You are on page 1of 10

Lightning in Tornadic Storms' A Review

DONALD R. MACGORMAN
National Severe Storms Laboratory, NOAA, Norman, Oklahoma 73069

1.

INTRODUCTION

There have been many reports of unusual lightning characteristics in tornadic storms. For example, eyewitnesses
have reported scorchingbeneath tornado funnels, a steady

or rapidly oscillatingglow inside funnels, rapidly recurring


small patches of light on the side of the thunderstorm,or
unusuallyhigh or low flashrates [e.g., ChurchandBarnhart,
1979; Vaughan and Vonnegut, 1976; Vonnegut and Weyer,
1966]. It is difficult, however, to quantify relationships
between lightningand tornadic stormsfrom these eyewitness reports.

has been learned from analysis of data from lightning mapping systems.
2.

LIGHTNING TERMINOLOGY

This sectionexplains lightningterminology and concepts


that will be used in the rest of the paper. Lightningcan be
classifiedinto two types: (1) Cloud-to-groundlightninghasat
leastone channelspanningfrom the cloud to the ground.(2)
Intracloud lightning does not have a channel to ground.
Becausethe locationsof lightningchannelsrelativeto visible
cloud boundaries are difficult to detect remotely, scientists
who studylightningand storm electrificationnormallygroup

Prior to 1975most quantitativemeasurementsof lightning in-cloud, cloud-to-air, and cloud-to-cloud lightning together
in the intracloud lightning classification. Cloud-to-ground
sferics, the electromagneticnoise radiated by lightning. lightningcan be further classifiedby the polarity of charge
These measurements were difficult to interpret, because thatit effectivelylowersto ground:Positivecloud-to-ground

in tornadic storms examined the rates and characteristics of

resultsdependedpartly on the characteristics


of the sferics
receiver, and becausesferics instrumentationprovided at
best only the bearingto the lightningflashgeneratingthe
sfericsand often provided no locationinformationat all.
Furthermore,manyof the sfericsstudieslackedradardata,
andonlyonehadanyDopplerradardata.Equallyimportant,
there was not an adequateappreciationof the variety of
storms that produce tornadoes.

lightninglowers positive charge;negativecloud-to-ground


lightninglowers negative charge. Most cloud-to-ground
flashesare negative.

A cloud-to-ground
lightningflashusuallybeginsinsidethe
cloudand is first apparentwhen a faint channel,calledthe
steppedleader,movesfrom the cloudto the groundinjumps
roughly50-100 m long. When the steppedleaderconnects
with the ground,a brightpulsemovesback up the lightning

Some of these limitations were overcome when systems channel


in a processcalleda returnstroke.After a pauseof
weredevelopedto mapthe locationof lightningflashes.For roughly20-150ms,anotherleadercantravelbackdownthe
example,beginning
in the late 1970s,variousorganizationsalreadyestablishedlightningchannel,followed by another
begandeploying
newlydeveloped
systems
for automatically return stroke. The combination of a leader and return stroke
mappingwherelightningchannels
strikegroundoverranges is called a stroke. All strokesgoingthroughessentiallythe
of a few hundredkilometers[e.g., Krideret al., 1980;Orville samechannelto ground make up a singlecloud-to-ground
et al., 1983;BentandLyons,1984;MacGorman
andTaylor, flash. There can be anywhere from one to a few tens of
1989].Now coverageby thesesystemsis continualacross strokes in a flash.
thecontiguous
UnitedStates,witherrorsin strikelocations Sfericsare generatedby changesin the vector electric
onthetime
typicallylessthan10km [e.g.,Machet al., 1986;MacGor- current,thefrequencyof the radiationdepending
man and Rust, 1989].
scaleof the changes.Since currentsin a lightningchannel
In this paper we first review what was learnedabout increasefaster than they decay, a lightningchannelwill
tornadic stormsfrom sfericsstudiesand then considerwhat generally
radiatehigherfrequencies
duringaninitialcurrent
surgethanduringits decay.The timerequiredfor a current
The Tornado:Its Structure,Dynamics,Prediction,andHazards.
pulseto propagate
througha channelsegment
depends
on
GeophysicalMonograph79
the
length
of
the
segment,
so
longer
channels
radiate
more
Thispaperisnotsubject
to U.S.copyright.
Published
in 1993bythe
energyat lowerfrequencies,if all elseis equal.
American GeophysicalUnion.

!74

REVIEW

OF LIGHTNING

IN TORNADIC

STORMS

The spectrumof sfericsfrom a typical return strokepeaks


at roughly 5 kHz. Most strong signalsat frequenciesbelow
about30 kHz are from cloud-to-groundreturn strokesand so
occur in discrete bursts, separated by tens of milliseconds.
At frequencies near or above 1 MHz, the amplitudesof
sferics from return strokes, from other processesof a

BURSTRATEversus FREQUENCY
4O

. 30

cloud-to-ground flash, and from intracloud flashes are all


comparable. Furthermore, at these higher frequencies, sfer-

ics are more numerous and more continual throughout a


lightning flash, and relatively few are associatedwith return
strokes.

The vertical componentof sfericsmeasuredat the stirface


of the conducting Earth is generally much larger than the
horizontal, even if the two components are comparable at
the source. Any horizontal component that exists is reduced
by currents that are induced in the ground by the stZrics.
'l"hcreforea sferics receiver at the ground receives little of
the energy radiated by horizontal components of current
change. Airborne receivers are less affected by currents
induced in the ground and so are better able to detect the
horizontal componentsof sferics.
3.

Initial

:I i : ::::l

:l ' : :'Jill

Tornadic
Storms
--.-.

X Severe
Storms

20

'"-' 10

H ...... e-.---.......
....../Thunderstorms
31.6 kHz

100 kHz

316 kHz

1 MHz

3.16 MHz

Frequency

Fig. 1. Average sferics burst rate as a function of receiver frequency for various storm classifications[from MacGorman et al.,

1989]. Data are from four tornadic storms, three nontornadic severe
storms, and seven nonsevere thunderstorms. Sferics burst rates
were measuredat 10 receiver frequencies between 10 kHz and 137
MHz, indicated by dots. There was little change above 3 MHz, so
higher frequencies are not shown.

