TELKOM
NIKA
, Vol.11, No
.1, Janua
ry 2013, pp. 63
~72
ISSN: 2302-4
046
63
Re
cei
v
ed Se
ptem
ber 28, 2012; Revi
se
d No
vem
ber
21, 2012; Accepted Novem
ber 28, 20
12
The Electrostatic Field Networking in Three Isolated
Thunderstorms
Zhenhui Wa
ng*
1, 2,a
, Qing
feng Ze
ng
1, 2,b
, Fengxia G
u
o
2,c
, Dongp
u Xu
3,d
, Hao Wang
2,e
1
Ke
y
L
abor
ator
y of Meteor
olo
g
ical D
i
saster
of Ministr
y
of E
ducati
on,
N
anj
i
ng Un
iversit
y
o
f
Information
Scienc
e & T
e
chno
log
y
, N
anj
i
ng 21
00
44, P. R. Chin
a
2
School of Atmosph
e
ric Ph
ysi
cs, Nanji
ng Un
i
v
ersit
y
of
Infor
m
ation Sci
enc
e &
T
e
chnol
og
y, Na
nji
ng 2
1
0
044,
P. R. China
3
Jiang
yi
n Mete
orol
ogic
a
l Bur
e
au, Jian
g
y
i
n
21
440
0, P. R. China
*corres
pon
di
ng
author, e-mai
l
:
a
eiap@n
u
ist.e
du.cn ,
b
zengq
i
ng0
41
9@si
na.
com,
c
aguo
_fx@
ya
h
oo.com.cn,
b
xudo
ng
pu@
hot
mail.com,
e
w
a
ngh
ao
911
@16
3
.com
A
b
st
r
a
ct
A metho
d
for
netw
o
rking
at
mos
p
h
e
ric
ele
c
trostatic fiel
d
by a
qu
asi-n
o
r
m
a
l
ch
arg
e
d
i
s
t
ributio
n
mo
de
l bas
ed
on ra
dar a
n
d
soun
din
g
dat
a in is
olat
ed
storm ce
lls h
a
s
bee
n pro
p
o
s
ed. T
he ch
ar
ge
distrib
u
tion p
a
r
ameters of thun
dercl
ou
d a
r
e firs
t estima
ted and i
n
ver
s
ed,
and the
n
the netw
o
rk
of
atmos
p
h
e
ric e
l
ectrostatic fiel
d
can b
e
calc
ul
ated w
i
th the
o
b
tain
ed
para
m
eters. T
he
met
hod w
a
s us
ed
to
ana
ly
z
e
thr
e
e
i
s
olate
d
th
und
e
r
storms th
at p
a
ssed
thro
u
g
h
the ex
peri
m
ent
site i
n
200
9. It w
a
s show
n th
at
the e
l
ectrostati
c fiel
d n
e
tw
orki
ng
an
d the
ch
a
r
ge
distrib
u
tio
n
w
e
re co
ncor
d
ant w
i
th th
e l
o
cation
of l
i
g
h
tni
ng
and radar echo. It is revealed
that
the m
o
del
and obtained param
e
ters ar
e reasonable to some extent and
the method for
netw
o
rking e
l
e
c
trostatic field
usin
g radar
an
d soun
di
ng dat
a is feasib
le.
Key
w
ords
:
a
t
m
o
spheri
c
el
ectro
s
tati
c field netwo
rki
ng,
Doppl
er wea
t
her ra
dar, isolated sto
r
m
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Thund
erstorms
are
on
e
o
f
the maj
o
r ca
use
s
of weat
her-rel
a
ted
h
u
man
inju
rie
s
, death
and treme
n
d
ous lo
ss
of
prop
erty. Wit
h
the
rapi
d
developm
ent
of e
c
ono
my and
extensi
v
e
appli
c
ation
of electro
n
ic t
e
ch
nolo
g
y, lightning
haza
r
d
traverse
man
y
http://www.ici
ba.com/inv
o
lv
e/ fields,
su
ch a
s
a
g
ric
u
ltur
e, a
v
iation, infrastru
c
tu
re a
nd
telecom
m
uni
cation
s.
Hen
c
e th
e works of lightni
n
g
monitori
ng,
f
o
re
ca
sting a
nd
d
e
fense have
become in
cre
a
sin
g
useful. Lightnin
g
acti
vity of
ten cau
s
e
s
a sig
n
ificant cha
nge o
f
EF near the
grou
nd. The
EF data can
reflect the charg
e
s
of th
unde
rcl
oud q
u
ickly and it is an impo
rta
n
t
para
m
eter u
s
ed fo
r
cha
r
g
e
dist
ributio
n
rese
arch
an
d
lightning
fore
ca
sting. S
c
ho
lars (e.g.
[1-2
])
attempted to
utilize the EF
data to
cha
r
g
e
dist
ri
bution
resea
r
ch
in
th
unde
rcl
oud
s. Other schol
ars
made
use of
the EF thresh
olds [3, 4], 0-1 rel
a
tion
s of
cu
rve q
u
ick j
i
tter [5] and
p
o
larity reversal
[4] as metho
d
s fo
r li
ghtni
ng n
o
wca
s
tin
g
. Some
sch
o
lars
also tri
ed to
fore
ca
st lightni
ng
by
combi
n
ing EF
data with lig
htning lo
catio
n
[6] or
ra
da
r
data [7]. However mo
st of them have
used
EF data from
a sin
g
le-stati
on. Thi
s
data
can
only refl
ect the
cha
r
g
e
situatio
n in
a sm
all re
gi
on
around the
field mill and cannot
show the
space and time
characteri
stics
of a whole
thunde
rsto
rm
. Therefore
i
f
EF can
be
netwo
rked,
it is po
ssible
to enh
an
ce
the mo
nitori
ng
function
an
d
enla
r
ge
the
dete
c
tion
ra
nge
of the fi
eld mill. T
h
e
mo
st comm
on meth
od f
o
r
netwo
rki
ng E
F
is sp
atial interpol
ation
usin
g
limited data. Due to
the eco
nomi
c
and manp
o
w
er
factors, the si
te setting of field
mills are sparse and discrete in
the observation region. There can
be
som
e
d
e
viation b
e
twe
e
n
net
wo
rked
EF with
in
terpolation
met
hod
s a
nd
act
ual o
b
servati
on
data. Since
EF data ca
n
reflect the
charg
e
situ
ation in a thun
derclou
d and
the size of the
cha
r
ge
d thu
n
derclou
d
can
be d
e
scribed
by ra
dar dat
a, we can try
to net
work the EF data base
on the
existin
g
cha
r
ge
stru
cture
wi
th thi
s
d
a
ta. Thi
s
work therefore p
r
e
s
ent
s a
n
an
alysi
s
of
EF
netwo
rki
ng u
s
ing rada
r an
d sou
ndin
g
d
a
ta bas
ed on
a qua
si-n
orm
a
l cha
r
ge di
st
ribution m
ode
l.
Evaluation Warning : The document was created with Spire.PDF for Python.
TEL
K
64
2. O
b
st
ud
y
with
N
is N
a
obse
r
Yan
g
mill
e
mill
m
st
at
o
r
prop
o
the s
t
11
-1
0
The
A
are s
A
t la
s
after
field
done
two
p
inten
s
plate
s
simu
l
data
the i
n
the d
and
v
dete
c
effec
t
longi
t
sit
u
a
t
pape
whi
c
h
con
s
i
com
p
of fie
l
K
OM
NIKA
V
b
serv
a
tion
m
More t
y
the law
an
d
N
a
n
jing new
a
nji
ng weat
h
r
vation a
r
ea
g
tz
e River
D
e
The E
F
e
lect
ric f
i
eld
m
ea
su
re
s E
F
r
which is
p
o
rtio
nal to E
F
t
ator. T
h
e al
t
0
-12
V and he
n
A
C voltage
s
e
n
t to a
pha
s
s
t the voltag
going th
rou
g
The a
m
E is
as
foll
o
u
s
ing th
e r
e
p
lates). Firs
t
s
ity bet
wee
n
s
and
t
he
p
l
atio
n of
ele
c
con
s
ist
s
of
o
n
stallation o
f
ata is finally
The E
F
v
ice versa.
