Professional Documents
Culture Documents
97JD03230 2
97JD03230 2
97JD03230 2
Joseph E. Borovsky
Space and Atmospheric SciencesGroup, Los Alamos National Laboratory, Los Alamos, New Mexico
and in AppendixC the electrostaticpotentialdifference ciated image charges. During the dart-leader phase,
between a thundercloud and the ground is estimated. electrostaticenergyis loadedaroundthe channel(with
somechannelheatingoccurringduringthis phase).Dur-
ing the return-stroke phase, this stored electrostatic
2. Energy Flow in Dart Leaders and energy is expendedto heat the lightning channel. In
Return Strokes all phases,the energy is transported by meansof an
electrodynamicPoynting flux. During the dart-leader
In a recentlydevelopedelectrodynamicdescriptionof phase,the energy flowschiefly from the thundercloud
lightning[Borovsky, 1995],the movement of chargeand generatoralongthe outsideof the lightningchannelto
energyis causedby propagatingelectricwavesthat are the downgoingdart-leaderwavefront, wherethe flowing
guidedby the conductinglightningchannel.The dart energyis loadedaroundthe outsideof the channel(as
leader is described by a downgoing wave propagating E 2/8• energydensityin space)
andontothechannelit-
slowlyalonga coolchannel,and the return strokeis de- self(asOhmicheatingof the channelgas).Duringthe
scribedby an upgoingwavepropagatingrapidly alonga return-strokephasethe Poyntingflux is radially inward
warm channel. The dart leader, as it propagates,loads at the upward-propagatingwave front, indicatingthat
chargeonto the channeland loadselectrostaticenergy the E2/8• storedelectrostatic energythat wasoutside
E2/8?raroundthe channel(whileslightlyheatingthe the channel is being deliveredonto the channelwhere
channel),and a return stroke,as it propagates, drains the channel heating occurs. As can be inferred from
this channel charge to the ground and dissipatesthis Figure 1, the net result of a dart leader and its sub-
stored electrostatic energy into channel heating. The sequentreturn strokeis a transfer of chargefrom the
aspect of the electrodynamicdescriptionwherein the
dart leader acts to load charge onto a channel is sim-
ilar to earlier descriptionsof leaders[e.g., $chonland,
1938; Meek, 1939; Wagner and Hileman, 1961], and li tnngcharge
f'low charge
on
ground
the aspect of the electrodynamicdescriptionwherein
the return stroke acts as a drainage-initiation wave for :
:_ /
• ' charge
in
cloud ' '%
charge
onchannel,• '--.....
chargeon the channelis similar to earlier descriptions
[e.g., Price and Pierce, 1977; Linet al., 1980; Dien-
dorfer and Uman,1990]. However,the electrodynamic
model differssubstantially from theseearlier treatments
of leaders and return strokesin that the electrodynamic
model accounts for the flow of energy associatedwith .'" I\
the flow of charge. Additionally, the propertiesof the
charge-movingwaves that propagate along the chan-
nels are calculated from first principles in the electro- i i I I , , I •
dynamic model.
In Figure I the flow of charge and energy during a
• efore• dart-leader/ after/
dart leader phase phase return stroke
dart leader and return stroke is depicted. In the top ' ' ' ' I ' ' ' ' I .... I ....
panel of the figure, the locationof the charge(•1owthat lightningenergyflow oh• heating,
dissociation, .
is lowered from the cloud to the ground is followedwith , energy
stored andionization•
time. (Note that (•1ow is not the total amountof charge ' /incloud
\ ,/ energy stored
.•
,' .....
in the thundercloud, but only the amount that will be "• along
channel . I
moved to the ground during one dart-leader/return- ß. I
strokepair.) Prior to the dart-leaderphase,the charge
residesentirely in the thundercloud. During the dart-
leader phase, the charge is transferred from the cloud
to the channel(onto the conductingchanneland into
the coronasheath). Duringthe return-strokephasethe
chargeis transferredfrom the channelto the ground.
For simplicity in Figure I the flow from the cloud is
taken to be shut off at the onset of the return-stroke
phase,but causally,this cannotreally be the case,since
• before
/•dart. leader/
dart leader
••ret•n-strok•
after
/ phase phase return stroke
information about the initiation of the return stroke
Time
takes a finite amount of time to reach the cloud.