SFERICS STUDIES

studies of sferics from tornadic

storms examined

radio bands fom 10 to 500 kHz. (Sferics normally are


classified by the frequency of the receiver used to detect
them.) Dickson and McConahy [1956] found that sferics
rates detected by a 10-kHz receiver often increased as

increasingseverity, until for tornadic storms, the energy is

independent of frequency.
There were exceptions to all of these findings. In several
cases, high sferics rates did not occur in tornadic storms
stormsgrew rapidly taller, but ratespeakedearlier in storms [Ward et al., 1965; Stan3'brdet al., 1971; Lind et al., 1972;
that had more violent severe weather. Rates peaked about Taylor, 1973; Greneker et al., 1976; Johnson et al., 1977;
1.5 hour before tornadoes and decreased during tornadoes to MacGorman et al., 1989]. Taylor, who examinedthe largest
about 40% of peak value. At frequenciesof 150 kHz and number of storms, found that roughly 20% of tornadic
higher, sfericsrateswere found to increaseuntil tornadoes storms did not have high sferics rates [MacGorman et al.,

were produced,and rates becameexceptionallyhighduring


periodswhen tornadoesoccurred[jones, 1951, I958; Jones
and Hess, 1952; Kohi, 1962; Kohl and Miller, 1963]. Kohl
and Miller [1963] observed that sferics rates at 150 kHz

1989].It also was possiblefor high sfericsrates to occur in

stormswithout tornadoes [Ward et al., 1965; StanJbrdet al.,


1971;MacGorman et al., 1989]. Taylor found that although

only 23% of nontornadicseverestormsand 1% of nonsevere


usuallypeakedduringsevereweatherand begandecreasing storms had sferics rates comparableto those of tornadic
prior to the end of severeweather.As might be expected storms, 70% of the storms with high sferics rates were not
from the above observations, Jones and Hess [1952] found
that the ratio of the numbers of higher-frequencyto lower-

tornadic, because the number of nonsevere storms was

and after tornadoes and in nonsevere thunderstorms versus

and the combined effect of several simultaneous

1:1 during tornadoes.

explainssomenonseverestormswith highsfericsrates[e.g.,

weatherwas greatestat frequenciesabove 1 MHz. This is


shownin Figure 1, which presentsa summaryof Taylor's
dataconcerning
the dependence
of sfericsrateson receiver
frequency.Silberg[1965]and Taylor[1973]alsofoundthat
theenergyradiatedassferics(notshown)increases
at higher
frequenciesas storm severity increases;radiated energy
peaksat roughly5 kHz for nonsevere
storms[Taylor,1972],
buttheenergyat higherfrequencies
graduallyincreases
with

storms.

muchlarger than the numberof tornadicstorms[MacGorfrequencysfericsincreasedfor sfericsof large amplitude man et al., 1989]. Sometornadic stormswithout high sferics
when storms were more severe; the ratio was 1:20 before rateshad only weak tornadoes[e.g., Johnsonet al., !977],
storms

Later studiesat frequencybandsup to 150 MHz [Silberg, MacGorman et al., 1989]. However, these situationsdid not
1965; Taylor, 1973;Stanford et al., 1971;Johnsonet al., accountfor all exceptionsto the rule that 3-MHz sferics
1977]found that the increasein sfericsrates during severe ratesare muchhigherin tornadicstormsthan in nonsevere
The mostobviousinferencefrom the increasingnumberof
sferics at high frequencies and the decreasingor constant
number at low frequencieswas that intracloudflash rates

usually increase while cloud-to-groundlightning return


strokes decrease or remain fairly constant during the tor-

nadicstageof storms.This interpretationwas further sup-

portedby the theoretical


work of Stanford[1971]and

MAcGORMAN

175

airborneobservations
of Shanmugatn
and Pybus[1971]. originatedalongazimuthsto low-levelreflectivitycoresand
Both reported that horizontalpolarizationof sfericsin- that sfericsrateswere largerfrom coreswith higherrefleccreasedand verticalpolarizationdecreased,indicatinga tivity. Furthermore, Brown and Hughes [1978]found that

greaterpreponderanceof horizontallightningchannels,as
the sfericsrate and stormseverityincreased.

the pattern in the number of VLF sferics(which are producedpredominantlyby return strokes)from the Union City
Other observationsalso supportedthis interpretation. tornadicstormwassimilarbothin timeandspaceto patterns
Jones[ 1958]reportedwatchinga stormwhenit wasproduc- in reflectivity at low levels of the storm.
ing high sfericsrates about 1 hour before a tornado;no
Although the sferics studies clearly demonstratedthat
groundflashescould be seen, but relatively dim circular lightningtended to evolve in characteristicpatternsin tor-

patches,about0.5 km in diameter,wereilluminated
in rapid nadic storms, there were a number of uncertainties. (1)
succession
on the sideof the storm.He suggested
that these There were uncertaintiesin extracting lightningflash type
were an unusualform of intracloudlightning.Electric field

and rates from sferics data. (2) No sferics study examined

recordsfor thissamestormindicatedthat the lightningflash the evolution of mesocyclones or considered differencesin

ratewas 12 min-1 duringthe tornado[Gunn,1956].Von- the types of storms that produced tornadoes in order to
negut and Moore [1957] inferred from recordingsof the

examinewhy sfericsrates evolved differently. (3) No sferics

cloud-to-ground
flashesnear the tornado. Unusuallyhigh

tornadic

closest electric field sensor to the tornado that there were no

lightningflash rates were also observedin two other tornadic


storm systemsby Orville and Vonnegut [1977] and Turrnan

and Tettelbach[1980], who usedsatellite-borneopticaldetectors. Films of 17 tornadoes and one funnel cloud docu-

mented low ground flash rates near tornadoes; Davies-Jones


and Golden [1975a] (see also Davies-Jones and Golden

[1975b, c], Colgate [1975], and Vonnegut [1975] for a


discussion)found cloud-to-groundlightningflashesnear only
two of the 18 filmed vortices. One of these two was a tornado

with 12 ground flashes;the other was a funnel cloudwith two


ground flashes.
From studies of sferics, however, it appears that the
tornado itself is not likely to be the cause of unusual sferics
rates. For example, Jones [1965] noted that sfericsrates can
be unusuallyhigh 60-90 min before tornadoes,althoughthey
may increase further when tornadoes occur. Scouten et al.
[1972] further analyzed the data of Jones [1958] at 150 kHz
and concluded

that touchdown

of the tornado

did not affect

the high sferics rate. Although larger-amplitudesfericsclustered more closely about a central core near the azimuth of
the tornado, sfericsas a whole were spreadamongazimuths
spanning most of the storm. A similar lack of correlation
with

the

time

and

location

of tornado

touchdown

was

reported for sferics in three bands between 300 kHz and 3


MHz [Taylor, 1973] and for sferics at 1-50 kHz [Brown and
Hughes, 1978]. Brown and Hughes [1978] also noted a lack
of obvious relationships with the tornadic vortex signature
[Brown et al., 1978]. In a study of sferics burst rates at high
frequencies (30-300 MHz) in two tornadic storms,Johnson
[1980]found that the tornadoesbegannear the time of peaks

study examined the three-dimensional vector wind field for


storms.