D
c
ted in a ran
t
on EF mea
t
ud
e [118.6
0
t
ed i
n
a
n
o
p
r. Sec
o
n
d
ly
,
h
i
s
r
e
st
ri
ct
e
d
i
stent pa
tter
n
p
ared, a
n
d
t
h
l
d mill is ab
o
ol. 11, No
. 1
m
ill and
dat
a
h
an 2
5
field
d
ch
ara
c
t
e
r
i
s
g
ene
rati
on
D
h
er rada
r,
it
s
is located i
e
lta.
Figure 1. Lo
c
F
me
ter
s
a
r
e
mete
r (it i
s
w
F
c
h
a
n
ges
th
p
ro
du
ced b
y
F
. The EF v
a
t
ern
a
ting cu
r
n
ce it is am
p
s
ignals as
w
s
e sen
s
iti
v
e
e sign
al whi
g
h low-pa
ss
f
m
plification i
s
o
ws
: U=kE,
w
e
lation betw
e
t
, s
o
me
v
o
lt
a
n
the two pl
p
ote
n
tiom
et
e
c
tric
field is
t
h
o
rdi
nal, me
a
s
f
field mills,
d
stored si
mu
l
F
i
s
de
fined
D
ep
endin
g
o
ge
c
l
os
e
to
2
surement
s t
w
0
-1
19.25E
],
p
e
n
a
nd fl
at
f
the EF wa
s
d
by gl
oba
l
a
n
[9, 10].
B
e
h
e
re
sult
s
sh
o
o
ut 1V/m a
n
d
, Janua
ry 2
0
a
mills have
b
t
i
c of
E
F
sp
a
D
o
pple
r
we
a
s
scan tim
e
n a low-lyi
n
g
c
ation of EF
e
of two
kind
s
w
ritten as
fi
e
roug
h a
win
d
y
the rotatio
n
a
lue can
co
n
r
re
nt sign
al
o
p
lified by I-V
w
ell as the s
y
detecto
r a
n
d
ch i
s
up
t
h
e
f
iltering, volt
a
s
linea
r
hen
w
he
re k
i
s
a
e
e
n
voltage
a
a
ge i
s
add
e
d
ates is
kno
w
e
r i
s
adj
ust
e
h
e sam
e
.
T
h
s
ur
e
d
va
lu
e
s
d
ata i
s
t
r
an
s
m
l
taneo
usly
o
n
he
re
as the
o
n the ma
g
n
2
0
k
m
an
d i
t
w
o
id
ea
s
ar
e
which inst
a
f
ield with
n
o
s
u
n
ified
an
d
a
nd lo
cal
dai
l
e
side
s, the
o
w that the
s
the accu
ra
c
0
13 : 63 – 7
2
b
een i
n
sta
l
l
e
a
ce
-time v
a
ri
a
ther rad
a
r s
t
is 6minute
s
g
plain a
nd
and rad
a
r
s
t
s
; ball
oon
-b
o
e
ld mill in th
e
d
mill-type d
e
n
of a mov
i
n
se
que
ntly
b
o
n the sen
s
o
conversio
n
c
y
nchro
n
izi
n
g
d
the
pola
r
it
y
r
e
qu
ir
e
m
en
a
ge reg
u
lati
o
c
e
the relati
a
cons
tant.
T
a
nd elect
r
ic
f
d
on
tw
o
pa
w
n. Based
o
e
d until the
h
e EF data i
s
s
, a
c
qui
sition
m
itted throu
g
n
two c
o
mp
u
ne
gative
c
h
itude a
nd l
o
t
s re
spo
n
se
e
used. First,
a
lled in sta
o
h
i
gh
ob
jec
t
d
identified
o
y variation
m
rationality o
s
ta
nd
ar
d
d
e
v
c
y of EF is
e
n
2
e
d in
Nanjin
g
ation. Th
e p
o
t
ation a
r
e s
h
s
and the d
e
a thunde
rst
o
t
ations in ex
p
o
rn tw
o
sph
e
e
pape
r)
wa
s
e
tec
t
ion prob
i
ng film is
c
b
e cal
c
ulate
d
o
r chi
p
is v
e
r
y
c
ircuit and
c
h
g
signal
s pr
o
y
of the EF is
t of a
nalog
-
t
o
n and the o
f
on b
e
twee
n
T
he
field mi
l
f
ield U = Ed
rallel
plate
s
o
n this prin
c
output of
s
packa
ged i
n
time, motor
g
h a
wi
rele
s
s
u
ters
.
h
a
r
ge
if th
e
c
o
cat
i
on of
ch
time is
1 s
.
The EF mill
ndard met
e
t
s aro
u
nd
t
h
e
o
n 0.13
kV/m
m
ech
ani
sm
a
f c
a
librated
v
iation
of 8 s
t
n
ough for th
e
g
and it
s su
r
o
sit
i
on
s of
t
h
h
own in Fi
g.1
e
tection ra
n
g
o
rm pron
e a
p
eri
m
ent sit
e
re
s a
nd the
f
s
use
d
in thi
e. First, the
c
onverted in
d
by measuri
y
w
e
ak
re
ac
h
h
ange
d into
A
o
d
u
ce
d by p
h
identified
u
s
t
o-digital
co
n
f
f
s
et ci
rcuit.
voltage sig
n
l
l bas
i
c cali
b
(d is
the dis
t
a
n
d fr
om t
h
c
iple, the de
v
t
he ele
c
tri
c
n
to a frame
a
speed and
c
s
co
nne
ct
io
n
c
ha
r
g
e
o
f
th
e
a
r
ge, the
cl
o
To we
aken
at latitude [
3
e
o
r
ologi
cal
o
e
m, has be
e
[8] based
o
a
n
d
the me
a
n
data are al
s
t
a
t
ions i
s
s
m
e
netwo
rki
n
g
ISSN: 230
2
r
roundi
ng a
r
e
h
e
EF mill to
g
. The dot in
g
e
is 230k
m
re
a kno
w
n
a
e
.
f
ield mill. Th
s p
a
per.
T
h
e
in
duced ch
a
to
voltage
t
ng this volt
a
h
ing a value
A
C voltage
s
h
otoele
c
tric
s
s
ing this
det
e
n
ve
r
t
e
r
i
s
ob
t
n
al U
an
d
e
b
ration p
r
in
c
t
an
c
e
be
tw
e
e
h
e form
ula
t
h
v
ice is put
i
fi
eld devi
c
e
a
nd
e
a
c
h
fr
a
c
he
ck code
s
n
to
a
r
e
c
e
i
v
e
e
cl
oud i
s
p
o
o
u
d
field co
u
th
e enviro
n
m
3
1.85-32.2
0
N
o
bservatori
e
s
e
n sel
e
ct
e
d
o
n t
h
e fair-w
e
n
va
lu
es
oc
c
u
s
o measure
d
m
all. The res
o
work
.