Figure 1. In the top panelthe flowof chargeduringa dart
In the bottom panel of Figure I the flow of energyis leader and return stroke is depicted and in the bottom panel
depicted.Initially,the energytakesthe formof E2/8• the flow of energy during the dart leader and return stroke
electrostaticenergy associatedwith the chargesepara- is depicted. Note that the plot is for the entire lightning
tion produced within the thundercloudand the asso- channel, not just a segmentof the channel.
BOROVSKY: LIGHTNING ENERGETICS 11,539
cloud to the ground and a transfer of energy from the RF emissions[e.g., Proctor, 1981, Figures14 and 15;
thundercloud to the channel. Proctor, 1983, Figures6-11]. Similarly,assumingthat
For a stepped-leader/return-stroke pair, Figure1 also the optical emissionsare caused by ohmic heating of
describesthe movementof chargeand energy. Similarly, air as current flows in channels, the network of feeder
the net result of a stepped leader and return stroke is channelscan be optically visualized in the casesof clear-
that charge is lowered from the thundercloud to the air lightning [e.g., Waldteufelet al., 1980, Figure 3].
ground and energy is transferred from the thundercloud The presence of a network of electrically conducting
generatorto the channel. Additionally, the channelis feeder channels is also consistent with radar probing
created. of lightningwithin thunderclouds[e.g., Holmeset al.,
1980], and feedernetworksare consistentwith electri-
cal dischargesof volume spacechargein the laboratory
3. The Energy per Unit Length Dissipated
[e.g.,Williams et al., 1985, Figure 2]. As a new region
in a Lightning Channel of a thundercloud is drained of charge with each sub-
To estimate the amount of energythat is depositedin sequent stroke in a lightning flash, a new set of feeder
a lightning channel,the quantity Q,owA•totis typically channelsforms [e.g.,Rustanet al., 1980, Figures 3-6;
used[e.g., Wilson,1929;Bruceand Golde,1941;Malan; Rhodeset al., 1994, Figure 4]. For variouslightning
1963; Golde,1977],whereQ•owis the amountof charge calculations, the various channels come into play. For
that is lowered from the thundercloud to the ground example, when calculating the charge per unit length
via the lightning channeland where A•to• is the elec- on a channel, the main channel and the blind chan-
trostatic potential differencebetween the thundercloud nels come into play for stepped leaders and first return
charge region and the ground. To estimate the energy strokes, but only the main channel comesinto play for
dissipationper unit lengthœ/L in a channel,the quan- dart leadersand subsequent
return strokes.(It is likely
tity Q•owA•o• is typically divided by the length of the that the feeder channels do not come into play, since
main channelLmai,[e.g., Uman,1984,1987],giving they are mostly drained of charge during these phases
as the negative charge is attracted downward toward
œ Qow/Xo the ground.) For another example, when calculating
Z • Lmain ' (1) the energy expended in creating channels, the feeder
channelsmust be included along with the main channel
It will be argued in this sectionthat expression(1) (and the blind channelsif the first strokeis being dealt
overestimatesthe energy dissipated per unit length in with) becausethesebranchchannelsrequireenergyfor
a lightning channel becausea large portion of the en- formation and heating. Accordingly,if one were to use
ergy (•owA•tot is expendedin branch channelswithin expression(1) to estimate the energy dissipationper
and below the cloud. (The overestimateoccursboth unit length in a channel,the length of the main channel
for stepped-leader-fed channels and for dart-leader-fed is not the appropriate value to use for L ....i,. Unfor-
channels.)Additionally,the usefulness
of expression
(1) tunately, accountingfor the feeder channelsand blind
is limited by the difficulty in estimating A•tot.
The channel structure of lightning is sketchedin Fig-
ure 2: as indicated there are branch channels below the
Channel Structure of Lightning
thundercloud, denoted here as blind channels(dead-
end channels),and there are branch channelswithin feeder
channels
the thundercloud, denoted here as feeder channels. The
blind channels are formed during the step-leader phase.
They are charged during this phase, and so during this
phasethere is an /•2/87r energystoredaroundthem.