4.

LIGHTNING MAPPING STUDIES

Studies using lightning mapping systems have begun to


address some of the shortcomingsof the sferics studies.
However, only two of the mapping studies thus far have
includedintracloudlightning; all others have examinedonly
cloud-to-groundlightning. Furthermore, mapping studies
have found considerablevariability in the relationship of
tornadoes to cloud-to-ground flash rates. In two tornadic
storms, ground flash rates were small before and during
tornadoes [Orville et al., 1982; MacGorman et al., 1989]. In
three, ground flash rates peaked during tornadoes [Kane,
1991;MacGorman and Nielsen, 1991;Keighton et al., 1991].
In two, tornadoes occurred after ground flash rates had
increased but when ground flash rates were not at their
maximum [Kane, 1991; MacGorman et al., 1985].

Results from two of the studies suggestsome reasonsfbr


this variability. In the Binger tornadic storm studied by
MacGorman et al. [1989], negative ground flash rates within

10km of the mesocyclone


centerwere lessthan1 min-

until the tornado began dissipating. Negative cloud-togroundflash rates reached a relative maximum after the last
tornado, as its mesocyclone core dissipated, and reached an

absolute
maximum(about4 min-) as the lastmesocyclone

core in the storm dissipated (see Figure 2). Intracloud flash


rates reached an absolute maximum of approximately 14

min- duringthemostviolenttornadoandwerewellcorrelated with low-level cyclonic shear when the shear was

greaterthan1 x 10-2 s- duringthe tornado.(Cyclonic

shear is one half vertical vorticity for ideal solid-body


rotation. In calculating this shear, tornadic winds were
ignored;shear was measuredacrossthe diameterof maximumtangentialwind speedfor the mesocyclone,typically a
distance of a few kilometers.) After the last tornado, when
sferics data and suggestedthat these resulted from the therewas a largedecreasein low-levelcyclonicshearbut the
cyclical intensificationand decay of thunderstormcells mesocyclonewas still strong at middle levels, intracloud
[Kohl and Miller, 1963; Stanford et al., 1971; Shantnugam flashrateswereonly0-2 min- .
In the Edmond storm studiedby MacGorman and Nielsen
and Pybus, 197!; Lind et al., 1972; Trost and Nornikos,
1975].Taylor [1973]reportedthat sfericsat highfrequencies [1991] there were no intracloud data, but negative

in the sferics burst rate but were not coincident with any

peaks.
Several studiesconsideredchangesin tornadicstormsthat
were correlated with the high sferics rates. For example,
severalinvestigatorsfound periodicitiesof 10-60 min in the

176

REVIEW OF LIGHTNING IN TORNADIC STORMS

12

30

KM

20

1.5

KM

30
1.5

KM

20

,.i':::
=:

r,;':

10

< 40

'i

3o

1.2

0.8

0.4

12

0.0

z 10

10 KM

1803

1815

1830

TIME
0,

1845

....

i"

1900

1920

(CST)

Fig. 3. Time series plots of cyclonic shear at the 1.5- and 5-km
levels and of ground flash rates within 10 km of the mesocyclone
center in the Edmond storm of May 8, 1986 [from MacGorman and

Nielsen, 1991]. The bars on the bottom indicate when tornadoes

occurred.

1833

1900

1930

20'00 '

20'30 ' 2054

TIM E (CST)

TORNADO
[

' ..'

.'..':'iI []

Fig. 2. Time series plots of cyclonic shear at the 1.5- and 6-km
levels and of ground and intracloud flash rates within 10 km of the
mesocyclone core in the Binger storm of May 22, 1981 [from
MacGorman et al., 1989]. The bars on the bottom indicate when
tornadoes occurred.

flash rates within 10 km of the mesocyclonecenter evolved


differently than in the Binger storm. Rates increased to a

peakof 11min- during


tornadic
activityandappeared
to be
correlated with cyclonic shear at the 5 km level (Figure 3).
Ground strike locationstended to cluster in the vicinity of a
reflectivity core north of the mesocyclone during

MAcGORMAN

activity and to be more scattered before and after the


tornadic stageof the storm (Figure 4). Keighton et al. [1991]
observed a similar increase in scattering of lightning strikes
after tornadic activity on another day.
MacGorman and Nielsen [1991] suggestedthat observed
differences between the Binget and Edmond storms might
account for the differences in the evolution of their lightning
flash rates. The Binget storm was a classic supercell storm,
with a prominentweak-echoregion, a strong,deep updraft
core, and a long-lasting mesocyclonehaving a family of

177

(a)

70

6o
so
40

mesocyclonecores [Burgesset al., 1982]. The Edmond


storm also was a supercell storm but was weaker than the
Binger storm by almost any measure:the duration of its
supercellstagewas only 30% that of the Binget storm;its
mesocyclonewas weaker, shallower,and shorterlived; and

30

features such as the weak-echo vault and tornadic vortex

signaturewere not as pronounced.In fact, the Edmond

(b)

storm did not form a tornado until the outflow boundary of

anotherstormbeganto overtakeit from the west. MacGorman and Nielsen [1991] suggested
that the increasein flash
rates in both storms when the mesocyclone was strongest

probablywasa resultof thestrongupdraftthatexistedthen.

They alsosuggested,
however,that the deeperand stronger

updraftcore and weak echoregionof the Bingetstorm


probably
delayed
ground
flashactivityandcaused
intracloud
activity to dominate.

To understandthe hypothesisthat MacGorman and


Nielsen [1991] offeredto explainwhy groundflash rates
behaveddifferentlyin thesetwo storms,it is necessary
first

35

to understanda little aboutthunderstormelectrification.One

-15

important
classof thunderstorm
electrification
mechanisms

the microphysics
of the storm.Subsequent
transport
is

dominated
initiallyby thesedimentation
of graupel
butmay
becomedominated
by the relativevelocityof the windsin

different
regions
asthegraupel
andicecrystals
movefarther
apart.Thunderstorm
measurements
[e.g.,KrehbieI,
1981;
Byrnne
et al., 1983;Chauzy
et aI., !985;Dyeet al., I986;
Koshakand Krider, 1989]suggest
that mostof the main

negative
charge
at middle
levels
ofthestorm
isongraupel
andprecipitation,
whilethemain
positive
charge
isgenerally
onicecrystals
atcolder
temperatures
(thenextsection
will

10

20

7O

control the sizesand numbersof graupeland ice particles

collisions
includethe strength
anddepthof theupdraftand

0
(kin)

30

(c)

' (... 40,,.-50--.I. /

that collide, and (2) by factorsthat controlsubsequent


thunderstorm.Factors affecting the number of relevant

-5
EAST

relieson microphysical
particleinteractions
to placecharge
of one signon graupelandchargeof the opposite
signon
cloudice. Suchmechanisms
are affected(1) by factorsthat

transport
of thecharged
particles
intodifferent
regions
of a

-10

,.-, 60
E

I-.. 50
o

,,..F"20

4O

..

