JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 90, NO. D4, PAGES 6013-6025, JUNE 30, 1985
Large-ScaleCharge Separation in Thunderclouds
EARLE R. WILLIAMS
Departmentof Earth, Atmosphericand PlanetarySciences,
Massachusetts
Institute of Technolo•Iy,Cambrid•Ie
Currently available evidencefor the mechanismsthought to be responsiblefor large-scalecharge
separationin thundercloudsis reviewed.The evidencesuggeststhat electricallychargedprecipitation
particlesare basicto the electrificationprocessbut that convectivemotionsare essentialin accounting
for the electricalenergyof activethunderstorms.The relativecontributionsof air and particle motions
remain largely unquantified,and the physical origins of thundercloudelectrificationhave yet to be
established.
1.
The electrification
INTRODUCTION
of clouds is a result
2.
of the relative
dis-
placement of positive and negative electric charge at the molecular scale and the subsequentmacroscopicseparation of
opposite charge by air and particle motions. This paper is
concernedwith a review of the evidencefor various processes
thought to be responsiblefor chargeseparationon scalescommensurate with the cloud dipole. Vital to a complete understanding of cloud electrification is a consideration of charge
separation on both large and small scales,and the interested
reader is referred to a paper by Illin•tworth [this issue] that
addressesthe microphysicsof chargeseparation.
Other review articles concernedwith large-scalecharge separation that are more comprehensivethan the presentone, in
the sensethat existing theories and mechanismsare described,
are those by Vonne•tut [1963], Moore and Vonne•tut [1977],
Latham [1981], and Lhermitte and Williams [1983].
The paper is somewhatarbitrarily subdividedinto nine categories of observational results, each of which places constraints(to a greater or lesserdegree)on the processes
responsible for large-scalecharge separation.Section 2 is concerned
with the many studies contributing to a description of the
main negative charge center in thunderclouds and with the
evidencethat cloud temperatureis influencingelectricalstructure. In situ charge measurements and vertically pointing
Doppler radar observationsof precipitation are reviewed in
section 3 and 4, respectively,in order to assessthe contribution of precipitation particle motions to electrification.This
contribution is further examined in section 5, in which the
gravitational power associatedwith falling precipitation is explored. The effectsof air motion on electrificationare investigated in section6 by consideringhow the electricalpower and
flash rate of a thunderstorm depend on its size; further interesting implications are exposed. Multiple Doppler radar
analysis,which is essentialin distinguishingthe contributions
of air and particle motions to electrification,is the subject of
section 7. Section 8 is concerned with trends in the paths
taken by lightning relative to precipitation structureand with
possible implications for the distribution of electric space
charge that may influencelightning paths. The still-unresolved
question of whether the first lightning in developing clouds
can be accounted for by currently considered precipitation
mechanismsis addressedquantitatively in section 9. Finally,
section 10 reviews present information about the large-scale
electriccurrentsflowing within and around thunderclouds.
Copyright 1985 by the AmericanGeophysicalUnion.
Paper number 5D0015.
0148-0227/85/005D-0015505.00
60i3
DESCRIPTION OF THE MAIN
NEGATIVE CHARGE REGION
Perhaps the most firmly establishedresult in all of cloud
electrification research is the systematic positive dipole
characterof terrestrialthunderclouds[Wilson, 1920]. A principal result of studiesin recent years is a vastly improved
descriptionof the lower negativeend of this dipole, which has
beenmore accessible
to observationbecauseof its closerproximity to the earth'ssurface.The resultsof thesemany observations as they pertain to the vertical location of the main negative charge region are summarizedin Figure 1. These observations include balloon and rocket soundingsof the electricfield
within the cloud [Chapman, 1956; Winnet al., 1974, 1981;
Marshall and Winn, 1982; Byrne et al., 1983-1,electric field
measurementsof the charge transferred by lightning [Jacobsonand Krider, 1976; Krehbiel et al., 1979; Krehbiel, 1981-1,
and the locations of acoustic and VHF
radiation
sources as-
sociatedwith.lightning [Weber et al., 1980; MacGorman et al.,
1981; Proctor, 1983].
All of the inferred locations for negative charge are in regions of the cloud colder than 0øC, a result that has been
frequentlycited as evidencefor a precipitation-basedcharging
mechanisminvolving the ice phase ILarham, 1981]. Although
the range of cloud temperaturesassociatedwith the various
height determinationsis 0øC to -45øC, a remarkable feature
of individual electric field soundings[e.g., Winnet al., 1981;
Byrne et al., 1983] is the evidencefor layers of negativespace
charge that are only ,-, 1000 m thick, a mere 10-20% of the
height of the parent thundercloud.This relatively small thickness and the stratified
character
often identified
with
these
negative regions are not readily accounted for by convective
theoriesfor cloud electrification[Vonne•tut,1955; Wa•tner and
Telford, 1981]. Indeed, extensivecharge stratification appears
to be most pervasivein cloud systemsin which the vertical air
motions are small. To cite a few examples,the longest horizontal lightning discharges ever documented [Li•7da, 1956-[
were observedin the rear portions of squall lines,regionsthat
Doppler radar observationshave sinceshownto be regionsof
small vertical air motion. Acousticreconstructionsof lightning
channels [Teer and Few, 1974] and the mapping of charge
participating in lightning [Krehbiel, 1981] show the most extensive discharge development in the dissipating stages of
thunderstormsthat are often convectivelyquiescent.Finally,
the largest charge transfers associatedwith lightning have
been recorded in relatively shallow Japanese winter snowstorms [Nakano, 1979; Brook et al., 1982], which are
characterized by large wind shear and strong horizontal air
motion but modest vertical air motion. These conditionsagain
suggesta stratifiedorganizationof charge,particularly in light
of laboratory scalemodels [Williams et al., this issue].
6014
WILLIAMS:THUNDERCLOUD
than those in New Mexico. Whatever the reason, this re•sult
12 :•;::•:;&......LIGHTNING CHARGE
:.
CHARGE SEPARATION
should be of interest to modelers
of cloud convection
and
r---.]JAC
OBSON-•A-K
RiDE
electrificationprocessesand to laboratory workers who are
............
:............ •..........
FLORI.DA_
•_.1___ (19761_i•
concernedwith the microphysicsof chargetransfer.
A secondtrend apparent in Figure 1 is for the lightning
10
chargeheightsto be systematicallygreaterthan the inferences
i.•.•<
.....
[ ......
; 7"•--Ne-'-'v•,/!-•'){iCOr=]
-KREHBIEL•AU
•.......
I ...............
:•-• .......
:.......
(1979).....
based on other methods. Furthermore, the charge heights
basedon total flash analysisin Florida [Jacobsonand Krider,
1976] are higher than thosebasedon individual stroke analysisin Florida [Krehbiel et al., 1979; Krehbiel,1981]. In light of
the apparent oblate shape of negative charge regions the
.......•
above discrepancymay be attributable to a finite sizeeffectin
the inversionof electricfield changescausedby lightning.It
6
• v.. •
...........•---•--•............
•:.a.......
•. ß
can easilybe shownthat a laterally extensivesourceof charge
will have an exaggeratedheight if it is modeledas a point
charge.This biasis most easilyunderstoodby consideringthe
•0.C
limiting case of an infinite horizontal sheet of charge whose
•-.......