2
-4
046
e
as
to
g
eth
e
r
Fi
g. 1
m
. The
a
s t
h
e
e fie
l
d
e
field
a
r
ge o
f
t
hat is
a
ge on
of 1
0
-
s
i
g
nal.
s
wit
c
h
e
ct
ion.
t
ained
e
le
c
t
r
i
c
i
p
le
is
e
n the
h
e EF
i
n tw
o
e
an
d
a
me of
. After
e
r and
o
sit
i
v
e
u
ld be
m
en
tal
N
] and
s
an
d
i
n th
is
e
ath
e
r
u
r i
n
a
d
and
o
lution
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Elect
r
ost
a
tic Field Networkin
g
in Th
ree Isol
ated T
hund
erstorm
s
(Zhen
hui Wa
ng)
65
3. Methodol
og
y
descriptions
The
cha
r
ge i
n
the thun
de
rsto
rm
can
b
e
refle
c
t by
EF data, whi
l
e the si
ze
o
f
the
cha
r
ge
d thu
n
derclou
d
can
be d
e
scribed
by ra
dar dat
a
,
therefo
r
e
th
e mai
n
cha
r
g
e
pa
ram
e
ters of
thunde
rcl
oud
can
be o
b
tain
ed ba
se
d on
EF, rada
r, so
undin
g
an
d the othe
r info
rmation throu
g
h
the model, a
nd then the
spa
c
e EF ca
n be re
ca
l
c
ul
ated and n
e
tworke
d acco
rding to the
s
e
obtaine
d ch
arge paramete
r
s.
3.1. Quasi-n
ormal charg
e
distributio
n model
In pra
c
tice, some mete
oro
l
ogical elem
e
n
ts, su
ch
a
s
temperature
and p
r
e
c
ipit
ation,
have the feat
ure in
som
e
frequ
en
cy
dist
ribution, i.e., the freq
uen
cy
is greate
s
t in its avera
ge a
nd
the frequ
en
cy of both sid
e
s will
de
cre
a
se g
r
a
dua
lly
. So many factors, su
ch
a
s
environme
n
t
a
l
temperature,
particl
e m
o
ving
spe
ed, p
a
r
ticle
si
ze,
fri
c
tion
rate
s
a
nd the
effect
of
coag
ulati
on,
colli
sion, hyd
r
ometeo
rs i
n
d
i
fferent pha
se
s (su
c
h
as li
q
u
id, crystal, snow
and
gra
u
pel),
will ma
ke
a different co
ntribution to the pro
d
u
c
tio
n
and di
st
rib
u
tion of cha
r
ges. In additi
on, the transfer
from po
sitive
ions to
negati
v
e and vice v
e
rsa in el
ec
t
r
ification p
r
o
c
e
ss i
s
st
ocha
stic. The
r
efore, it
is quite
difficult to describ
e the
cha
r
ge
distrib
u
tion in
a thund
erclo
ud. T
he
ch
arge di
stributio
n of
thunde
rcl
oud
s i
s
ta
ken
a
s
the
re
sults p
r
odu
ce
d by
a
num
bers
of
“ra
ndom
” fa
ct
ors h
e
n
c
e it i
s
descri
bed
with norm
a
l or q
uasi
-
no
rmal f
unctio
n
in
this study. The
cha
r
ge di
stri
b
u
tion of a sin
g
le-
polar
cha
r
g
e
d
mass in term
s of Gau
ssi
an
function can
be expre
s
sed
as:
(1)
Whe
r
e A (
0
x
,
0
y
,
0
z
) is ch
arg
e
co
n
c
entration ce
nter of ch
arg
ed mass,
0
is the cha
r
ge d
e
n
s
ity
at this point a
n
d
,
,
are pa
ra
meters for th
e cha
r
ge d
e
n
s
ity that decrease with di
stance
from the ce
nter, whi
c
h is
si
milar to the st
anda
rd deviat
i
on in Gau
s
si
an functio
n
.
Acco
rdi
ng to the kno
w
le
dg
e of physics,
becau
se
of the balan
ce eff
e
ct of total electri
c
field, the ind
u
ce
d charge
will app
ear
on the g
r
ou
n
d
wh
en a th
unde
rsto
rm o
c
curren
ce. T
he
electri
c
field
near th
e grou
nd is the j
o
int
functi
on of t
he charge in
the clo
ud an
d
on the g
r
oun
d.
The ele
c
tri
c
field measured nea
r the grou
nd is
el
ectro
s
tati
c field and ca
n be de
scribe
d
by
Coul
omb la
w and mirror i
m
age. The
EF intensity
at point (x, y, z) arou
nd
the single
-
p
o
lar
cha
r
ge
d clo
u
d
woul
d be
(2)
Whe
r
e
,
is an
atmosphe
ric
diele
c
tric
co
n
s
tant an
d pa
rameters
,
,
are the
limi
t
s for i
n
tegration calculatio
n wh
ich
sta
n
d
for th
e 3
-
di
mensi
onal
si
ze
of the
cha
r
ge
d ma
ss an
d is e
s
ti
mated from t
he ra
dar
ech
oes of
the
thunde
rcl
oud. The
po
sition of
cha
r
ge
de
nsit
y cente
r
A
(
,
,
)
and
charge
distrib
u
tion
p
a
ram
e
ters
(
,
,
)
are
al
s
o
estimated a
c
cording to the
rada
r echo feature
s
of clo
ud.
Whe
n
a cha
r
ged cl
oud m
o
ve with a certain s
pee
d to a mill station, the EF at this
station would
chan
ge with
time and ca
n be cal
c
ulat
ed usi
ng Eq. (2) by igno
rin
g
the sho
r
t time
variation
s
of
cha
r
ge
in th
e
mature, stab
le t
hund
erclo
ud. In the
me
antime, Eq. (2)
can
be
used
for cal
c
ulatin
g the EF cau
s
ed by a cl
ou
d with mu
lti-p
o
lar cha
r
ge
structu
r
e afte
r modificatio
n
.
The charge
structu
r
e of th
unde
rcl
oud i
s
not
singl
e-p
o
lar b
u
t com
p
licate
d
(muti
-
pola
r
)
in practi
ce,
and it i
s
n
e
c
e
s
sary to
u
s
e
som
e
ch
arge
st
ru
cture in the
inv
e
rsi
on
of ch
arge
para
m
eters. Krehbi
el
[11]
found that alt
houg
h the he
ights of
the
main ne
gative cha
r
g
e
re
gion
were differen
t
in three
different
are
a
s, t
hey we
re i
n
t
he same te
m
peratu
r
e
zo
n
e
. Although t
h
e
cha
r
ge
stru
ct
ure of thun
de
rclo
ud i
s
co
m
p
lex, the
main cha
r
g
e
stru
cture
ca
n de
scribe
as di
p
o
le
or tri
pole
in t
he mai
n
cha
r
ge a
r
ea
[12
-
15] in
a m
a
tu
re
conve
c
tion
zo
ne. T
he
n
u
meri
cal
re
su
lts
[16] of three
region
s i
n
Chi
na a
nd [1
7]
in
Chi
n
e
s
e plat
eau re
gion
al
so co
nfirme
d this con
c
lu
sio
n
and foun
d th
at the cha
r
ge
stru
cture ca
n be de
scri
b
ed as
dipole
or tripol
e cha
r
ge
stru
cture
in
mature
sta
g
e
.
The
cha
r
g
e
stru
cture research i
n
Nanj
ing i
s
few an
d there a
r
e
n
o
othe
r m
ean
s to
22
2
00
0
0
1
(
,
,
)
e
x
p
[
(
(
)(
)(
)
)
]
2
xx
y
y
z
z
xy
z
xy
z
x
y
z
00
0
00
0
22
2
2
2
2
00
0
0
0
0
/2
/2
/2
0
22
2
3
22
2
/2
/2
/
2
''
'
'
'
'
11
e
xp[
(
(
)
(
)
(
)
)
]
e
xp[
(
(
)
(
)
(
)
)
]
22
(,
,
)
[
4
[
(
')
(
'
)
(
')
]
[
(
'
)
(
'
)
(
'
)
z
z
yyx
x
zz
y
y
x
x
xx
y
y
z
z
xx
y
y
z
z
rr
xy
z
x
y
z
Ex
y
z
xx
y
y
z
z
xx
y
y
z
z
3
]'
'
'
]
dx
dy
dz
[
(
')
,
(
')
,
(
')
]
rx
x
y
y
z
z
x
y
z
0
x
0
y
0
z
x
y
z
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 1, Janua
ry 2013 : 63 – 72
66
detect di
re
ctl
y
. The simpl
e
tripole
stru
cture
is
us
ed
in this m
o
d
e
l, i.e. the u
pper po
sitive, the
central
neg
ative and
the
l
o
we
r
po
sitive ch
arge
(if
th
e lo
we
r p
o
siti
ve ch
arge i
s
less o
r
ze
ro,
it
become
dipol
e st
ru
cture
)
.