'4.0

30
I

-20

-lO

.......I

o
EAST

lO

,I

1__

20

(kin)

discussonepossibleexception).

Fig.4. Lightning
ground
strikes
superimposed
onradarreflectivityatthe3-kmleveloftheMay8, 1986,
Edmond
storm
during
three
Several
investigators
haveobserved
thata strengthening
periods:
(a)
1802-1806
CST,
(b)
1814-1817
CST,
and
(c)
1834-1838
updraft
ataltitudes
colder
than-20Cincreases
flashrates
[fromMacGorman
andNielsen,199!].Therewasa tornado
(seediscussion
byMacGorman
etal.[!989]).
ThisprobablyCST
onlyduring
theperiodin Figure
4b. Minuses
indicate
thestrike
explains
theobserved
increases
in flashratesduring
the pointofflashes
thateffectively
lowered
negative
charge
toground;
tornadic
stage
oftheBinget
andEdmond
storms.
MacGor-pluses,
flashes
thateffectively
lowered
positive
charge.
Thelarge
core.Radarreflectivity
is
manet al. [1989]alsoobserved
increasing
areasof large dotmarksthecenterof themesocyclone

reflectivity
atheights
between
6kmand8kmshortly
beforelabeledin dBZ.NotethatFigure4b hasa differentdistance

178

REVIEW

OF LIGHTNING

IN TORNADIC

STORMS

and duringperiodsof high flashrates.They suggested


that
the increasedflashratesprobablywere causedby increased
particleinteractionsleadingto reflectivitygrowthat 7-9 km
in and near the strong updraft. Similar correlationsof
lightningrateswith the horizontalareaof largereflectivityat
midlevelsof stormshave been observed,for example,by

TOTALCHARGEDENSITY (NC/M**3)
2{

i '

i'

i ....

Larson and Stansbury [1974], Lhermitte and Krehbiel


[1979], and Keighton et al. [1991].

One of the primary problemsposed by the Binger and


Edmondstormsis explainingwhy groundflashrateswere so
small near and during the time when total and intracloud
flash rates were large in the BingeTstorm and why this
apparently was not true for the Edmond storm. MacGorman

et al. [1989] suggestedthat the strong,deep updraftand the


resultingpronouncedweak-echoregionof the Bingerstorm
enhancedproductionof intracloud flashesand delayed or

suppressedproduction of ground flashes because of effects


on the thunderstorm charge distribution.
For example, one primary effect on the chargedistribution
is to keep negative charge higher than in most storms. This

would increasethe energyrequiredfor lightningto spanthe


distance to ground and would decrease the electric field at
most heightsbelow the negative charge. Negative charge is

higherfor three reasons.(1) The strongupdraft core rapidly


lifts all but the largestparticles to upper levels of the storm.
(2) Large updraft speedsdo not allow most hydrometeorsto
remain long in a given layer. The short residence time
prevents hydrometeors from acquiring much charge and
prevents positive and negative particles from moving far
apart after becoming charged. Therefore there is little net
negative or positive charge in a given layer of a strong
updraft. (3) Temperatures in stronger updraft cores are
generally warmer, and this causesthe ice processesthought
to be responsiblefor charging (see discussionof II!ingworth
[1985]) to occur somewhat higher.
A secondimportant effect of a strong, deep updraft is that
regions of net positive and negative charge would be relatively close together near the top of the storm updraft. This
occurs for two reasons. (1) As noted above, sedimentation

and differing relative velocities of positively and negatively


charged particles would not have time to separate the
charges very far in a fast updraft. (2) The vertical shear of
horizontal winds in a classic supercell storm causesgraupel
to fall to the side of the updraft core, so much of the
storm-scaleseparationof charge occurs outside the updraft.
When two charged regions are closer together, less net
charge is needed in each region to create electric field
magnitudessufficientto initiate lightningbetween them. This
enhances intracloud flash rates. It also reduces the negative
charge available for cloud-to-ground lightning near the updraft, because the threshold for initiating lightning regulates

2
0.

1 0.

2{3.

30.

4.

50.

60.

HORIZONTAL DISTANCE
Fig. 5. Total space charge density in a southwest-northeastvertical cross section through the main updraft, from a numerical
simulation of the Binger storm of May 22, 1981. Solid contours
indicate zero or positive net spacechargedensity; dashedcontours,

negativenet chargedensity.The contourintervalis 0.5 nC m-3.


Stippling
delineates
the regionof updraftexceeding
5 m s-1. The

thin horizontal curve near a height of 4 km indicates the melting


level. The bold curve outlines the cloud and precipitation boundary.

negative charges move farther apart (thereby increasing the


amount of net negative charge that can be separated before
causing intracloud lightning).
Since MacGorman et al. [ 1989] first suggestedthis hypothesis, there has been support from two sources. First, the
lowest charge detected by the only published balloon-borne
electric field soundingthrough a mesocycloneupdraft was a
region of negative charge starting at 9.5 km above mean sea
level, where the temperature was approximately -37C in
the storm's environment and -31C in the updraft [Rust et
al., 1990]. This lower boundary is 3-5 km higher, and at
environmental temperatures 20-30Ccolder, than the lower
boundary normally reported for the main negative charge in
continental thunderstorms. Second, Ziegler and MacGorman [1990] used a kinematic retrieval that included electri-

ficationprocesses[Ziegler et al., 1991]to model the electric


field and charge structure of the Binger storm. In agreement
with the balloon-borne measurements, the lower boundary

of negative charge inside the updraft core in the retrieved


charge distribution was elevated to a higher altitude than

normal (Figure 5).