•y•ne e•al (t98•)
•4
effectat the groundis reproducedby a point chargeat infinite
VHF
ACOUSTIC
NEGATIVE
height. Additional negative charge height bias (P. Krehbiel,
3__•SAD!ATION.
SOURCES_
C.HARGE personalcommunication,1984) may result from the subsidiary
....... SOURCES .........................
LAYERS'_
lower positive charge center, which may participate in cloud
to ground discharges.
Another remarkableresult pertainingto the inferredheight
of the negativechargeregionin thundercloudsis its apparent
.....
time stability. Lightning charge location analysisfor a se•-;
....
quence of 15 dischargesin one storm [Krehbiel et al., 1984;
Fig. 1. Observations pertaining to the vertical locations of nega-• see Figure 2] indicates that the inferred negative charge
tive charge regions in thunderclouds.Lightning charge locations are
heightsvary by lessthan 1 km over a period of severalminplotted in histogramform. Negative chargecentersinferredfrom balloon and rocket soundingsand the mean heightsof VHF and acous- utes.Simultaneously,the inferred positiveend of the thundercloud dipole risessteadily with time. One would like to know
tic source distributions are both denoted by horizontal arrows. An
approximate scale for cloud temperature correspondingwith msl whether this behavior is due to an ever-expandingregion of
height is included.
selective charge transfer to precipitation particles or to
7 -20c........ ]
..
.
;.
,
upward transport of space charge by vertical air motions.
The negativechargeheight comparisonsin Figure 1 reveal
two significanttrends:
First, the inferred MSL heightsfor negative charge regions
in Florida are systematicallyhigher and at lower temperature
than those in New Mexico. This differencemay be attributable
to a difference in moisture regimes in Florida and New
3.
IN SITU MEASUREMENTS OF CHARGE CARRIED BY
PRECIPITATION
The nature of the chargedparticlesthat composethe prominent negativechargeregion in thundercloudsis a fundamental
questionin cloud electrification.Although very few measurementsof cloud particle chargeshave beenmade, severalinvestigators have succeededin measuringthe chargesand local
Mexico or to a difference in characteristic cloud size (see sec-
tion 6), as the Florida clouds are recognizedas being larger
Average Height of LightningCharge Centers for First 15 Discharges in Small Storm
Day 220 (Aug 8) 1977 (Florida TRIP 77)
-5O
- q0
-30
-20
(-)• • l-l'l • t' .'
Krehbielet al (1984)
,. :I'
-]_•.sm
CG
Discharges
•
- l0
.
Tl•,
SIllliB
Fig. 2. Lightning charge structure,derived by Krehbiel et al. [1984] for August 8, 1977, Florida thunderstorm,
illustratingtemporalstabilityof negativechargecenterheightsand systematicrise of positivechargecenterheights.Also
shownare VHF radiation sourcesprovidedby the LDAR (lightningdetectionand ranging)system.
WILLIAMS:THUNDERCLOUD
CHARGESEPARATION
6015
TABLE 1. Summaryof In Situ Measurements
of PrecipitationParticleCharge
Altitude
Range
Investigator
negative
1-9 nA/m2
3/16
negative
3-10 nA/m2
4-5 km msl
GaskeN et al.
all
dissipative
4/4
( + 4øC-0øC)
[1978]
4-6 km msl
et al.
negative
2-20 nA/m2
positive
1-25 nA/m2
negative
1-5 nA/m2
(+ løC to - 8øC)
[1980]
3-7 km msl
and Winn
mostly
generafive
mostly
dissipative
(+ 10øC to - 15øC)
[1982]
Gatdiner
GenerationCases/
DissipationCases
( + 7øC-0øC)
[1974]
Marshall
Magnitude of
CurrentDensity
3-4 km msl
Rust and Moore
Christian
Preferential
Signof Change
4-5 km msl
et al.
(- 3øC to - 7øC)
[1984]
currents carried by precipitation particles in situ. These The findingthat the negativespacechargedensitycould be
measurements are summarized in Table 1.
roughlyaccountedfor by the negativechargefound on preThe natural variability of thundercloudsrequiresa large cipitationwithin the inferrednegativechargeregionsuggests
This
number of suchmeasurementsto establishrepresentativecon- an importantrole for precipitationin the electrification.
ditions,and no grand generalizations
can be drawn on the findingis particularlynoteworthyin light of the likely possibasis of the available information in Table 1. In four of these bility that the total surfacearea of the cloudparticlesin situ
studiesa tendencywasfoundfor negativelychargedprecipi- was 1-2 ordersof magnitudegreaterthan that of the precipitation particlesbeneaththe inferrednegativechargeregion, tation particles.Sorelyneededare additionalin situ measurewith corresponding
particlemotionsthat were dissipating ments of this kind, both within the main negative charge
electrostaticenergy.Marshall and Winn [1982], on the other regionand, perhapsmore importantly,in the centraldipole
hand,founda strongtendencyfor generafiveparticlemotion. regionof thunderclouds.
Theselatter measurements
(Figure 3) deserveparticular atten4. DOPPLER RADAR MEASUREMENTS OF PRECIPITATION AT
tion, sinceboth precipitationparticle chargesand the vector
VERTICAL INCIDENCE
electric field were recorded from the same balloon. These data
alloweda comparisonbetweenthe chargedensityassociated Additional efforts to quantify the role of precipitationin
with the precipitationand the net spacechargedensityin situ. electrificationon somewhatlargescalesand at higherlevelsin
150
100
AUGUST 7, lg7g.
•r-5500m,
Bc:lloon E ?ield
d MSL
cloud
1015
'
'
' 10'20 ....
10'25 '
1029 •
I
-•0MST MST MSTMST
-100
-.,_so
L
7500m,
MSL••15
øC)
•odE/dz
.
(,•ono-Cowl/cub
i ½ me'• e r")
.
.
10'15
i
i
•
i
MST
1
1020
NST
,
Prec i p. c¬ar•e densi Ly Ill
MST
,
r'•
, i-t-,
1020 4, M'ln,,,,,•Tn,,[,,}•[;•,,•,'•,
MST
MST
MST
.
Marshall & Winn
Fig. 3. Electric
fieldeo(dE/dz)
andthespace
charge
density
associated
withprecipitation
•orAugust
7, 1979,
NewMexico
thunderstorm[Marshall and Winn, 1982].
6016
WILLIAMS:THUNDERCLOUDCHARGESEPARATION
TABLE 2. Summaryof VerticallyPointingDopplerRadar
Measurements
of Precipitation
at theTimeof NearbyLightning
Discharge
Investigator
10
Location
Findings
L. J. Battan(unpublished) Arizona
Zrnidet al. [1982]
Oklahoma
WilliamsandLherrnitte
New Mexico
negativeresult
negativeresult
velocityperturbations
Mazur et al. [1983]
negativeresult
[1983]
Negative Charge LocationData-
l
) Jacobson
and
Krider,
1976
1
in < 1% of cases
Oklahoma
the cloud have been made with verticallypointing Doppler
radar [Zrni6 et al., 1982' Williams and Lhermitte, 1983; Mazur
et al., 1983]. Theseeffortswere motivatedby the predictions
of precipitation chargingmechanisms[Levin and Ziv, 1974;
Chiu, 1978] that the terminalvelocitiesof precipitationparticleswould be appreciablymodifiedby fieldsof thunderstorm
magnitude.At the time of a nearby lightningdischargethe
electricalforcesacting on the precipitationparticlesshould
suddenlychange,resultingin a changein the Doppler velocity.
