
10
,
no
.
8
,
p
p.
2941
,
2018
.
[20]
M.
Dem
bél
é
an
d
S.
J.
Zw
art
,
“
Eva
lu
at
ion
and
com
par
ison
of
satelli
t
e
-
base
d
ra
i
nfa
ll
produ
ct
s
i
n
Burkina
Faso
,
W
est
Afric
a
,
”
in
Inte
rnat
ional
Jo
urnal
of Re
mot
e Sensing
,
vo
l. 37, no. 17, pp. 3995
–
4014,
Sep
.
201
6.
[21]
S.
N.
M.
Za
d
,
Z
.
Zul
ka
fli,
and
F.
M.
Muharra
m
,
“
Sate
ll
ite
r
ai
nf
all
(TRMM
3B42
-
V7)
per
form
ance
assess
m
ent
an
d
adj
u
stm
ent
ov
er P
aha
ng
riv
er
b
asin
Mal
a
y
s
ia,”
in
Re
mote
Sensing
,
vol. 10, no.
3,
p
p.
388
,
Mar
.
201
8.
[22]
K.
C.
T
an,
M.
L.
Ta
n,
J.
Paska,
a
nd
A.
M.
S.
L
au,
“
Eva
lua
t
ion
of
TRMM
3B42V7
produc
t
on
ext
r
eme
pre
c
ipi
t
at
io
n
m
ea
surem
ent
s
o
ver
pen
insula
r
Malay
s
ia,”
Proc
ee
dings o
f
th
e
S
PIE
,
vol
.
10421
,
p
p.
12,
Nov
.
20
17.
[23]
A.
Chen,
D.
Che
n,
and
C.
Azori
n
-
Molina
,
“
As
ses
sing
rel
ia
b
il
i
t
y
o
f
pre
ci
pi
ta
t
ion
d
at
a
over
th
e
Mek
ong
Rive
r
Basin:
A
compari
son
of
ground
-
base
d,
satelli
t
e,
and
re
an
aly
s
is
dat
ase
ts
,”
in
Inte
rnational
Journal
of
Cli
matology
,
vol.
38,
no.
11
,
Jul
.
2018
.
[24]
M.L
.
Ta
n
,
K.C
.
Ta
n
,
V.P
.
Chu
a,
and
N
.
W
.
C
han,
“
Ev
al
ua
ti
o
n
of
TRMM
pr
oduct
for
m
onitoring
drought
i
n
the
Ke
la
nt
an
R
iv
er
Basin
Mal
a
y
s
i
a,
”
in
Wate
r
,
vol
.
9
,
no
.
1
,
pp
.
57
,
Jan
.
2017.
[25]
M.
L.
L
ee,
K.
Y.
Ng,
Y.
F.
Hua
ng,
and
W
.
C.
L
i,
“
Rai
nfa
ll
-
indu
ce
d
l
andsli
des
in
Hulu
Kela
ng
ar
ea
Mal
a
y
s
ia,”
i
n
Natural
Haz
ards
,
vol
.
70
,
no
.
1
,
p
p.
353
–
375
,
Jul
.
2014.
Evaluation Warning : The document was created with Spire.PDF for Python.