MacGorman and Nielsen [1991] suggestedthat the charge
the net negative(and positive) chargethat can be separated. distribution near the updraft of the Edmond storm was more
This, combined with the unusually large height of negative like that of an ordinary thunderstorm and that cloud-tocharge discussedabove, makes cloud-to-groundlightning groundflashesthereforewere able to occur frequently when

much less likely to occur near a very strong, deep updraft


core. Ground flash rates increase later, as the negative

charge moves closer to ground and as the positive and

the mesocyclonewas strong.They concludedthat negative


charge was lower, a larger fraction of oppositelycharged
particleshad time to separate,and positive and

MAcGORMAN

charge were farther apart because(1) the updraft core was


weaker, shallower, and shorter lived, (2) the resultingweakecho region was shallower and contained larger reflectivities, and (3) the 55-dBZ reflectivity cores extended to a
height of only 5-6 km in the Edmond storm versus over 10
km in the Binger storm. Similarly, Keighton et al. [1991]
observed another storm in which cloud-to-ground flash rates
did not decreaseduring the tornadic stageand noted that it
also had a weaker mesocyclone than the Binger storm.
5.

179

30
w

20

FO

POSITIVE CLOUD-TO-GROUND

Ill

F1

F4

F1

FO

1800

1900

LIGHTNING IN TORNADIC STORMS

Data on the less common flashes that effectively lower

positivechargeto ground(+CG flashes)havebeenstudied

30

in detail for only one tornadic storm, the Edmond storm

investigatedby MacGorman and Nielsen [1991]. (Reliable


positive cloud-to-grounddata were not availablefor the
Bingerstormor for othertornadicstormsbefore1983.)In
the Edmond storm, positive cloud-to-groundflashesbegan

% 20

to occur a few minutes before the tornado touched down,

andpositivegroundflashratesincreased
to theirmaximum
value during the tornado, near the time when negative
groundflashratesalsopeaked(seeFigure3). MacGorman

15()0

1600

1700

2000

2100

TIME (CS'I)

and Nielsen [1991] noted from watchinggroundflash dis-

playsfor severaltornadicstormsin realtimethatpositive


plotsof thenumber
of positive
andnegative
groundflashesoftenoccurredduringtornadoes.
Sometimes Fig.6. Timeseries
cloud-to-ground
flashes
in each15-min
intervalduringa stormthat
a fewpositivegroundflashes
wereclustered
onthesouthern produced
a violent
tornado
nearHutchinson,
Kansas,
onMarch26,
flank of stormsin which negativecloud-to-ground
lightning 1991. Bars indicatethe time of tornadoes,labeledwith F scale
Theapproximate
timeoflargehailreports
areindicated
by
still dominatedgroundflashactivity (as in the Edmond ratings.

storm),and sometimes
mostof the groundflashesin the

the leadingedgeof Hs.

stormwere positivegroundflashes.

Althoughmanysevereor tornadicstormsdo not have

largenumbers
of positiveground
flashes,
preliminary
evidencesuggests
that whenrelativelyhighflashratesand tornadonearHutchinson,Kansas,on March26, 1991.This
produced
severe
weather
throughout
muchofitslife.
densities
of positivecloud-to-ground
lightning
occur,they storm
are associatedwith severeweather. Reap and MacGorman

In the secondmode the dominant polarity of ground


flashesin the stormswitchesfrompositiveto negative,with

[1989]reportedin a climatological
studythattherewas a
groundflashratescomparable
to orlarger
correlation
betweena highdensityof positivegroundflashes thepeaknegative
flashrates.An example
andsevereweather.Furthermore,in August1990andspring thantheearlierpeakpositiveground
modeisgivenin Figure7, whichshows
ground
1991,MacGorman
and Burgess
[1991]observed
several ofthesecond
flash
rates
for
the
storm
that
produced
an
F5
tornado
in
stormsthat wereunusualbecause(1) mostcloud-to-ground
Plainfield,
Illinois,
on
August
28,
1990.
Prior
to
the
F5
flashes
werepositive
ground
flashes
and(2)ground
strike
pointsof positive
cloud-to-ground
lightning
occurred
in tornadomostgroundflashesin the stormwerepositive.

ratedupto
denseclusters,muchlike the clusters
of negative
ground Duringthisperiod,largehailandfourtornadoes
F2
were
produced.
Three
of
these
tornadoes
occurred
during
flashesobserved
in mostelectrically
activestorms.(Positive
the
period
of
maximum
positive
ground
flash
rates.
ground
flashes
aregenerally
infrequent
anddiffuse.)
Allof
lightning
thestorms
withfrequent
positive
ground
flashes
produced The dominantpolarity of cloud-to-ground

largehail,andmanyproduced
tornadoes.
Similar
observa-switchedas the Plainfieldtornadobegan.Most subsequent
flashes
werenegative,
asshown
clearly
in thetime
tionswerereported
by Rustet al. [1985],CurranandRust ground
series
plots
in
Figure
7.
Also
note
that
there
was
a decrease
[1992],andBranickandDoswell[1992].
flashactivityshortly
beforeandduring
the
These stormswith unusualpositivelightningactivity inoverallground
tornado,muchas described
earlierfor negative
ground
appear
to occurin at leasttwomodes.
In one,positiveF5
flashesin sometornadicstorrns.Negativegroundflashrates
ground
flashes
dominate
cloud-to-ground
activity
throughout
increased
to theirlargestvaluesat theendof thetimeseries
the lifetimeof the storm,including
all tornadicactivity.An

wasmerging
withotherstorms
toform
example
ofa storm
inthismode
isshown
inFigure
6,which plot,whenthestorm
behavior
observed
presents
ground
flash
rates
forthestorm
thatproduced
anF4 a squallline (similarto lightning

180

REVIEW OF LIGHTNING IN TORNADIC STORMS


400

high-precipitationsupercell was related to the transition in

,1
.....
,i,i i

350

groundflashpolarity. A similartransitionwas observedby


Curran and Rust [1992].

MacGorman and Nielsen [1991] suggestedtwo reasons


why positive ground flashes might occur in severe storms.

200

:5 5o

First, the structureof supercellstorms,in whichprecipita-

tion at middleto low levelsis displacedhorizontallyfrom the


updraft core, might lead to a tilted bipolar chargedistribution, with a lower negativechargeoccurringin the vicinity of
precipitationandpositivechargeoccurringon ice crystalsat
the top of the updraftcore, abovethe weak echoregion.The
tilt might be sufficientfor the upper positive chargeto be
effectively exposed to the ground, as for winter storms
observedby Brook et al. [1982]. Branick and Doswell [1992]
have suggesteda variation of this: not only does the horizontal displacementof the updraft and precipitationtilt the
dipole, but the lack of low-to-middlelevel precipitationin
low-precipitationstormsinhibitsformation of a large region

o, oo
5O

F1

FO-2

F5

IIIII

HH HHH

400

350
/

200

5 5o
+

lOO
50
o

1200

1300

1400

1500

1600

TIME (CST)

Fig. 7. Same as Figure 6 but for the Plainfieldstormon August28,


1990. l,arge hail reportswhose H symbolswould overlap earlier Hs

are omitted.