The resultsof thesestudies,summarizedin Table 2, are,for
the most part negative.The few positiveresultsare too few in
numberto supportany broad generalizations
but do provide
evidencethat on occasionthe precipitationparticleswithin
Precipitat
3--
-- Power
Distribution
• x•
2--
1--
I
00
0.5
I
1.0
I
1.5
Gravitational
Power(gigawatts/km)
the regionof radarsurveillance
(severalhundredcubicmeters)
Fig. 5. Verticaldistributionof the gravitationalpowerassociated
were chargedpredominantlywith a singlepolarity. No evi- with precipitation
in August13, 1979,Florida thunderstorm
[Wildencefor systematicgenerativeprecipitationparticlemotions liamsand Lhermitte,1983] and comparisonwith inferredelectrical
structure[JacobsonandKrider, 1976].
has yet appeared.
5.
GRAVITATIONAL POWER ASSOCIATEDWITH FALLING
PRECIPITATION
The vertically pointing Doppler radar measurementsde-
scribedin section4 were motivatedin additionby considerations of the gravitationalenergyavailable to precipitation
mechanisms[Williams and Lhermitte,1983], a quantity at
leastan orderof magnitudelessthan the convective
energyof
thunderclouds[Braham, 1952]. Radar measurementsof the
gravitational power associatedwith falling precipitation
(Figure 4) indicatethat for widespreadstorm systemswith
modestconvectiveactivityand lightningflashrateslessthan 1
min-•, thegravitational
energy
of precipitation
caneasilyaccount for the electrification.In suchcases,no detectableveloc-
ity perturbations
are expected
in the verticallypointingDoppler radar measurements.
For convectivelyvigorousstorms
with lightningflash rates of severalper minute the gravitationaland electricalpowerare of comparablemagnitude.If
fallingprecipitationis the principalcontributorto the electrification of these clouds, then it follows that the electrical and
gravitationalforcesacting on theseparticleswill be of comparable magnitude and that the charging processmust be
consideredhighly efficient[Williams and Lhermitte,1983].
This predictionis not consistentwith the generalabsenceof
O
ß Thunderstorms
W•despreadRa•nshower Systems
T
J_
i
velocitychangesin the Doppler measurements.
If one examinesthe vertical distributionof precipitation
gravitationalpower in a thunderstorm(Figure 5) and compares it with presentinformation on grosschargestructure
(recall Figure !), one finds that less than 10% of the total
precipitationgravitationalpower lies in the inferredcentral
dipoleregion,whereaccordingto the classicalprecipitation
hypothesis,negativelycharged precipitationparticlesfall
againstthe local field to maintain the positivethundercloud
dipole.As discussed
in section7, the centraldipoleregionin
electrically active thundercloudsis not so much characterized
T
0.101
.
I
1
ß
I
lO
Electrical l Power(gigawatts)
Gravitational
lOO
by largequantities
of fallingprecipitation
asby strongupward
air motionsin whichprecipitation
particlesmayberisingrelative to the ground.
A vital but poorly known quantityis a representative
value
for the total electrostatic
energytransformed
by a lightning
Fig. 4. Radarmeasurements
of the gravitationalpowerassociated
If thisenergyis 109J, thenthegravitational
power
with falling precipitationin thunderstormsand estimatesof the simul- discharge.
to accountfor
taneouselectricalpower basedon lightningflashrate [Williams and in the centraldipoleregionappearsinadequate
Lherrnitte, 1983].
the observedelectrification.
If a representative
valuefor flash
WILLIAMS' THUNDERCLOUDCHARGE SEPARATION
6017
energyis 108J, the availableenergyis adequate,
but detect-
l
able velocity changesin the Doppler radar observationsare
expected.
6.
The
SCALING LAWS FOR CLOUD ELECTRIFICATION
well-established
correlation
between
+
the convective
+ +
+
vigor of clouds and their simultaneouselectrification leaves
little doubt that air motionsare playinga fundamentalrole in
large-scaleelectrification.Nevertheless,a quantitative assessment of this contributionis still unavailable.A semiquantitative assessment
is affordedby a considerationof how the electrical output of a thunderclouddependson its size[Vonnegut,
+
+
+
+
t
1963; Williams, 1981].
Simplescalingrelationshipsfor the steadystate electrical
power output of a thunderclouddipole structurewith characteristicsizeL were outlinedby Vonnegut[1963] and are summarized in Figure 6. Accordingto Gauss' Law, if the space
chargedensityis scaleindependent,the electricfield will scale
with the size of the cloud and the total potential differenceas
L 2. The chargingcurrentthat maintainsthe positivedipole
will dependon the productof the chargetransportvelocityV
and the cross-sectional
area of the cloud, or VL 2. The electrical power generatedis the product of the chargingcurrent
and the potentialdifferencethroughwhichit passesand scales
Electric
like L ½timesthechargetransportvelocity.
A fundamentalquestionconcernsthe variation of charge
Gauss'
transport velocity with cloud size. If convectiveair motions
areimportant,onemay expectthe chargetransportvelocityto
increasesubstantiallywith the sizeof the cloud.A compilation
of maximum air velocitydeterminationsfor cloudsrangingin
size from a few kilometersto 17 km is shown in Figure 7.
Empirically, the convectivevelocityis proportional to the size
Potential
Law
Pield
{ E)
?'E
= 0/½
Difference
(P.D.)
of the cloud.
The relationshipbetweenprecipitation charge transport ve-
locity and cloud size will dependon the vertical scaleover
which precipitation is making an electrical contribution. If
precipitation in a relatively narrow range of msl (mean sea
level) altitude, or equivalently,cloud temperature,is important, one may expectthat the chargetransport terminal velocity would be independentof cloud size,in which caseelectrical
P.D.= • E'dl - pL2 ' nz
Charging
Current
{I)
J ß (area)
- pVL
Electrical
Power
- VL
powerwouldscalelikeL '•.If, on theotherhand,precipitation
particlesovera substantial
depthof thecloudare contributing
to electrification,one must considerthe variation of particle
terminal velocitywith heightin the atmosphere.In sucha case
thechargetransportvelocitywill increase
with cloudsizeand,
given the natural variability in quantitiesof this kind, will be
indistinguishable
from convectivechargetransport,as will be
P = I
'
(P.D.)
~ VL
(P)
ß L
' VL
Fig. 6. Scaling relationships for cloud electrification [Vonnegut,
1963].
demonstrated.
Cloud height and flash rate data from three geo•aphical
locatiOns--New England [$chackford,1960], Forida [Jacobsonand Krider, 1976; Livingstonand Krider, 1978], and
New Mexico [Williams, 1981]--were used to test the scaling
law predictions.Flash rate valueswere averagedwithin 1-km
cloudheightintervalsto producethe plots in Figure 8. Least
squaresfits exhibitpowerlaw slopesof 5.0 (New England),4.7
Brook, personalcommunication,1979) on flash rate and cloud
height for a particularly vigorousFlorida thunderstorm.The
fifth power of the cloud height is very well correlated with
flashrate for a period of severalhours.
The empirical fifth power dependenceof flash rate on cloud
size has severalimportant implicationsin light of the scaling
First of all, lightningflashrate appearsto be
(Florida),
and4.9(NewMexico).