The
ration
ality of re
su
lt
s
will be ve
rified
by the in
dep
e
ndent EF
dat
a,
lightning lo
cat
i
on data an
d the inversion
cha
r
ge d
e
n
s
ity.
3.2. EF net
w
orking
It is shown from Eq. (2) th
at if the di
sch
a
rg
e
di
stributi
on (fu
n
ctio
n
Eq. (1) )
and
the si
ze
s
feature
of thu
nderclo
ud i
s
kno
w
n,
the
space EF di
stribution n
ear
grou
nd can
b
e
cal
c
ul
ated with
the model. As mention
ed
above, the p
a
ram
e
ters
,
,
,
,
,
,
,
,
can be
estimated fro
m
the differen
t
types of data. T
he EF dat
a is obtai
ned
by obse
r
vatio
n
and the
r
efo
r
e
only para
m
et
ers
0
of the three layers are unkno
wn. Thi
s
ca
n be
cal
c
ulated from t
h
ree
or mo
re
than three E
F
value
s
. Th
e pa
ramete
rs of cha
r
g
e
d
ensity is i
n
ve
rse
d
a
s
follo
ws, first, the
EF
values of diff
erent
layers i
n
si
mple
tripo
l
e st
ru
ct
ure
o
f
inversion
st
ations are
cal
c
ulate
d
with t
he
cha
r
ge d
e
n
s
ity of 1C/km
3
and then the
rea
s
on
able
charg
e
den
sity range (0-30
C
/km
3
) a
nd step
(0.001
C/km
3
) of ea
ch
laye
rs are
set to
be
comp
osed
of differe
nt
charg
e
d
e
n
s
ity com
b
inatio
n
s
.
Finally the tot
a
l erro
r p
e
rce
n
tage
and
sta
ndard d
e
viation
b
e
twee
n simulated and
observed
EF of
inversi
on stations
u
nde
r
different cha
r
ge den
sity
combi
nation
s
are calcul
ated ci
rcularly u
s
ing t
he
least
sq
ua
re
metho
d
. T
he
smalle
st
erro
r p
e
rce
n
tage
of ch
arge
de
nsity
co
mbinatio
n
s
i
s
con
s
id
ere
d
a
s
optimum. If the error p
e
rcenta
ge bet
ween si
mulate
d and o
b
serv
ed EF whi
c
h
is
greate
r
th
an 100% rea
c
h 60%
in 6
mi
n,
the
p
a
ra
m
e
ters
,
,
,
,
and
will
be
adju
s
ted
and
simul
a
ted
a
gain. To
verif
y
the si
m
u
lat
i
on p
a
ra
mete
rs
obtain
ed, t
he EF
of oth
e
r
station a
r
e simulated with
the model
and t
he obta
i
ned pa
ram
e
ters
and
com
pare
d
with t
h
e
observation
d
a
ta. After a different adju
s
t
m
ent, if
there
are no
para
m
eters satisf
ying the abo
ve
requi
rem
ent or
the error b
e
twee
n
test result
s
an
d
ob
servatio
ns i
s
t
oo large,
this simulatio
n
of th
e
stage i
s
con
s
ide
r
ed
as
a
failure. If the EF pa
ra
meters is i
n
verse
d
successfully, the EF
netwo
rki
ng
n
ear gro
und will
be re
cal
c
ulated.
W
hen
the ob
se
rvation site
s a
r
e
adeq
uate, tim
e
synchro
n
ization data
of di
fferent
site
s
are
used in
the solution
o
t
herwi
se
the
data obtai
ne
d at
different time intervals from
the same
site are u
s
ed.
4. Case an
aly
s
is and results
The i
s
olate
d
storm
cells of
July
15
(20
0
9071
5), Au
gu
st 14
(2009
0
814) an
d Au
g
u
st 2
4
(200
908
24) i
n
2009
which passe
d throug
h the re
sea
r
ch re
gio
n
s were a
n
a
l
yzed. The
EF
measurement
may be affe
cted by ch
arg
e
d
pre
c
ipit
atio
n and
hen
ce
may not refle
c
t the cha
r
ge
of
the thunde
rcl
oud. The
r
efo
r
e the EF
data
unde
r strong
rada
r e
c
ho i
s
not in
clud
e
d
in the process
of inversio
n a
nd interp
olati
on. For in
sta
n
ce, NJXJ
st
ation whi
c
h
was covere
d b
y
radar e
c
h
o
was
exclud
ed in
the ca
se of 2009
0814. T
he slo
w
EF
variation
s
(3
00s ave
r
ag
e
value) of time
synchro
n
ization data
in diff
erent
site
s was u
s
e
d
in th
e 200
908
14
and 2
009
082
4 ca
se,
while
the
data from different time intervals in the sam
e
site was u
s
e
d
in the 20090715 case. The
variation
s
of cha
r
ge di
stri
bution in the
time in
terval of 10s is ignore
d
, i.e., a
charge den
sity
combi
nation i
s
calculated
every 10
s. Beca
use of
the
simple
tripol
e structu
r
e u
s
ed i
n
the m
odel
,
the isol
ated
cells a
nd th
e close
d
sta
ge
o
f
the
ra
dar-e
cho tightly are
anal
yzed
an
d networke
d
i
n
this pap
er.
4.1. The cas
e of 200
908
1
4
A thunde
rsto
rm o
c
curre
d
and
develop
ed a
r
ou
nd
NJXJ statio
n a
t
about
074
8
UT
C
August 14, 2
009. The
r
e were 8 syn
c
hroni
ze
d EF
observation
stations (NUIS
T
, NJQL, NJXJ,
NJTQ, ZJJR, NJJP, NJSH
and Z
J
EL)
at the experim
e
n
t
site. To av
oid the effect
of pre
c
ipitatio
n,
NJX
J
station
whi
c
h was
co
vered by ra
d
a
r echo was
exclud
ed wh
en inversion
and net
worki
ng.
The EF ob
servation of
NJQL, NJT
Q
and
NJLD
wa
s the inve
rsio
n data,
while
NJPK wa
s
indep
ende
nt test data in this ca
se. Before inversi
on,
first, with NJXJ station as th
e origin an
d the
increa
se in the dire
ction
s
of longitude
and latit
ude
as X, Y axis se
parat
ely, the coordina
tion
system was
built up, and then the
information of latitude and lo
ngi
tude, su
ch as the site as well
as the thun
de
rsto
rm cl
oud
path we
re co
nverted into
Carte
s
ia
n co
ordin
a
tes.
x
y
z
0
x
0
y
0
z
x
y
z
x
y
z
0
x
0
y
0
z
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Elect
r
ost
a
tic Field Networkin
g
in Th
ree Isol
ated T
hund
erstorm
s
(Zhen
hui Wa
ng)
67
4.1.1. Parameter s
e
lectio
n
Takin
g
into account the charg
e
stru
ctu
r
e
re
sea
r
ch [18,19], the numeri
c
al si
m
u
lation
results
[16] in China as
well as
the
Chines
e inland sounding exper
iments [17],this
s
t
udy
sele
cted
the
regio
n
b
e
lo
w-5
as the
lowe
r p
o
sitiv
e
charge
reg
i
on of thu
n
d
e
rsto
rm
and
the
cente
r
wa
s a
bout 0 ; the
regi
on
-5
~ -20
as t
he ne
gative
cha
r
ge
re
gio
n
and
the
ce
nter
height was
a
bout -1
0 ; the re
gion
-2
0 ~ -
40
as up
pe
r po
sitive ch
arg
e
regio
n
and t
h
e
cente
r
h
e
ight
wa
s a
bout
-30 . Ambi
ent tempe
r
at
ure
height
which th
e
stu
d
y req
u
ire
d
wa
s
estimated
b
a
s
ed
on
the
correspon
ding
so
undi
ng
da
ta. Taki
ng th
e p
r
o
c
e
s
s of
0812
UTC a
s
an
example to e
x
plain ho
w to determi
ne th
e
param
eters of the model required.