Goodman and MacGorman [1986] when storms merged to


form a mesoscaleconvective complex).
Visual

observations

indicate

that

the

character

of the

Plainfield storm changedat approximately the time when the


last and most violent tornado beganand groundflashpolarity
switched. Initially, the storm grew rapidly into a supercell,
and it remained a supercell for much of its life. During this
period the storm had a prominent, deep weak-echo region
and produced large hail. At about the time that the Plainfield
tornado began, heavy precipitation began to wrap around the
mesocyclone, filling in the weak-echo region. Observers of
the Plainfield storm noted that the tornado was obscured by
heavy precipitation.
Curran and Rust [1992] and Branick and Doswell [1992]

have observed that low-precipitation supercell storms often


produce high densities of positive ground flashesand that
nearby supercell storms with heavy precipitation produce
the normal preponderance of negative ground flashes. At
least some severe storms (including the earlier stagesof the
Plainfield storm) that have high numbers and densities of
positive groundflashesare not low-precipitationstorms,but
there are no reportsof this positivecloud-to-groundlightning
signaturein high-precipitationsupercellstorms.Therefore it
appears that the transition of the Plainfield storm to a

of negative charge.
The second mechanism discussedby MacGorman and
Nielsen [1991] involved a possible region of significant
positive charge beneath the main negative charge of the
thunderstorm. Such a region could be formed in different
ways. For example, several investigators have suggested
that a lower positive charge could be formed in thunderstorms by a noninductive mechanism studied in the laboratory by Takahashi [1978], Jayaratne et al. [1983], and
others. In most regions in which this mechanism for microphysical charge separation might occur in thunderstorms,
collisions between graupel and ice crystals place negative
charge on graupel and positive charge on ice crystals. These
collisionsresult in the normal tendencyfor the main positive
charge to be generally above the main negative charge. At
temperatures warmer than some threshold between -10
and -20C, however, this is reversed: positive charge is
placed on graupel; negative charge, on ice crystals. If
enough collisions occur between 0C and this reversal temperature, they could create a region of positive charge on
graupel beneath the main negative charge of thunderstorms.
6.

CONCLUSION

Obviously, the studiesreviewed in this paper fall far short


of fully characterizingthe diversity of lightning evolution in
tornadic storms. Most studies examined lightning behavior
relative just to the existenceof tornadoes. Only a few studies
examinedlightningevolution relative to the evolution of the
reflectivity structure, mesocyclone,or wind field of tornadic
storms. To test adequately the hypothesized relationship
between updraft characteristicsand suppressionof ground
flashesrequiresa more systematicstudy of large data setsof
lightning and storm characteristics,possibly supplemented
by numerical simulations.Such studiesalso may be able to
determine better the underlying relationship that caused the
observed close correlation of low-level cyclonic shear and
intracloud flash rates in the Binger storm.
If the hypothesesdiscussedin this paper are correct,

MAcGORMAN

expect that more comprehensive studies of tornadic storms

will findthefollowing.Almostall stormsthathavemesocyclones will have an increase in intracloud and total flash rates

as the updraft increasesin depth, horizontalextent, and


speedat altitudesabovethe -20C level,therebyincreasing
the numberof hydrometeorcollisionsthat generatecharge.
Intracloud flash rates will be enhanced even more in classic

181

relationship between supercell structure and lightning ground


strike polarity, Weather Forecasting, 7, 143-149, 1992.
Brook, M., M. Nakano, P. Krehbiel, and T. Takeuti, The electrical
structure of the Hokuriko winter thunderstorms, J. Geophys.
Res., 87, 1207-1215, 1982.
Brown, R. A., and H. G. Hughes, Directional VLF sfericsfrom the
Union City, Oklahoma, tornadic storm, J. Geophys. Res., 83,
3571-3574, 1978.

,supercell
storms,which tendto have very largeand strong Brown, R. A., L. R. Lemon, and D. W. Burgess,Tornado detection
by pulsed Doppler radar, Mon. Weather Rev., 106, 29-38, 1978.
updrafts that are quasi-steady,in contrastto updraftsin Burgess, D. W., V. T. Wood, and R. A. Brown, Mesocyclone

lesser thunderstorms.In some tornadic storms, cloud-toevolution statistics, in Preprints, 12th Conference on Severe
groundflasheswill begin and peak shortly after intracloud
Local Storms, pp. 422-424, American MeteorologicalSociety,
Boston, Mass., 1982.
lightning (roughly 10-15 rain later), as in most nonsevere
storms. However, in storms that have updraftslarge and Byrnne, G. J., A. A. Few, and M. E. Weber, Altitude, thickness,
and charge concentrationsof charged regions of four thunderstrongenoughto create prominent, deep weak-echoregions
stormsduringTRIP 1981 basedupon in situ balloonelectricfield
(with high lower boundariesof negativecharge)we expect
measurements,Geophys. Res. Lett., I0, 39-42, 1983.
cloud-to-groundlightningnear the updraft and mesocyclone Chauzy, S., M. Chong, A. Delannoy, and S. Despiau,The 22 June
to be suppressed,so that few ground flashesoccur when the
updraft is strong and ground flash rates peak when the
updraft finally weakens.
There are far fewer data on positive cloud-to-ground
lightning in severe storms, so observed relationships and
suggestedhypothesesshould be regarded as more tentative.
If a tilted chargedistributioncausespositivecloud-to-ground
lightning, then we would expect to see positive ground
flashes in storms in which the top of the main updraft is
displaced sufficientlyfar horizontally from the reflectivity
core. If the described mechanism for the lower positive
charge causespositive cloud-to-ground(+CG) flashes,then
we would expect to see + CG lightningin stormsin which the
number of collisions between graupel and ice crystals is
enhanced at temperatures between freezing and roughly
-15C. Such a situation could occur, for example, if there is

significantrecirculationof graupel,as hasbeenreportedfor


some hail storms.

tropical squallline observedduringCOPT 81 experiment:Electrical signatureassociatedwith dynamicalstructureand precipitation, J. Geophys.Res., 90, 6091-6098, 1985.

Church, C. R., and B. J. Barnhart, A review of electrical phenomena associatedwith tornadoes,in Preprints, 1Ith Conferenceon
Severe Local Storms, pp. 337-342, American Meteorological
Society, Boston, Mass., 1979.

Colgate,S. A., Commenton "On therelationof electricalactivityto

tornadoes," by R. P. Davies-Jonesand J. H. Golden, J. Geophys.


Res., 80, 4556, 1975.
Curran, E. B., and W. D. Rust, Positivegroundflashesproducedby

low-precipitationthunderstormsin Oklahomaon 26 April 1984,

Mon. Weather Rev., I20, 544-553, 1992.