All threevalues
arein good law predictions.
agreementwith the predictedslopeof 5.
The agreementwith the simplesteadystate predictionis
quite remarkable in light of the obvious departuresfrom
an honest measure of the quasi-steady state electrical power
producedby a thunderstorm.This conclusionin turn requires
that the energyper flash be independentof the size of the
steadystate behavior in actual thunderclouds,and indeed, the
cloudin whichit occurs.
Thisresultis surprising
andrepre-
naturalvariabilityin thesedatawassuppressed
by theaverag- sentsa departurefromcertainelectrostatic
predictions
[Voning procedures
adoptedin Figure8. HOwever,
the largerand negut,
1963].Furthermore,
because
theenergy
disslpated
will
moreelectricallyactivethe cloud,the betterthe steadystate goliketheproductof chargetransfer
andthepotentialdifferassumption becomes.Figure 9 shows unpublished data (M.
ence,one has a prediction that the charge transfer AQ will be
6018
WILLIAMS:THUNDERCLOUD
CHARGESEPARATION
lOO
Data from:
LU
Adler and Fenn (1979)
Battan (1975)
Battan and Thei$$ (1965,1970)
Kropfli and Miller
(1976)
Lawson (1979)
Malku$ (1954)
Schmeter and Silaeva (1966)
Shenk (1974)
Steiner
Telford
Warner
and Rhyne (1964)
and Warner
(1969)
(1962)
,<
1
3
10
20
CLOUD SIZE (KM)
Fig. 7. Compilation
of maximumverticalair motionsin cloudsof varioussizes,illustrafin8empiricalincrease
of
convectivevelocity with cloud size.
smallerfor largeactiveflashrate storms,a predictionin qualitative agreementwith observations[Livingstonand Krider,
1978;Lhermitteand Krehbiel,1979; Williams,1981]. A related
implicationconcerns
the cloudchargingcurrent.A stormproducingflashesat 10 timesthe rate of anotheris not expected
to have a chargingcurrent10 timesas large but rather more
like a factorof 3 as large.Finally,the fifth powersizedependencedoesnot permita distinctionbetweenchargetransport
by convectionand by falling precipitationover a substantial
depth of the cloud.
A differentkind of scalinglaw implicationconcernslightning in warm clouds [Moore et al., 1960], whoseoccurrence
is everywheregreater than 0øC is constrainedto be no more
than 3-4 km in height.If the scalinglaw trend in Figure 8 is
extrapolatedto thesesmallclouds,the predictionfor flashrate
is lessthan one flash every 10 min. Sincethe lifetimeof clouds
of this size is also of this order, it seemslikely that small
cloudshave a small probability for reachingthe lightning
stage,and it is perhapsnot surprisingthat warm cloudlightning has beeninfrequentlyobserved.
7.
DOPPLER RADAR DETERMINATION OF AIR AND PARTICLE
MOTIONS
Very valuableinformationusefulin distinguishing
the roles
has not beensatisfactorily
explained.Becauseof the temper- of air and precipitationparticlemotionsin electrificationhas
ature structurein the troposphere,a cloud whosetemperature comefrom multiple Doppler radar analysis.The resultsof one
WILLIAMS'THUNDERCLOUD
CHARGESEPARATION
6019
......
•ß
...........................................
:
,
.•.....
'•.i,.r
2 .. • .;L.,25•f2•:
::......•.'.;::::T::.,. •...,
•
,,
....
: ..
.
.:.:.:::.
:
: :;:-.-.-
...........
_
............
..........
..........
..,.,
L'
.
...................................
..........•ß
...............................
'..,,;.,,.•.;,•
................................
•::::..4'..•..•.,.;?;};,,,:;.5..•:•:•:
•::':
:•:-•:
...................
, ...............
•
, ........ ...........
.......
•
............... •
...,.•
.......
•.:•:•
'•:':
::...,.:.:.:.:•
i•:..•.:..•..•..'-;:•:.
,t:•.:-:
•:,.
:•:•::•.:.•
ß
................
• .............
6020
WILLIAMS'THUNDERCLOUD
CHARGE
SEPARATION
ß
FloSh
ß
14
Ternporol
Vatlotion
of
Lightning
Flash
•Rate
and Cloud Height
Fig.9. Temporal
variation
oflightning
flash
rate,
cloud
height,
and
fifth
power
ofcloud
height
forAugust
10,1976,
Floridathunderstorm
(M. Brook,personal
communication,
1979).
of theseanalysesfor a Florida thunderstorm[Lhermitteand
Krehbiel, 1979] is shownin Figure 10, in which both the
scalar field of radar reflectivity and the vector field of air
motion are illustrated.Superimposed
on this storm structure
altitudeof maximumreflectivity
is comparable
withtheheight
at which negativecharge has been located, and indeed one
expectsto have negativespacechargecolocatedwith the intenseprecipitationif the fallingprecipitationis responsible
for
is the histogramof lightningchargeheights[Krehbiel,1981] the electrification.
Otherobservations,
to be discussed
in sec-
as it appears in Figure 1. A prominent feature of the air tion 8, on this storm and a number of others are inconsistent
motionwithin the radar reflectiveportion of the cloudis the with the view that negativespacechargecoincides
with the
occurrence
of an updraftwith maximumamplitudehighin the mostintenseprecipitationaloft.
cloudand a regionof maximumreflectivityaloft whoseintenAlsoevidentin this Doppleranalysisof air motionis the
sityis unmistakably
correlated
with theelectrical
activity.The existenceof both updraftsand downdrafts,which are most
pervasive
in the inferredcentraldipoleregionandin the up--'
' ,',/ -'
19
a07a•TO19
I•09"GM---'•
permost few kilometers of the radar reflectivecloud. These are
features of cloud circulation that are essential to convective
electrification
and that were advocatedby Vonnegut[1955,
Reflectivity
Structure
1963] long beforesuch Doppler observationsbecameavailAugust13 1978
able.The downdraftair is on the veryedgeof the detectable
radar echoand thereforeis not fully resolved,therebyillus(1979
I
trating a limitation of centimetricradars in providingthe
....-.-.---..\.. %,•,•,,
-'iL•.'..,,,,,,.,
'•• • I 1%•, Velocffy
ond
-' • ,t 4
...._
:•[.•/...
,,_......
,,,-•..
,, [I/ Lherm,,e
on
Krehb,el
"_•.•
' ' - '•'*'"'""'"'/""
"'""'-""'"
CHARGE.•D,UR•
....
, •,
*K•'ehbieJ •(AGI) ?)'
completepicture of air motion in and around the cloud. Since
spacecharge screeninglayers have been identified at cloud
boundaries[Winnet al., 1978; Byrne et al., 1983; Marshall et
al., 1984], new methodsfor observingperipheralair motion
are essential
in quantifyingthe contributionof screening
layer
convection to cloud electrification.
8.
LIGHTNING PATHS AND PRECIPITATIONECHOES
A puzzling result in need of theoreticalattention concerns
t0
km
Fig. 10. Dual-Dopplerradar analysisoœAugust]3, ]978, Florida
thunderstorm
rœhermitte
and Krehbiel,]979-1and comparison
with
inferredheightsof negativechargeregionfor a numberof Florida
storms[Krehbiel,1981].
the relationship
betweenlightningpathsand radar precipi-
tationechoes.