Figure 2. The radar CAPPI at 0812 UT
C on
August14,2009. (a)5km; (b) 7km;(c) 9km.
Accordi
ng to the
sounding dat
a,
the
radar
CAPPI (Constant Altitude Plan P
o
sition
Indicatin
g
) of
5, 7, 9 km can well re
pre
s
ent
the
rada
r e
c
ho
of lower,
ce
ntral and
up
per
respectively. Fig.2 i
s
showing
the
radar CAPPI of
5, 7, 9
km
at 0812
UT
C on
August
14, 2009.
The di
stributi
ons of statio
ns in Fig.2 a
r
e con
s
i
s
tent with Fig.1 an
d t
he box around the e
c
h
o
is
ech
o
cent
rali
zed
a
r
ea. T
h
e pa
ram
e
ters
0
x
,
0
y
,
and
of different laye
rs we
re
determine
d by
the center an
d the
scale
of echo
cent
rali
zed
area
(g
re
ater tha
n
2
5
dBZ in lo
we
r
cha
r
ge
regio
n
;
greate
r
th
an
15 dBZ
in u
p
per
and
central re
gion
) of
corre
s
p
ondin
g
hei
ght CAP
P
I respe
c
tively.
The inte
gratio
n-limits pa
ra
meter
o
f lower po
sitive
charge
regio
n
we
re dete
r
min
ed
by the e
c
h
o
thickne
ss
whi
c
h the e
c
h
o
wa
s greater t
han 25
dB
z b
e
low 6.5
k
m
(environ
menta
l
temperatu
r
e
wa
s
about -5
). T
he pa
ramete
r
o
f lowe
r po
sitive cha
r
ge re
gion was a
b
o
u
t 4.5km. Th
e integration
limits pa
ram
e
ter
o
f cent
ral
and
upp
er
cha
r
ge
re
gio
n
were
dete
r
mined
by ra
dar
echo
an
d
sou
ndin
g
te
mperature
p
r
ofile. The
p
a
ra
m
e
ter
o
f central
neg
ative ch
arge
region
（
-5
~-
20
）（
wa
s a
b
out 2.5
k
m
while u
ppe
r p
o
s
itive cha
r
ge
re
gion
-20
~-
40
）
wa
s abo
ut 3
k
m.
The
para
m
eters
,
were d
e
termin
ed by the scale of
ech
o
centrali
zed a
r
ea whi
c
h
wa
s greater
than 30 dBZ
(lower
cha
r
g
e
regio
n
) o
r
2
0
dBZ (u
ppe
r and central region
). As for the paramet
er
, many re
sea
r
ch
es [14,15,
17,20]
poi
nte
d
out that it range
d from
h
undred
s of m
e
ters to 2
k
m.
The pa
ramet
e
r
o
f upper a
n
d
cent
ral re
gi
on wa
s value
d
1.5km in th
is pap
er. The
param
eter
of
lower po
sitive
ch
arg
e
reg
i
on wa
s
d
e
termined
by e
c
h
o
thickn
ess th
at the e
c
h
o
was
gre
a
ter
than 3
5dBZ
belo
w
6.5
k
m
,
and
it was abo
ut 3.5
k
m. In ad
ditio
n
, the
echo
thickne
s
s in
the
simulatio
n
proce
s
s chan
g
ed little, so
the paramet
ers
0
z
,
,
were co
nsid
ere
d
to
be the
x
y
z
z
z
z
x
y
z
z
z
z
z
(a
)
(b)
(c)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 1, Janua
ry 2013 : 63 – 72
68
same. Th
e pa
ramete
r
0
z
of different laye
rs
wa
s the cent
ral
height of these laye
rs a
n
d
determi
ned
by the Nanji
ng statio
n so
undin
g
temp
erat
u
r
e p
r
ofil
e. Mean
while
, the param
e
t
ers
,
,
,
in the adja
c
e
n
t rada
r scan
time is con
s
id
ered to be
ch
ange
d linea
rl
y in inversion
cal
c
ulatio
n.
4.1.2. The simulated res
u
lts and co
n
t
ras
t
analy
s
is
The thund
erstorm pro
c
e
ss betwee
n
08
12-0
830
UT
C was
simulat
ed and the result
s
are
sho
w
n i
n
fig 3. The
station
s
of
NJQL,
NJT
Q
and
NJLD
were inve
rsi
on data, a
n
d
the
measured an
d simulate
d EF are in fig 3a,b.
The NJPK was inde
pend
ent test station, and the
measured a
n
d
simul
a
ted E
F
values
we
re sh
own in
Fi
g 3a. The
si
mulated EF o
f
test station
wa
s
cal
c
ulate
d
wi
th the mod
e
l
and the
o
b
tained
par
a
m
eters. The
coeffici
ent
of NJPK
sta
t
ion
betwe
en
me
asu
r
ed
a
nd
simulate
d EF
wa
s 0.86.
T
he inve
rsion
and i
nde
pen
dent te
st EF
all
indicated that
the obse
r
vat
i
ons a
nd si
m
u
lated EF we
re in go
od a
g
ree
m
ent. It wa
s sp
eculat
e
d
that the mod
e
l
and the
obta
i
ned p
a
ra
met
e
rs were
rati
o
nal to
some
e
x
tent. Fig 3c i
s
the
evolutio
n
of
ch
arg
e
cen
t
er
d
ensity of this ca
se.
It
was exhi
bited
that the
neg
ative ch
arg
e
d
e
n
sity of
cent
ral
regio
n
h
ad t
he la
rge
s
t v
a
lue
while
p
o
sitive
cha
r
g
e
de
nsity of
lower
and
uppe
r
we
re
small
relatively. Especi
a
lly, the positive
cha
r
ge de
nsity
of
lowe
r laye
r
wa
s small (m
ost of them
were
zero)before 0
824
UT
C an
d
the stru
ctu
r
e
wa
s dip
o
le a
t
this sta
ge,
whi
c
h
corre
s
pond
ed
with the
developm
ent
of thunderst
orm; After 0
824 UTC th
e
cha
r
ge d
e
n
s
ity of lower layer incre
a
s
ed
grad
ually an
d the structu
r
e assum
ed a
tripole
cha
r
g
e
stru
ctu
r
e, whi
c
h corre
s
pond
ed
with the
mature
d sta
ge of thund
ersto
r
m. Me
anwhile, li
ght
ning lo
cation
data indicated that the two
negative fla
s
hes o
c
curred
at a
bout
08
25
UT
C
(t
he
erect
dotted
line
in th
e
Fig 3
c
and
t
he
measured EF
in Fig 3a,b)
.The ch
arg
e
stru
cture wa
s tripole when
CG lightnin
g
occurred. Th
e
EF netwo
rki
ng characte
ristics of 08
18 an
d
082
5 UT
C of
different cha
r
ge
stru
cture
of
thunde
rsto
rm
resp
ectively were analy
z
e
d
in this pap
e
r
.
Figure 3. Measu
r
ed a
nd si
mulated EF a
nd ch
arg
e
distribution on A
ugu
st14,20
09
. (a) Mea
s
u
r
e
d
and sim
u
late
d EF of NJQL
and NJPK; (b) Mea
s
u
r
ed
and sim
u
late
d EF of NJL
D
and NJTQ; (c)
The ch
arge e
v
olution on Augu
st14,20
09
. T
he cha
r
ge
den
sity is absolute value.