Davies-Jones,R. P., and J. H. Golden, On the relation of electrical


activity to tornadoes,J. Geophys.Res., 80, 1614-1616,1975a.
Davies-Jones,R. P., and J. H. Golden, Reply, J. Geophys.Res., 80,
4557--4558, 1975b.

Davies-Jones,R. P., and J. H. Golden,Reply, J. Geophys.Res., 80,


4561--4562, 1975c.

Dickson, E. B., and R. J. McConahy, Sferics readingson windstorms and tornadoes, Bull. Am. Meteorol. Soc., 37, 410--412,
1956.

It may be difficult to determinewhich of theseor other Dye, J. E., J. J. Jones,W. P. Winn, T. A. Cerni, B. Gardiner,D.
mechanismsare responsiblefor positive cloud-to-ground Lamb, R. L. Pitter, J. Hallet, and C. P. R. Saunders,Early
lightning.For example, the large separationreportedbeelectrificationand precipitationdevelopmentin a small isolated
Montanacumulonimbus,J. Geophys.Res., 9I, 1231-1247,1986.
tweenthe mesocyclone
andprecipitationfor somestormsin

which positive ground flashesdominate [Branick and Goodman,S. J., and D. R. MacGorman, Cloud-to-groundlightning

Doswell, 1992] could result from either scenario. There

probablywouldbe significanttilt in the chargedistribution,


and a significantnumberof recirculatinggraupelparticles
mightpassthroughthe layerbetween0and-15C.Distinguishingwhich mechanismdominateswill requireat least
someof the following:simultaneous
electricfield measurementsin upperregionsof the updraftandin heavyprecipitation,improvedDopplerradardata, simulations
of severe
storms,and three-dimensional
lightningmappingsystems.
REFERENCES

activityin mesoscaleconvectivecomplexes,Mon. WeatherRev.,

114, 2320-2328, 1986.


Greneker, E. F., C. S. Wilson, and J. I. Metcalf, The Atlanta
tornado of 1975, Mort. Weather Rev., 104, 1052-1057, 1976.
Gunn, R., Electric field intensityat the groundunderactive thunderstorms and tornadoes, J. Meteorol., 13, 269-273, 1956.

Illingworth,A. J., Chargeseparation


in thunderstorms:
Small-scale
processes,
J. Geophys.Res., 90, 6026-6032,1985.
Jayaratne,E. R., C. P. R. Saunders,and J. Hallett, Laboratory
studiesof the chargingof softhail duringice crystalinteractions,
Q. J. R. Meteorol.Soc., I09, 609-630, 1983.
Johnson,H. L., R. D. Hart, M. A. Lind, R. E. Powell, and J. L.
Stanford,Measurementsof radiofrequencynoisefrom severeand
nonsevere thunderstorms, Mort. Weather Rev., 105, 734-747,
1977.

Bent, R. B., andW. A. Lyons,Theoreticalevaluations


andinitial Johnson,R. L., Bimodal distributionof atmosphericsassociated
with tornadicevents,J. Geophys.Res., 85, 5519-5522,1980.
operational
experiences
of LPATS(Lightning
Position
andTrackandtracking,
ing System)to monitorlightning
groundstrikesusinga time-of- Jones,H. L., A sfericmethodof tornadoidentification
arrival(TOA)technique,
in Preprints,7thInternational
Confer- Bull. Am. Meteorol. Soc., 32,380-385, 1951.
ence on AtmosphericElectricity, pp. 317-324, American Jones,H. L., The identificationof lightningdischargesby sferic
characteristics,in Recent Advancesin AtmosphericElectricity,
MeteorologicalSociety,Boston,Mass., 1984.
editedby L. G. Smith,pp. 543-556,Pergamon,
New York,
Branick,M. L., and C. A. DoswellIII, An observation
of the

182

REVIEW

OF LIGHTNING

IN TORNADIC

STORMS

Jones, H. L., The tornado pulse generator, Weatherwise, 18, 78-79,


85, 1965.
Jones, H. L.,

and P. N. Hess, Identification of tornadoes by


observation of waveform atmospherics, Proc. IRE, 40, 10491052, 1952.
Kane, R. J., Correlating lightning to severe local storms in the
northeastern United States, Weather Forecasting, 6, 3-12, 1991.

Keighton, S. J., H. B. Bluestein, and D. R. MacGorman, The


evolution of a severe mesoscale convective system: Cloud-toground lightning location and storm structure, Mon. Weather
Rev., 119, 1533-1556, 1991.

Kohl, D. A., Sfericsamplitudedistributionjump identificationof a


tornado event, Mon. Weather Rev., 90, 451-456, 1962.

Kohl, D. A., and J. E. Miller, 500 kc/sec sferics analysisof severe


weather events, Mort. Weather Rev., 91,207-214, 1963.
Koshak, W. J., and E. P. Krider, Analysisof lightningfield changes

duringactive Floridathunderstorms,
J. Geophys.Res., 94, 11651186, 1989.

Krehbiel, P. R., An analysisof the electric field changeproducedby


lightning,Ph.D. dissertation,400 pp., Inst. of Sci. and Technol.,
Univ. o1Manchester, Manchester, England, 1981. (Available as
Rep. T-11, Geophys.Res. Center, N. Mex. Inst. of Mining and
Technol., Socorro.)

Krider, E. P., A. E. Pifer, and D. L. Vance, Lightning directionfindingsystemsfor forestfire detection,Bull. Am. Meteorol.Soc.,
61, 980-986, 1980.

Larson, H. R., and E. J. Stansbury, Associationof lightningflashes


with precipitationcoresextendingto height7 kin, J. Atmos.Terr.
Phys., 36, 1547-1553, 1974.
Lhermitte, R., and P. R. Krehbiel, Doppler radar and radio observations of thunderstorms, IEEE Trans. Geosci. Electron., GE-17,
162-171,

1979.

Lind, M. A., J. S. Hartman, E. S. Takle, and J. L. Stanford,Radio


noise studies of several severe weather events in Iowa in 1971, J.
Atmos. Sci., 29, 1220-1223, 1972.

pp. 448-451, American Meteorological Society, Boston, Mass.,


1982.

Orville, R. E., R. W. Henderson, and L. F. Bosart, An east coast

lightning detection network, Bull. Am. Meteorol. Soc., 64, 10291037, 1983.