Whiletherenowseems
littledoubtthatlight-
ningpropagates
withi
n bothstrong
precipitation
andpreciPi-
tation
freecloud,
aswellasintheclearairoutside
thecloud'
a
number
ofobservations
indicate
a trendforlightning
to avoid
WILLIAMS: THUNDERCLOUD CHARGE SEPARATION
6021
TABLE 3. Spatial RelationshipsBetweenLightning and Precipitation
Technique
Radar
measurements
Investigator
Szymanskiand Rust
[1979]
"lightning echoes... where precipi-
Mazur
"maximum lightning density tends
to be near leading edge of precipitation cores"
"lightning is within high reflectivity region and ahead of it in
lessintenseprecipitation"
"most of the negativechargeis
associatedwith strong precipita-
and Rust
[1983]
Mazur
et al.
[1984]
Electric field
change measurements
VHF radiation
location
Results
Krehbiel
et al.
[1979]
Krehbiel [ 1981]
Proctor [ 1976]
Williams [1981]
tation
tion but some is associated
Cloud-to-ground
stroke location
Holle
et al.
[1983]
Geotis and Orville
[1983]
with
weaker precipitation"
"tendencyfor lightning to track
the edgesof precipitation regions"
"radiation sourcessystematically
displacedfrom upper level precipitation
Proctor [1983]
echo ... less than maximum"
echoes"
"most flashesprogressedfrom
regionsof higher reflectivity to
regionsof lower reflectivity"
"a strong decreasein likelihood of
lightning ... above 50 dBZ"
"strokesfound primarily in peripheral regionsrather than in or
near cores"
Acoustic(thunder)
Few [1974]
source location
MacGorman [1978]
MacGorman
et al.
[1980]
"lightning activity is concentrated
(if not confined)to regionsof the
cloud with low radar reflectivity"
"lightning channelstend to go
around regions of
high reflectivity"
"tendencyfor lightning channelsto
closelyparallel reflectivity
contours"
regionsof maximum radar reflectivity.Table 3 summarizesthe
findingsbased on radar observationsof the lightning plasma
[Szymanskiand Rust, 1979; Mazur and Rust, 1983; Mazur et
al., 1984], electricfield measurementsof the chargetransferred
by lightning [Krehbiel et al., 1979; Krehbiel, 1981], acoustic
reconstructionsof major lightning channels[Few, 1974; MacGorman, 1978], locations of VHF radiation sourceswithin the
cloud [Proctor, 1976, 1983], and fixes on the vertical channel
of cloud to ground lightning events [Geotis and Orville, 1983;
Holle et al., 1983].
Although the interpretation of these observations is still
open to question, experimentswith dischargesin laboratory
spacechargeclouds [Williams et al., this issue]provide graphical evidence(Figure 11) that dischargespropagate into regions of enhancednegativespacechargeand often completely
avoid regionsdevoid of spacecharge.The implication here, in
light of the trends in the lightning observations(Table 3), is
that regions of intense precipitation are not regions of maximum spacecharge density. It is important to note that such a
prediction is contrary to the resultsof theoreticalmodels for
cloud electrification[e.g., Rawlins, 1982; Takahashi,1984], in
which the maximum space charge density is colocated with
maximum reflectivity.
There are of course other possible explanations for the
trends in lightning paths. Some direct electrical interaction
between the lightning and the precipitation particles in its
vicinity may interfere with its progression.It is also possible
that a few very large precipitation particlescompletelydominate the radar reflectivity but contribute little to the local
spacechargedensity becauseof their small total surfacearea
[Lathare, 1981]. Yet another possible explanation for the
lightning avoidance phenomenonaloft is that the regions of
;..' "-'-. ' .....•.
...:'"...,
.......{.•;-:':.'.-:::.--..•',
: ,,.;;'
..
.....
.• ..........
--....
:.•'.•....:...
..:
............
:.:•:.•.::.:•:
............
,:....
......>
%.'
.......
ß
':.':"7::'":7::":'::'"',:-':::
.,,.
.....
.....
:.....:
ß
.-' .........
:.-'.'.:'.......
:':"...X':'
:'-":'""'
•
...'..:::•
:-:•-.':
.
...•.......
...... . ..........-""
Fig.11. Electrical
discharge
structure
within
a laboratory
scale
spacechargecloud [Williams et al., this issue],illustratingtendency
for pervasionof high spacechargeregionsand avoidanceof regions
devoid of spacecharge.
6022
WILLIAMS: THUNDERCLOUDCHARGESEPARATION
TABLE 4. Summaryof ParametersContributingto the Triple Productn. Aq. P
Ice Crystal
Concentration
Jones [1960]
Latham
and Stow
n,
per L
ChargeTransfer
per Collision
Aq,
fC
Maximum Precipitation
Rate at First Lightning
P,
mm/h
80
Reynoldset al.
[1957]
Takahashi[1978]
Gaskell and Illingworth
[1980]
Jayaratneet al.
[1983]
140
Moore [1965]
Moore [1974]
Holmeset al. [1977]
Lhermitte and Krehbiel
[1979]
Atchleyet al. [1980]
Christianet al. [1980]
Krehbiel et al. [1984]
1-3
3
6
24
10-100
[19693
Mossop et al.
[1972]
1-100
Hallerr
1-60
et al.
[1978]
Gatdiner
et al.
1-5
30
30
10
2
21
11-24
[1984]
larger reflectivityrepresentupwellingzonesthat conveypositively charged subcloudair to higher levels,as in the convective theory [Vonnegut, 1963]. This positivechargecould dilute
or completely offset the negative charge at these levels and
result in a dischargestructuresimilar to the one illustrated in
Figure 11.
9.
PRECIPITATION
MECHANISMS
AND THE FIRST
LIGHTNING
DISCHARGE
An old problem [Moore, 1974, 1976; Mason, 1976] in cloud
electrification,but one still in urgent need of experimental
attention on both the cloud scaleand the particle scale,concerns the adequacy of precipitation charging mechanismsin
accountingfor the first lightning dischargein a developing
cloud.
The quantity of interest[e.g.,lllingworth and Latham, 1977]
is the triple product n. Aq.P, where n is the small particle (ice
crystal)concentration,P is the precipitationrate and is representativeof the large particles,and Aq is the selectivecharge
transfer that occurs when small particles collide with large
particles.Table 4 presentsa summaryof valuesfor thesethree
quantitiesfrom the literature, where valuesof ice crystal concentration are based on in situ measurements within clouds,
valuesof precipitationrate P are inferredfrom radar measurements,and chargetransfervaluesAq are basedon laboratory
measurements.
Noteworthy here is the fact that the values for all three
quantitiesvary by at leastan order of magnitude,a resultthat
leadsto a very large uncertaintyin the triple product.Vitally
important but still poorly known are representativevaluesfor
each of the three quantities. Illingworth and Latham [1977]
have referred to their cloud model as being crude, but in this
reviewer'sopinion, charge transfer and separation by sedimenting precipitation can be modeled unambiguouslyand
probablyhavebeenadequatelytreatedfor the purposesof this
problem. As it stands,with a charge transferAq of 140 fC
[Reynoldset al., 1957], the largestvalue in this category,and
an ice crystalconcentrationof 80 per liter, a high value in this
category,the modelsstill cannotaccountfor the first lightning
in the available time with precipitation rates lessthan 10-20
mm/h [e.g., Illingworth and Latham, 1977]. Factor-of-2 discrepancies between theory and experiment are not too
troublesome in cloud electrification but order-of-magnitude
discrepancies
deserveattention, and as it standsthe discrepancyis closerto an order of magnitude.