Figure 4a
an
d b a
r
e
sho
w
ing th
e val
ues
of EF n
e
tworkin
g
wh
ich
we
re o
b
tained
by
cal
c
ulatin
g wi
th the mod
e
l
at 0818
and
0825
UT
C re
sp
e
c
tively. The rada
r e
c
h
o
in this figure is
rada
r PPI (pl
an po
sition i
ndicating)
at 14.6° el
evation (the
sa
m
e
belo
w
). T
h
ere
we
re n
o
CG
x
y
x
y
(a
)
(b)
(c)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Elect
r
ost
a
tic Field Networkin
g
in Th
ree Isol
ated T
hund
erstorm
s
(Zhen
hui Wa
ng)
69
lightning
o
c
curred
at 08
18
UT
C
and
Fig
u
re
3
sh
owed
that the
charge
st
ru
cture of
thund
erclo
u
d
wa
s dip
o
le.
Becau
s
e
the
cha
r
ge
struct
ure
wa
s di
pol
e as well a
s
the neg
ative
clou
d cha
r
ge
of
central layer
wa
s larg
e, the EF in radar echo surrou
nding a
r
ea h
ad a larg
e po
sitive value, such
as
Figu
re
4a.
The
LLS
disclo
sed
that th
ere
we
re
two
neg
ative CG
lightnin
g
fla
s
hes o
c
curred
at
0825
UT
C. In
the me
antim
e, Figu
re 3
c
showed that
th
e ch
arge
stru
cture
chan
ge
d from
dipol
e
at
0818
UTC to
tripola
r
before the lightnin
g
happ
ened.
Except for the positive EF
region
simila
r to
0818
UT
C, there
app
ea
red a n
egativ
e EF area
a
r
oun
d the
ra
dar e
c
h
o
ce
nter
which was
resulted f
r
om
the p
o
sitive
cha
r
ge
s i
n
lo
wer pa
rt
of t
he thu
nde
rcl
oud, a
s
sh
own in
Figu
re
4
b
,c.
Figure 4c i
s
the spa
c
e-d
i
stributio
n of EF net
work at 0825 UT
C, whi
c
h ca
n exhibit the EF
distrib
u
tion visually. It is exhibited that the
positive cha
r
ge a
r
ea
wa
s small an
d therefore the
negative
EF
regio
n
wa
s a
l
so sm
all.
Fig
u
re
4d
sh
o
w
s the
cha
r
ge
den
sity of
di
fferent laye
rs in
thunde
rcl
oud
whe
n
the li
g
h
tning o
c
cu
rred, the
filled
cont
our ma
p in
colo
r i
s
negative
cha
r
ge
den
sity of ce
ntral laye
r, th
e red
and
bl
ack
cont
o
u
rs we
re
po
sitive charge
den
sity of lo
we
r
and
uppe
r laye
r.
The
ch
arg
e
d
ensity di
strib
u
tion in
figure
wa
s co
mpa
r
ed with rada
r
echo and
it wa
s
found that the cha
r
ge
cent
er den
sity
wa
s coi
n
ci
dent
with the rad
a
r echo. It also
sho
w
e
d
that the
CG li
ghtning
occurre
d
at t
he pla
c
e
bet
wee
n
the
ce
n
t
er of n
egativ
e charge i
n
central l
a
yer
a
n
d
positive charge in lower la
yer, whi
c
h in
dicate
d that
locatio
n
of lightning was in
accordan
ce
with
cha
r
ge cente
r
density.
Figure 4. The EF netwo
rki
ng in the ca
se of
20090
81
4.(a) T
he EF netwo
rki
ng wi
th model at
0818
UTC; (b
) The EF net
workin
g with
model at 082
5 UTC; (e) Th
e EF distributi
on at 0825
UTC;
(f)Th
e
charg
e
dist
ribu
tion at 0825
UTC. Th
e ch
arge d
e
n
s
ity is ab
solute val
ue.
4.2. The cas
e of 200
908
2
4
and 200
90
715
Usi
ng th
e
sa
me meth
od
as
used i
n
t
he
ca
se
200
9081
4, the
case
20
090
82
4 wa
s
reverse
d
with
data of
time synchro
n
ization
of
different s
i
tes (NJ
X
J
,
XJ
LD,
NJ
TQ were used
as
inversi
on dat
a and
NJQL
wa
s used a
s
test data), while 200
907
1
5
wa
s reve
rsed with d
a
ta of
different time
s of the
same
site (thi
s thu
nderclo
ud
wa
s be
sid
e
NJX
J
an
d NJL
D
station, therefo
r
e
NJX
J
wa
s u
s
ed a
s
an inve
rsio
n d
a
ta a
n
d
XJL
D
wa
s
use
d
a
s
te
st
data). T
he
evolution
of cha
r
ge
cente
r
den
sit
y
in different layers of the
two ca
se
s is
sho
w
n in figu
re 5a &b. Th
e simulate
d a
n
d
measured EF
values of inv
e
rsi
on a
nd te
st we
re
simila
r to the case
of 2009
0814.
Combi
ned
wi
th
the ra
dar
dat
a of the two
ca
se
s,
it wa
s found that t
he evolutio
n
of cha
r
g
e
ce
nter d
e
n
s
ity was
simila
r to th
e ca
se
of 2
0090
814
and
the ch
ar
ge
stru
cture eva
l
uated li
ke t
h
is: the
cha
r
ge
(b)
(a
)
(c)
(d)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 1, Janua
ry 2013 : 63 – 72
70
stru
cture of thund
erclou
d in
developm
e
n
t phase wa
s dipole, the mature
d stag
e wa
s tripole
and
it beca
m
e di
pole
whe
n
th
e thund
erclo
ud was
ex
tinct. The
cha
r
ge de
nsity e
v
olution of case
2009
0824
i
s
sho
w
n
in fig
u
r
e
5a
and
th
e e
r
ect
dotte
d line
ab
out
0718
UT
C is
the be
ginni
ng
o
f
lightning o
ccurred while t
he erect d
o
tted line a
bout
0736
UT
C is the en
d of
lightning. Th
ere
occurre
d
15 lightning flash
e
s du
ring thi
s
stag
e
and
fig.5a indicat
ed that
the charg
e
structu
r
e
durin
g the occurre
n
ce of lightning
was t
r
ipole. Fi
g 5b
is the cente
r
char
ge de
nsity evolution
of
2009
0715
an
d there
wa
s o
ne lightnin
g
o
c
curred
(the
pl
ace at the e
r
ect d
o
tted lin
e) in this
ca
se
.
The sa
me as in the case
s of 2009082
4
and 2009
08
14,
the lightning ch
arg
e
structure wa
s al
so
tripole
whe
n
the lightnin
g
o
c
curred. Fig
u
r
e 5
c
&d are
sho
w
in
g the
EF netwo
rk
o
f
200908
24 a
n
d
2009
0715
wh
en lighting
occurre
d
re
sp
e
c
tively. The
n
e
tworke
d EF
wa
s sim
u
late
d acco
rdin
g to
the model. It is indicated
that the EF netwo
rk
and
the rad
a
r e
c
ho we
re in g
ood ag
ree
m
e
n
t.
Meantime, b
e
ca
use of th
e existen
c
e
o
f
positive
cha
r
ge i
n
the lo
wer layer, th
ere
wa
s a
n
small
negative EF
area
ben
eath
the str
ong ra
dar echo,
whi
c
h we
re simil
a
r
to
the ca
se
of
200
908
1
4
.
Mean
while, t
he location o
f
lighting occurred an
d ch
arge
den
sity distrib
u
tion
were al
so in
g
ood
agre
e
me
nt, which
wa
s also
the same a
s
that in the ca
se of 200
908
14.