Reap, R. M., and D. R. MacGorman, Cloud-to-ground lightning:


Climatological characteristics and relationships to model fields,
radar observations, and severe local storms, Mon. Weather Rev.,
117, 518-535, 1989.
Rust, W. D., D. R. MacGorman, and S. J. Goodman, Unusual
positive cloud-to-groundlightningin Oklahoma stormson 13 May
1983, in Preprints, 14th Conferenceon Severe Local Storms, pp.
372-375, American Meteorological Society, Boston, Mass., 1985.
Rust, W. D., R. Davies-Jones,D. W. Burgess,R. A. Maddox, L. C.
Showell, T. C. Marshall, and D. K. Lauritsen, Testing a mobile
version of a cross-chain Loran atmospheric sounding system
(MCLASS), Bull. Am. Meteorol. Soc., 71, 173-180, 627, 1990.
Scouten, D.C., D. T. Stephenson,and W. G. Biggs, A sferic rate

azimuth-profileof the 1955 Blackwell, Oklahoma, tornado, J.

Atmos. Sci., 29, 929-936, 1972.

Shanmugam,K., and E. J. Pybus,A note on the electricalcharacteristicsof locally severe storms, in Preprints, 7th Conferenceon
SevereLocal Storms,pp. 86-90, AmericanMeteorologicalSociety, Boston, Mass., 1971.

Silberg, P. A., Passiveelectricalmeasurements


from three Oklahoma tornadoes, Proc. IEEE, 53, 1197-1204, 1965.

Stanford,J. L., Polarizationof 500 kHz electromagneticnoisefrom


thunderstorms:A new interpretationof existing data, J. Atmos.
Sci., 28, 116-119, 1971.

Stanford,J. L., M. A. Lind, and G. S. Takle, Electromagneticnoise


studies of severe convective storms in Iowa: The 1970 storm

season,J. Atmos. Sci., 28, 436-448, 1971.

Takahashi,T., Rimingelectrificationas a chargegenerationmech-

anism in thunderstorms, J. Atmos. Sci., 35, 1536-1548, 1978.


andseverestorms,in RemoteSensing
MacGorman, D. R., and D. W. Burgess,Positive cloud-to-ground Taylor, W. L., Atmospherics

flashesduring strongconvectionin severe storms,Eos Trans.

of the Troposphere,
editedby V. E. Derr, pp. 17-1-17-17,U.S.

MacGorman,D. R., andK. E. Nielsen,Cloud-to-groundlightningin


a tornadic storm on 8 May 1986, Mon. Weather Rev., 119,

Taylor, W. L., Electromagnetic


radiationfrom severestormsin
OklahomaduringApril 29-30, 1970,J. Geophys.Res., 78, 8761-

AGU, 72, Fall Meeting Suppl., 93, 199!.


I557-1574,

1991.

GovernmentPrintingOffice, Washington,D.C., 1972.


8777, 1973.

Trost, T. F., andC. E. Nomikos,VHF radioemissions


associated
with tornadoes,J. Geophys.Res., 80, 4117-4118, 1975.
LPATSlightning
groundstrikemappingsystems,in Preprints,5th
International Con3rkrence
on Interactive Information and Pro- Turman, B. N., and R. J. Tettelbach, Synoptic-scalesatellite
lightningobservationsin conjunctionwith tornadoes,Mort.
cessing
Systems,
pp.249-254,AmericanMeteorological
Society,

MacGorman,D. R., and W. D. Rust, An evaluationof the LLP and

Boston, Mass., 1989.

Weather Rev., 108, 1878-1882, 1980.

lightningdetectionby a direction-finder
network,J. Geophys.

nomenaassociatedwith nocturnal tornadoesin Huntsville, Ala., 3

MacGorman, D. R., and W. L. Taylor, Positive cloud-to-ground Vaughan,O. H., Jr., and B. Vonnegut,LuminouselectricalpheRes., 94, 13,313-13,318, 1989.

MacGorman,D. R., W. D. Rust, andV. Mazur, Lightningactivity


and mesocyclone
evolution,17 May 1981, in Preprints,I4th

Conferenceon SevereLocal Storms,pp. 355-358,American

MeteorologicalSociety,Boston,Mass., 1985.
MacGorman,12).R., D. W. Burgess,V. Mazur, W. 12).Rust,W. L.

April 1974,Bull.Am. Meteorol.Soc.,57, 1220-1224,1976.


Vonnegut,
B., Comment
on "On therelationof electricalactivityto
tornadoes,"by R. P. Davies-Jones
andJ. H. Golden,J. Geophys.
Res., 80, 4559-4560, 1975.

Vonnegut,
B., andC. B. Moore,Electricalactivityassociated
with
the Blackwell-Udall tornado, J. Meteorol., 14, 284-285, 1957.

B., andJ. R. Weyer,Luminous


phenomena
in nocturnal
Taylor,andB.C. Johnson,
Lightning
ratesrelativeto tornadic Vonnegut,
storm evolution on 22 May 1981, J. Atmos. Sci., 46, 221-250,

tornadoes,Science,153, 1213-1220, 1966.

Ward,N. B., C. H. Meeks,andE. Kessler,Sfericsreceptionat 500


kc/sec,radarechoes,and severeweather,in Paperson Weather
Mach, D. M., D. R. MacGorman,W. D. Rust, andR. T. Arnold,
Radar,Atmospheric
Turbulence,Sferics,and Data Processing,
Site errors and detectionefficiencyin a magneticdirection-finder
Tech. Note 3-NSSL-24,pp. 39-71, Natl. SevereStormsLab.,
networkfor locatinglightningstrikesto ground,J. Atmos.Oce1989.

anic Technol., 3, 67-74, 1986.

Norman, Okla., 1965.

stadt, Germany, 1977.

1990.

Observed
lightning
flash
Orville,R. E., andB. Vonnegut,
Lightning
detection
fromsatellites, Ziegler,C. L., andD. R. MacGorman,
ratesrelativeto modeledspacechargeandelectricfielddistribuin ElectricalProcessesin Atmospheres,editedby H. Dolezalek
tions in a tornadicstorm (abstract),Eos Trans.AGU, 71, 1238,
and R. Reiter, pp. 750-753,DietrichSteinkopffVerlag,Darm-

P.S. Ray,andJ. E. Dye, A


Orville,R. E., M. W. Maier,F. R. Mosher,D. P. Wylie,andW. D. Ziegler,C. L., D. R. MacGorman,
graupel-ice
charging
in the early
Rust,The simultaneous
displayin a severestormof lightning modelevaluationof noninductive
electrification
of
a
mountain
thunderstorm,
J.
Geophys.
Res., 96,
ground
strikelocations
ontosatellite
images
andradarreflectivity
patterns,
in Preprints,
I2th Conference
onSevere
LocalStorms, 12,833-12,855,

You might also like