Larger precipitationrates at the time of the first lightning
have been recordedin Florida storms [Lhermitte and Krehbiel,
1979; Krehbiel et al., 1984], but the observationsare too few
in number to draw any firm generalizationsconcerningpossible geographicaldifferences.Additional observationsof this
kind are essentialin resolvingthis important issue.
10.
THUNDERSTORM CURRENTS
This final sectionis concernedwith a review of present
knowledgeabout the large-scaleelectriccurrentsflowing in
and around thunderclouds.
Some of these currents are better
knownthan others,but considerations
of currentcontinuityin
a quasi-steadystate dipole permit inferencesabout current
magnitudesand directions.This information is summarizedin
Figure 12. Flowing in the clear air above the cloud is the
Wilsonconductioncurrent,whichin many respectsis the current most easilymeasuredand in other respectsthe most difficult. The averagecurrent measuredby Gishand Wait [1950]
was 0.5 A, with a maximum value of 6.5 A for a deepcloud,
which is of interest in light of the scaling law predictions
(section6). Based on the lightning charge transferanalysis
shown in Figure 2, Krehbiel et al. [1984] have inferred an
intracloud lightning current in the range 0.1-0.7 A. Representative values for the anvil advection current are still una-
vailable, but if this current is small and if the Gish and Wait
averagevalue is representative,then it may be inferredthat a
dipole charging current of the order of 1 A must flow to
maintain the upper positivecharge.Krider and Musser[1982]
have establisheda lower bound of 0.5 A for the dipole
charging current based on an analysis of Maxwell current
beneath a thundercloud.
The nature of the dipole charging current must be consideredone of the biggestunknownsin cloud electrification.
Based on the in situ precipitation current density measurements in Table 1, one may expect the magnitude of these
currents within
the cloud to be of the order of hundreds of
milliamperes.The systematiclocationsand directionsof these
currentshave not yet been established.
The current contributionof screeninglayer spacechargeby
convective motions, which is a fundamental element of one
convectivetheory [Vonnegut, 1955], is likewisepoorly specified.
Again to maintain current continuity in Figure 12, the
large-scalecurrents must be upward directed in the region
beneath the cloud. Results of recent studies [Livingston and
Krider, 1978; Standlerand Winn, 1979] suggestthat the largest contributor here is the cloud-to-groundlightning current.
Beneath electricallyactive thunderclouds,the corona current
densityis small in comparisonwith the situation beneathelectrified clouds whose lightning rate is small or nonexistent.
Standlerand Winn estimatea representativevalue of 0.1 A for
the corona current beneath a thundercloud.
Since this value is
small in comparisonwith estimatesfor intracloud lightning
current [Krehbiel et al., 1984], positivecorona spacechargeis
an unlikelyprincipalcontributorto positivechargeaccumulation in the upper regions of thunderclouds,contrary to the
predictionof Vonnegut's[1955] convectivetheory.
WILLIAMS: THUNDERCLOUD CHARGE SEPARATION
C.T.R.
CLOUD
TOP
WILSON
6023
CURRENT
• AVG
0.5AMP
LIGHTNING
cURRENT(?)
MAX6.5AMP
•k'Gish
&Wait
(1950)
SCREENING
LAYER
CURRENTS (?)
•
/
--
--
-
+.+++
0.5AMP ( .
I
Krider
&Musser
(19•) _ - - - -
1
TOTAL• [ - '
PRECIPITATION
C•RENT8
(?)
•
_- -
•
•
Krehbiel
(1981)
CLOUD
TOGROUND
/
UHTNN,CURRENT
0.1-1AMP
Livingston
& Krider(1978)
• 8tandler&
CORONA
CURRENT
Winn (1979)
O. 1 AMP
•
Fig. 12. Summaryof knowledge
concerning
quasi-steady
statethundercloud
currentson dipolescale.
11.
CONCLUSIONS
Although the origin of thundercloud electrificationremains
an open qustion, recent systematicresults shed new light on
this difficult problem. The stable stratified character of negative chargeregionsis qualitativelymore consistentwith largescale charge separation by differential particle motions than
by convectiveoverturn. The additional consistentfinding that
the cloud environmentassociatedwith negativechargeregions
is colder than 0øC is indirect evidencethat the ice phase is
basic to electrification.
Bothersomein supporting precipitation as the sole contributor to large-scalecharge separationis (1) the paucity of
evidencefor systematicgenerativeparticle motions (sections3
and 4), (2) the quantitative inadequacyof the currently popular ice phase precipitation mechanismin accountingfor the
first lightning in a developingcloud (section9), and (3) the
indirect evidencethat electric spacecharge is displacedboth
verticallyand horizontallyfrom the bulk of the precipitation
within fully developedthunderclouds(sections5 and 8). The
scaling behavior of cloud electrification(section 6), the longrecognizedassociationbetweenvertical air motion and lightning activity (sections6 and 7), and the energyrequirementsof
electricallyactive clouds(section5) all support an important
role for convectiveair motions in large-scalecharge separation.
The contributionsof precipitationand convectionto largescale electrification remain largely unquantified and require
further study. We shall need many more observationsof the
chargescarried by the various particles in thundercloudsas
well as an improved understandingof the convectivemotions
that affect the displacementof all chargedparticles.We shall
also need additional studiesin the laboratory that isolate and
realistically quantify those charging phenomena that occur
systematically in thunderclouds. Most of all, we shall need
new ideas about the physicalorigins of chargeseparationand
the meansfor distinguishingone contribution from another in
the field.
Acknowledgments.The author is gratefulto Marx Brook, Chathan
Cooke, Paul Krehbiel, Roger Lhermitte, Thomas Marshall, Charles
Moore, and Bernard Vonnegutfor supplyingmaterial for this review.
He is gratefulto theseindividualsand to manyothersfor illuminating
discussions
on the mysteriesof cloudelectrification.
REFERENCES
Adler, R. F., and D. D. Fenn, Thunderstormintensityas determined
from satellitedata, J. Appl.Meteorol.,18, 502-517, 1979.
Atchley,A. A., et al., TRIP '79 casestudyof August7 thunderstorm
over Langmuir Laboratory, paper presentedat Sixth International
Conferenceon AtmosphericElectricity,Am. Meteorol. Soc.,Manchester,England, 1980.
Battan, L. J., Doppler radar observations
of a hailstorm,J. Appl.
Meteorol., 14, 98-108, 1975.
Battan, L. J., and J. B. Theiss, Observations of vertical motions and
particle sizesin a thunderstorm,J. Atmos.Sci.,23, 78-87, 1965.
Braham,R. R., Jr., The water and energybudgetsof the thunderstorm
and their relation to thunderstormdevelopment,J. Meteorol.,4,
227-242, 1952.
Brook, M., M. Nakano, and P. R. Krehbiel, The electricalstructureof
the Hokuriku winter thunderstorms,
J. Geophys.Res., 87, 12071215, 1982.
Byrne, G. J., A. A. Few, and M. E. Weber, Altitude, thickness,and
charge concentrationof charged regions of the thunderstorm
during TRIP 1981 based upon in situ balloon electric field
measurements,Geophys.Res. Lett., 10, 39-42, 1983.