Figure 5. Cha
r
ge an
d EF networkin
g dist
ribut
ion in the
case of 2009
0824 a
nd 20
0
9071
5.(a
)
Cha
r
ge evol
u
t
ion in ca
se o
f
200908
24; (b) Ch
arge
ev
olution in case of 20090
71
5; (c) T
he EF
netwo
rki
ng wi
th model at 0718 UT
C in case of
20
090
824; (d
) The
EF netwo
rkin
g with model
at
0825
UTC in
ca
se of 200
9
0715; (e
) The
EF networkin
g with Krigin
g
interpolatio
n method at
0825
UTC in
ca
se of 200
9
0715.
(e)
(b)
(a
)
(d)
(c)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Elect
r
ost
a
tic Field Networkin
g
in Th
ree Isol
ated T
hund
erstorm
s
(Zhen
hui Wa
ng)
71
The su
rfa
c
e
electri
c
field
distrib
u
tion in
experimenta
l
site can be
obtained through
limited ob
servation data with spa
c
e inte
rpolatio
n
met
hod. Fig 5
e
is the EF network
at 082
5 UT
C
in the case
2009
0715
ba
sed
on the
Kriging i
n
te
rp
olation meth
od. It was
shown that th
ere
existed
som
e
deviation
bet
wee
n
mo
del
EF network
with the lo
catio
n
of lig
htning
and
rad
a
r e
c
ho.
Meantime, th
e sa
me resul
t
was
re
ceiv
ed in oth
e
r t
hund
erstorm
s
and
with ot
her int
e
rp
olat
ion
method
s. So we spe
c
ulate
d
that it was result
e
d
due t
o
the inade
qu
ate intensive
site data.
From t
he a
n
a
ly
sis of
t
h
r
e
e ca
se
s,
it
wa
s sh
own that the EF netwo
rki
ng a
nd the
cha
r
ge
de
nsit
y distri
bution
whi
c
h
obtain
ed from
rad
a
r and
soun
din
g
data
in i
s
ol
ated
storm
cells
were in
ag
re
ement
with t
he lo
cation
o
f
rada
r e
c
h
o
es
and li
ghtn
i
ng. Mea
n
whi
l
e, the cha
r
ge
stru
ctures of
three thund
ersto
r
m
s
we
re simila
r,
i.e., the charge
stru
cture
was dip
o
le in
the
developm
ent, tripole i
n
th
e mature st
age a
nd di
p
o
le after m
a
tured
stag
e. It was
also
in
coin
cid
e
n
c
e
with the f
a
ct
that the th
ree thu
nde
rst
o
rm
s
were o
f
the same t
y
pe. The
ch
arge
stru
cture of different stage
s
in thunderclo
ud al
so sho
w
s t
hat the cha
r
ge st
ru
cture
of
thunde
rcl
oud
is compli
cate
d. It was al
so
found that
th
e evolution of
rada
r e
c
h
o
a
nd the
cha
r
g
e
stru
cture in th
ree thu
nde
rcloud
s were
ba
sically con
s
i
s
t
ent to the re
sults of
r
e
s
e
a
r
c
h
es
[2
1
]
ab
ou
t
stron
g
upd
raf
t, electrificati
on activity and EF.
Mean
while, the lig
htning o
c
curred in the trip
ole
stage i
n
thre
e thund
erclo
uds
and the
lightning lo
cation was b
e
t
ween th
e ce
nter of po
siti
ve
cha
r
ge d
e
n
s
ity in lower lay
e
r and n
egati
v
e charge in
central layer,
whi
c
h we co
n
j
ectured that the
existen
c
e
of positive
cha
r
ge in th
e lo
wer laye
r of th
unde
rcl
oud
m
a
y be e
a
si
er
to be ex
cited
the
CG lightnin
g
, becau
se the
majority of
CG li
ghtnin
g
flash beg
an
with negative
leader an
d the
existen
c
e of
stron
g
EF a
r
ea cau
s
ed
by
the lo
we
r pa
rt of thund
ercloud
can
ea
si
ly be excite
d
the
leade
r a
nd th
e spre
ad to t
he g
r
ou
nd. T
h
is
re
sult
co
nforme
d to t
he
con
c
lu
sio
n
abo
ut light
ning
flashe
s an
d lowe
r po
sitive cha
r
ge [22].
5. Conclusio
n
With the help
of the model, EF data and the re
lation
of radar reflectivity, thunderstorm
initial feature, firstly the e
n
v
ironme
n
tal tempe
r
ature a
nd cha
r
ge
di
stributio
n a
r
e
impleme
n
ted
in
the inve
rsio
n
of ch
arge
distribution
pa
ra
meters,
an
d t
hen th
e EF
n
e
tworkin
g
ca
n be
calculat
ed
with the
mo
d
e
l an
d th
e p
a
ram
e
ters
ob
tained.
T
h
re
e
thund
erstorms
ob
serve
d
at the
site
of
experim
ent covered by rad
a
r and field m
ills we
re
an
al
yzed in this p
aper. Th
e re
sults sh
ow that
:
(1)
There exists
some
deviati
on between t
he net
worke
d
EF received
from Krigin
g i
n
terpol
ation
method an
d the location of lightning and
rada
r ech
o
. We conje
c
ture that it is resulted fro
m
the sparse data of EF. It illu
strates the im
portance of the layout of EF mill as well.
(2)
The EF net
workin
g an
d
the cha
r
g
e
density whi
c
h a
r
e o
b
tained by the
qua
si-n
orm
a
l
distrib
u
tion m
odel i
s
in a
c
cordan
ce
wit
h
t
he lo
catio
n
of the rada
r e
c
ho
and
CG lig
hting,
whi
c
h i
ndicated the
ratio
n
a
lity of the m
odel
and
pa
rameters to
some extent. I
t
is reveale
d
that the method of netwo
rked EF usin
g
radar a
nd sou
nding d
a
ta is
feasibl
e
.
(3)
The ch
arge structu
r
e an
d cha
r
ge di
strib
u
tion
of different stage
s in
these
ca
se
s have som
e
indication an
d referen
c
e functio
n
for the resea
r
ches about the ch
ara
c
teri
stics of
electri
c
ity
and lighting f
o
re
ca
sting.
Becau
s
e of the absen
ce o
f
effective means, such as
sou
ndin
g
EF to detect the cha
r
ge
stru
cture, the cla
ssi
c
cha
r
g
e
stru
ctu
r
e is
use
d
in mod
e
l
and at the
same time onl
y isolated
sto
r
m
cell
s are
anal
yzed in this
pape
r. Furth
e
r
study is n
e
eded to ma
ke the inversi
on pro
c
e
s
s
more
obje
c
tive so that the retrie
vals wo
uld b
e
more reliab
l
e and uniq
u
e
.
The shieldi
n
g effect of clo
ud
boun
dary
an
d the
disch
a
rge of
lighting
whi
c
h
b
r
ing
s
so
me
errors
to re
sult
s a
r
e
igno
red.
The
s
e
deficienci
e
s
will be consi
dered in the future
study. Meanwhile, the EF
networking with m
u
lti-
polar
(mo
r
e t
han trip
ole)
model a
s
we
ll as mu
lti-po
lar thun
derst
orm will b
e
carri
ed out un
der
permi
ssion of
study con
d
itions.
Ackn
o
w
l
e
dg
ements
We a
c
kno
w
le
dge Professo
r Qilin Zh
ang
and
Wen’a
n
Xiao, and o
u
r frie
nd Mu
h
a
mmad
Ha
san Ali bai
g and Jame
s Wanjo
h
i Nya
ga for their v
a
luabl
e co
ntri
bution in this
resea
r
ch wo
rk.