Chapman,S., Electrostatic
fieldmeasurements,
coronadischarge,
and
thunderclouds,Rep. C.A.L. 68, Cornell Aeronaut. Lab., Ithaca, N.
Y., 1956.
Chiu,C. S.,Numericalstudyof cloudelectrification
in an axisymetric,
time-dependent
cloudmodel,J. Geophys.
Res.,83, 5025-5049,1978.
Christian,H., C. R. Holmes,J. W. Bullock,W. Gaskell,A. J. Illingworth, and J. Latham, Airborne and groundbasedstudiesof thunderstorms
in the vicinityof Langmuirlaboratory,Q. J. R. Meteorol.
$oc., 106, 159-175, 1980.
Few,A. A., Lightningsources
in severethunderstorms,
paperpresented at Conferenceon Cloud Physics,Am. Meteorol. Soc.,Tucson,
Ariz., 1974.
Gardiner, B., D. Lamb, R. Pitter, and J. Hallett, Measurementsof
initial electricfield and ice particlechargesin Montana thunderstorms, paper presented at Seventh International Conference on
AtmosphericElectricity,Am. Meteorol.Soc.,Albany,N.Y., 1984.
Gaskell,W., and A. J. Illingworth,Chargetransferaccompanying
individualcollisionsbetweenice particlesand its role in thunderstormelectrification,Q. J. R. Meteorol.$oc.,106, 841-854, 1980.
Gaskell,W., A. J. Illingworth, J. Latham, and C. B. Moore, Airborne
studiesof electricfield and the chargeand size of precipitation
elementsin thunderstorms,Q. J. R. Meteorol. $oc., 104, 447-460,
1978.
6024
WILLIAMS:THUNDERCLOUD
Geotis, S. G., and R. E. Orville, Simultaneousobservationsof lightning ground strokesand radar reflectivitypatterns,paper presented
at 21st Conference on Radar Meteorology, Am. Meteorol. Soc.,
Edmonton, September19-23, 1983.
Gish, O. H., and G. R. Wait, Thunderstormsand the earth's general
electrification, d. Geophys.Res., 55, 473-484, 1950.
Hallett, J., R. L. Sax, D. Lamb, and A. S. R. Murty, Aircraft measurements of ice in Florida cumuli, Q. d. R. Meteorol. Soc., 104, 631651, 1978.
Holle, R. L., R. E. Lopez, and W. L. Hiscox, Relationshipsbetween
lightning occurrencesand radar echocharacteristicsin South Florida, paper presentedat International Aerospaceand Ground Conferenceon Static Electricity,U.S. Dep. Transp./Fed.Aviat. Agency,
Fort Worth, Tex., 1983.
Holmes, C. R., C. B. Moore, R. Rogers, and E. Szymanski, Radar
study of precipitation development in thunderclouds,in Electrical
Processesin Atmospheres,edited by H. Dolezalek and R. Reiter,
Steinkoff Verlag, Darmstadt, Fed. Repub. Ger., 1977.
Illingworth, A. J., Charge separation in thunderstorms--Small-scale
processes,d. Geophys.Res.,this issue.
Illingworth, A. J., and J. Latham, Calculations of electric field growth,
field structure and charge distributions in thunderstorms,Q. at. R.
Meteorol. Soc., 103, 281-295, 1977.
Jacobson, E. A., and E. P. Krider, Electrostatic field changes produced by Florida lightning, at.Atmos.Sci.,33, 103-117, 1976.
Jayaratne,E. R., C. P. R. Saunders,and J. Hallett, Laboratory studies
of the charging of soft hail during ice crystal interactions, Q. at. R.
Meteorol. Soc., 109, 609-630, 1983.
Jones,R. F., Size-distributionof ice-crystalsin cumulonimbusclouds,
Q. at.R. Meteorol. Soc.,86, 187-194, 1960.
Krehbiel, P. R., An analysisof the electricfield changeproducedby
lightning, Ph.D. thesis,Univ. Manchester,England, 1981.
Krehbiel, P. R., M. Brook, and R. A. McCrory, An analysis of the
charge structure of lightning dischargesto ground, at.Geophys.Res.,
84, 2432-2456, 1979.
Krehbiel, P. R., R. Tennis, M. Brook, E. W. Holmes, and R. Comes, A
comparative study of the initial sequenceof lightning in a small
Florida thunderstorm, paper presented at Seventh International
Conference on Atmospheric Electricity, Am. Meteorol. Soc.,
Albany, N.Y. 1984.
Krider, E. P., and J. A. Musser, Maxwell currents under thunderstorms,d. Geophys.Res.,87, 11,171-11,176, 1982.
Kropfli, R. A., and L. J. Miller, Kinematic structure and flux quantities in a convectivestorm from dual-Doppler radar observations,
at.Atmos. Sci., 33, 520-529, 1976.
Latham, J., The electrification of thunderstorms, Q. J. R. Meteorol.
Soc., 107, 277-298, 1981.
Latham, J., and C. D. Stow, Airborne studiesof the electrical properties of large convectiveclouds, Q. at.R. Meteorol. Soc.,95, 486-500,
1969.
Lawson, R. P., A system for airborne measurementof vertical air
velocity, at.Appl. Meteorol., 18, 1363-1368, 1979.
Levin, Z., and A. Ziv, The electrification of thunderclouds and the
rain gush,at.Geophys.Res., 18, 2699-2704, 1974.
Lhermitte, R., and P. R. Krehbiel, Doppler radar and radio observations of thunderstorms, IEEE Trans. Geosci. Electron., GE-17, 162171, 1979.
Lhermitte, R., and E. Williams, Cloud electrification, Rev. Geophys.
SpacePhys.,21, 984-992, 1983.
Ligda, M. G. H., The radar observationsof lightning, J. Atmos. Terr.
Phys., 9, 329-346, 1956.
Livingston,J. M., and E. P. Krider, Electric fieldsproducedby Florida thunderstorms,at.Geophys.Res.,83, 385-401, 1978.
MacGorman, D. R., Lightning location in a storm with strong wind
shear, Ph.D. thesis, Rice Univ., Houston, Tex., 1978.
MacGorman, D. R., A. A. Few, and T. L. Teer, Layered lightning
activity, d. Geophys.Res.,86, 9900-9910, 1981.
MacGorman, D. R., W. L. Taylor, and A. A. Few, Lightning location
from acousticand VHF techniq•es relative to storm structure from
10-cm radar, paper presentedat Sixth International Conferenceon
AtmosphericElectricity,Am. Meteorol. Soco,Manchester,England,
1980.
Malkus, J. S., Some results of trade-cumulus cloud investigation, at.
Meteorol., 11,220-237, 1954.
Marshall, T. C., and W. P. Winn, Measurements of charged precipitation in a New Mexico thunderstorm: Lower positive charge centers, at.Geophys.Res.,87, 7141-7157, 1982.
CHARGE SEPARATION
Marshall, T. C., W. D. Rust, and W. P. Winn, Screeninglayers at the
surfaceof thunderstormanvils,paper presentedat SeventhInternational Conference on Atmospheric Electricity, Am. Meteorol. Soc.,
Albany, N.Y., 1984.
Mason, B. J., In reply to a critique of precipitation theoriesfor thunderstorm electrification by C. B. Moore, Q. at. R. Meteorol. Soc.,
102, 219-235, 1976.
Mazur, V., and W. D. Rust, Lightning propagation and flash density
in squall lines as determined with radar, at. Geophys.Res.,88, 14951502, 1983.
Mazur, V., W. D. Rust, and D. Zrni•, Correlations betweenlightning
and precipitation velocity spectra determined with a vertically
looking radar, Eos Trans. AGU, 64, 662, 1983.
Mazur, V. M., J. C. Gerlach, and W. Rust, Lightning flash density
versus altitude
and storm structure
from
observations
with VHF-
and S-band radars, Geophys.Res.Lett., 11, 61-64, 1984.
Moore, C. B., Charge generation in thunderstorms,in Problemsof
Atmosphericand SpaceElectricity, edited by S.C. Coroniti, pp. 275,
159, Elsevier, New York, 1965.
Moore, C. B. An assessmentof thundercloud electrification mecha-
nisms,paper presentedat Fifth International Conferenceof Atmospheric electricity, Am. Meteorol. Soc., Garmish-Partenkirchen,
Fed. Repub. Ger., 1974.
Moore, C. B., Reply (to B. J. Mason), Q. J. R. Meteorol. Soc., 102,
225-238, 1976.
Moore, C. B., and B. Vonnegut, The thundercloud,in The Physicsof
Liqhtninq,edited by R. H. Golde, pp. 51-98, Academic,New York,
1977.
Moore, C. B., B. Vonnegut, B. A. Stein, and H. J. Survilas,Observations of electrification and lightning in warm clouds, J. Geophys.
Res., 65, 1907-1910, 1960.
Mossop, S.C., R. E. Cottis, and B. M. Bartlett, Ice crystal concentration in cumulus and stratocumulusclouds, Q. at.R. Meteorol.
Soc., 98, 105-123, 1972.
Nakano, M., Initial streamersof the cloud dischargein winter thunderstormsof the Hokuriku coast, at. Meteorol. Soc. atpn.,57, 452457, 1979.
Proctor, D. E., A radio study of lightning, Ph.D. thesis,Univ. Witwatersrand,Johannesburg,S. Afr., 1976.
Proctor, D. E., Lightning and precipitation in a small multicellular
thunderstorm,at.Geophys.Res.,88, 5421-5440, 1983.
Rawlins, F., A numerical study of thunderstorm electrification using a
three-dimensionalmodel incorporating the ice phase, Q. at. R. Meteorol. Soc., 108, 779-800, 1982.
Reynolds,S. E., M. Brook, and M. F. Gourley, Thunderstormcharge
separation,at.Meteorol., 14, 426-436, 1957.
Rust, W. D., and C. B. Moore, Electrical conditions near the base of
thundercloudsover New Mexico, Q. at. R. Meteorol. Soc., 100, 450468, 1974.
Shackford, C. R., Radar indications of a precipitation-lightningrelationship in New England thunderstorms,at. Meteor., 17, 15-19,
1960.
Shenk,W. E., Cloud top height variability of strong convectivecells,
at.Appl. Meteorol., 13, 917-922, 1974.
Shmeter, S. M., and V. I. Silaeva, Vertical streams inside cumulonimbus clouds, Meteorol. Gidrol., 10, 1966.
Standler, R. B., and W. P. Winn, Effects of coronae on electric fields
beneaththunderclouds,Q. at.R. Meteorol. Soc.,105, 285-302, 1979.
Steiner, R., and R. H. Rhyne, Atmosphericturbulenceand airplane
responsein convective-typeclouds,at.Aircraft, 1, 13-17, 1964.
Szymanski,E. W., and W. D. Rust, Preliminary observationsof lightning radar echoesand simultaneouselectricfield changes,Geophys.
Res. Lett., 6, 527-530, 1979.
Takahashi, T., Riming electrificationas a charge generation mechanism in thunderstorms, at.Atmos. Sci., 35, 1536-1548, 1978.
Takahashi, T., Numerical simulation of winter thunderstorm electrifi-
cation, paper presented at Seventh International Conferenceon
AtmosphericElectricity, Am. Meteorol. Soc.,Albany, N.Y., 1984.
Teer, T. L., and A. A. Few, Horizontal lightning, at.Geophys.Res., 79,
3426-3441, 1974.
Telford, J. W., and J. Warner, On the measurementfrom an aircraft of
buoyancy and vertical air velocity in cloud, at.Atmos.$ci., 19, 415423, 1962.
Vonnegut, B., Possiblemechanismfor the formation of thunderstorm
electricity,Conferenceon AtmosphericElectricity, Res. Pap. 42, pp.
169-181, Bedford, Mass., 1955.
Vonnegut, B., Some facts and speculationsconcerningthe origin and
WILLIAMS: THUNDERCLOUD CHARGE SEPARATION
6025
role of thunderstorm electricity, Meteorol. Monogr., 5, 224-241,
Wilson, C. T. R., Investigationson lightning dischargesand on the
1963.
electric fields of thunderstorms, Trans. R. Soc.,221A, 73-115, 1920.
Winn, W. P., G. W. Schwede, and C. B. Moore, Measurements of
electric fields in thunderclouds, J. Geophys.Res., 79, 1761-1767,
Wagner, P. B., and J. W. Telford, Cloud dynamics and an electric
chargeseparationmechanismin convectiveclouds,J. Rech.Atmos.,
15, 97-120, 1981.
Warner, J., The microstructure of cumulus cloud, 2, The effect on
droplet sizedistributionof the cloud nucleusspectrumand updraft
velocity, J. Atmos.Sci.,26, 1272-1282, 1969.
Weber, M. E., and A. A. Few, A balloon-borne instrument to induce
corona currents as a measure of electric fields in thunderclouds,
Geophys.Res. Lett., 5, 253-257, 1978.
Weber, M. E., H. J. Christian, A. A. Few, and M. F. Stewart, A
thundercloudelectricfield sounding:Charge distribution and lightning, J. Geophys.Res.,87, 7158-7169, 1982.
Williams, E. R., Thunderstorm electrification: Precipitation versus
convection,Ph.D. thesis,Mass. Inst. Technol.,Cambridge, 1981.
Williams, E. R., and R. Lhermitte, Radar tests of the precipitation
hypothesisfor thunderstorm electrification,J. Geophys.Res., 88,
10,984-10,992, 1983.
Williams, E. R., C. M. Cooke, and K. A. Wright, Electrical discharge
propagation in and around spacecharge clouds, J. Geophys.Res.,
this issue.
1974.
Winn, W. P., C. B. Moore, C. R. Holmes, and L. G. Byerly III, A
thunderstormon July 16, 1975, over Langmuir Laboratory: A case
study,J. Geophys.Res.,83, 3080-3092, 1978.
Winn, W. P., C. B. Moore, and C. R. Holmes, Electric field structure
in an active part of a small, isolated thundercloud,J. Geophys.Res.,
86, 1187-1193, 1981.
Zrni•, D. S., W. D. Rust, and W. L. Taylor, Doppler spectra of
lightningand precipitationat verticalincidence,J. Geophys.Res.,
87, 7179-7191, 1982.
E. Williams, Department of Earth, Atmospheric,and Planetary Sciences,MassachusettsInstitute of Technology,Cambridge, MA 02139.
(ReceivedJuly 5, 1984;
revised November 2, 1984;
acceptedNovember 26, 1984.)