The wo
rk is sup
porte
d by China Com
m
onweal
Ind
u
stry Re
se
arch Proj
ect (G
YHY200
806
0
14)
and P
r
og
ram
for Po
stg
r
ad
uates Research In
novatio
n
of Ji
ang
su
High
er E
d
u
c
ation In
stitutions
(CXLX1
1_0
6
24).
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No
. 1, Janua
ry 2013 : 63 – 72
72
Referen
ces
[1]
Liu
Q , Guo
C
M
, W
ang
CW
, Yan
MH.
T
h
e
surface
e
l
ectr
ostatic fil
ed-c
h
ang
e
prod
uce
d
by
lig
htni
ng
flashes
an
d th
e lov
e
r p
o
sitiv
e
char
ge
layer
of the thu
n
d
e
r
s
torm
. Acta M
e
teoro
l
og
ica S
i
nica.1
98
7; 45:
500-
504.
[2]
Z
hao Y, Z
h
a
n
g
YJ, Dong
DS, Z
hang
HF
, Ch
en CP, Z
h
a
ng
T
.
Prelimin
ar
y
ana
l
y
sis of ch
a
r
acteristics o
f
light
nin
g
in the
Nagq
u area of
the Qingha
i-
Xi
zang p
l
ate
au.
Chin
ese Jo
urn
a
l of Geophys
i
cs
. 2004; 47:
405-
410.
[3]
Liu CQ, Ho
u
JX, Li
u K. T
h
und
erstorm
w
a
rni
ng a
nd
for
e
cast w
i
th
at
mosphere
electric field data.
Ningx
ia En
gi
ne
erin
g T
e
chn
o
lo
gy
. 2010; 9: 20
2–2
05.
[4]
Arang
uren
D,
Montan
ya
J, S
o
la G, March
V. Ro
mero
D, T
o
rres H. On the l
i
ghtn
i
n
g
h
a
z
ard
w
a
rni
n
g
usin
g el
ectrost
a
tic fiel
d: Ana
l
ysis
of summe
r thund
erstorm
s
in S
pai
n.
Jo
urna
l of El
ectrostatics
. 20
09;
67: 507-
51
2.
[5]
Chai
R, W
a
n
g
Z
H
, Xi
ao W
N
,
Yang Z
J
, Z
h
a
n
g
HL, Z
h
ang
W
B
. Applicati
o
n of atmos
pher
ic el
ectric fiel
d
data in l
i
g
h
tnin
g
w
a
rn
ing.
Met
eoro
l
og
ical Sci
ence a
nd T
e
ch
nol
ogy
. 20
09; 37: 724-
72
8.
[6]
Meng Q, Lv W
T
, Yao W
,
He P, Z
hang YJ,
Liu Q, Li
L, Z
han
g M, Ch
an
g C.
App
licati
o
n of det
ectio
n
data from e
l
ect
r
ic field m
e
ter
on gr
oun
d to l
i
ghtni
ng
w
a
rn
in
g techn
i
q
ue.
M
e
teoro
l
og
ica
l
Monthly
. 20
05
;
9: 30-33.
[7]
Gao T
C
, Huan
g Z
Y
, Z
hang P, Yang B. Method of
am
al
gamate atmos
pher
ic electric
field d
a
ta an
d
radar d
a
ta.
Jou
r
nal of PLA Un
i
v
ersity of Scien
c
e and T
e
ch
no
logy
. 20
06; 3: 302-
306.
[8]
Montan
ya
J, Bergas J, Herm
oso
B. Electric
field me
asur
e
m
ents at gr
o
u
n
d
lev
e
l as
a ba
sis for lig
htni
ng
hazar
d w
a
rn
in
g.
Journa
l of El
ectrostatics
.20
04; 60: 24
1-24
6.
[9]
Che
n
W
M
. 2003. Princip
l
es of
lightn
i
ng sci
en
ce.
Chin
a Mete
orol
ogic
a
l Pres
s
, Beijin
g.
[10]
Golde R H. Li
g
h
tnin
g, Vol. 1,
Acade
mic Pres
s
, Londo
n.19
7
7
.
[11] Krehb
iel
PR.
T
he el
ectrical st
ructure of
thun
derstor
ms, in the eart
h
’
s
el
ec
trical env
iron
ment
. National
Academ
y Pres
s, W
a
shingt
on,
D.C. 1986; 90-
113.
[12]
Williams E. T
h
e tripol
e
structure of thun
ders
t
orms.
Journal
of Geophysic
a
l
Researc
h
.19
8
9
; 94: 131
51-
131
67.
[13]
Stolzen
bur
g M
.
An o
b
serv
ati
ona
l stud
y
of
electric
al struc
t
ure i
n
co
nvec
tive re
gio
n
s
of Mesosc
al
e
convectiv
e
s
y
st
ems.
Ph D diss, University of Oklahoma
, Nor
m
an.19
96.
[14]
Stolzen
bur
g M, Rust W
D
, Marsha
ll T
C
. El
ectric
al structu
r
e in thun
ders
t
orm convectiv
e
regi
ons.1.
Mesoscale convective s
y
st
em
s.
Journal of Geop
hysica
l Res
earch.
19
98; 1
03: 140
59-
140
78.
[15]
Stolzen
bur
g M
,
Rust W
D
, Marsha
ll T
C
. Electrical
structu
r
e in
thun
derst
orm conv
ectiv
e
reg
i
o
n
s. 2.
Isolated storms
.
Journal of Ge
ophys
i
cal R
e
se
arch
. 199
8; 10
3: 1407
9-1
409
6.
[16]
Z
hang, YJ, Ya
n MH, Z
hang,
CH, Liu QS. Si
mulati
ng
ca
lcul
ation
of charg
e
structure in th
und
erstorm for
different are
a
s.
Acta Meteorol
ogic
a
Sinic
a
. 2
000; 58: 6
17-6
27.
[17]
Z
hao Z
K
, Qie
XS, Z
han
g T
L
, Z
hang T
,
Z
hang HF
,
W
ang Y, Yu Y, Sun
BL, W
ang HB. Electric field
soun
din
g
s an
d the
ch
arg
e
stru
cture
w
i
t
h
in
a
n
isolat
ed t
hun
de
rstorm.
Chi
nes
e Sci
Bul
l
. 2
0
0
9
; 54:
35
32-
353
6.
[18]
W
o
rkman EJ, Holzer
RE. A prelim
in
ar
y
inv
e
stigat
i
on of t
he el
ectrica
l
st
ructure of thu
nderstorms.
Natio
nal Adv
i
s
o
ry Co
mmittee
for Aerona
utics
.1942; 8
50: 1-2
9
.
[19]
Krehb
iel PR,
Brook M, McCroc
y
RA. An a
nal
ysis
of the
charg
e
structure
of lightn
i
ng
disch
arges to
grou
nd.
Jour
na
l of Geophysic
a
l Res
earch
. 1
979; 84: 2
432-
245
6.
[20]
Marshal
l T
C, Rust W D. E
l
ectric
fie
l
d s
o
und
ings
thro
u
gh th
un
derstor
ms.
Journ
a
l of
Geop
hysica
l
Research
.19
9
1
;
96: 2229
7-22
306.
[21]
Guo F
X
, Z
han
g YJ, Yan MH. A numeric
al st
ud
y of
the ch
ar
ge structure i
n
t
hund
erstorm i
n
Nag
qu ar
ea
of the Qingh
ai-
X
iz
an
g plat
eau
.
Chines
e Jour
nal of Atmosp
h
e
ric Scie
nces
. 200
7; 31:28-
36
.
[22]
Shao
XM,
Li
u XS.
A pre
limi
n
ar
y
a
nal
ys
is of
intracl
o
u
d
l
i
g
h
t
ning
flash
e
s
a
nd l
o
w
e
r p
o
siti
ve ch
arge
of
thund
erclo
ud.
Platea
u meteor
olo
g
y
. 198
7; 6: 318-3
25.
Evaluation Warning : The document was created with Spire.PDF for Python.