• blind
chann
(first stroke only)
channelsin calculationssuchas expression(1) is diffi- ing channel. The total amount of electrostaticenergy
cult becausethe total length of the channelnetwork is per unit lengthœstorea/L storedalongthe outsideof the
difficult to estimate. Hence, expression(1) is of little chargedchannelis obtainedby integratingexpressions
value, unlessone wishesto use it in the form Leff - (4) overthe volumeoutsidethe channel:ϥtorea/L -
i•lowA(•tot/(•/L) to estimatethe effectivelengthLeff of f•e(E2/8•r)
2•r9o
d9o.
Breaking
theintegral
upasf0*ed9o
the branch-channel networksoncea valuefor œ/L is at
hand. : •rbreak
do dr+ frbøCreak
drand
usingexpressions
(4)for
E gives
A more-direct and more-accurate estimate of the en-
ergy dissipationœ/L in the main channelcomesfrom
using the storea.....
channelfor the energybudget[Borovsky,
"2/8•r aroundthe
electrosfarm energy e-
1995]. The
œ•tøred
L = 1,k2q-/k2
log(
2 9o•ut
q) , (5)
q 9obr
eak
storedenergyper unit lengthœstorea/L on the channel where a cutoff radius rcut has been introduced into the
is calculated as follows. For a channel with a charge secondintegralto preventthe integralfrom logarithmi-
per unit lengthof &qon it, the electricfielda distance callydivergingasr -• oo. (Notein expression (5) and
r awayfromthe centerof the channelis E - 2&q/r,as elsewhere,log - loge') Two physicalchoicesfor select-
obtainedby applyingGauss'law 57.E - 4•rnqin cylin- ing limits for rcutcometo mind. The first (whichis the
drical geometry(wherenq is the volumechargeden- largestrcutvalue)is to choose rcutequalto the lengthof
sity). Accordingto this expression,if all of the charge the main channel;at this radius the leader has ceased
resides on the channel itself, the electric field E will to havethe propertiesof an infinity long cylinder.The
have a magnitudethat is abovethe air-breakdownlimit second(whichis the smallestrcutvalue)is to choose
Ebrea
k at small r outside the channel, where Ebrea
k• rcutto be the radius at which the electric field of the
2 x 106V/m at sealevel[Cobinc,1941]. (If processeschannelequalsthe backgroundelectric field. Because
such as streamer formation comeinto play around light- routappearsin a logarithm,ϥtorea/L
will only vary by
ning channels,then the valueof Ebrea
k may be substan- about a factor of 2 in goingfrom onechoiceof rcutto the
tially reduced
from2 x 10• V/m [Zeleny,1907](seealso other. The secondchoice(the radiusat whichthe field
WagnerandHileman[1961],Figure1.) However,wher- equalsthe backgroundfield) will be usedhere. Taking
ever E exceedsEbreak,the electrical conductivity of the
the backgroundelectricfieldto be the electricfieldun-
air rapidly risesand free chargethen movessothat E is der the thundercloud Ecloua,the radius at which E -
rapidly reducedto E <• Ebreak'This effectivelymoves Ec,oua is givenby expression(3b) to be
some of the charge off the highly conductingchannel 2
and into the weakly conducting air surroundingthe
channel(the coronasheath). So, arounda lightning
9ocut
= /•qEciou
d ' (6)
channel, the electric-field strength will be E • Ebreak (Typicalvaluesof rcu•will be severaltensof meters.)
out to a radius of
With the useof expressions(2) and (6), expression
(5)
2 becomes
9obreak
----AqEbreak (2)
Beyond 9obrok the field will fall off approximatelyas
Estøred
Z __--
A2
q [•1 + log(Ebreak
/ 1 ß (7)
Ecloud
l/r, appropriateto a uniformlycharged,infinitelylong As an aside,note that the 1/2 in expression (7) corre-
cylinder. Thus, the radial dependenceof the electric spondsto the electrostaticenergyinsider - rbreak and
field around a charged lightning channel is approxi- the logarithm correspondsto the electrostaticenergy
mately outside r - rbreak;if the air is sufficiently conducting
2
within r - rbreak,then all of the charge on the chan-
1 05 ß
ß ß
ergy dissipated œ in the channel by a return stroke is ' ' ' '''1 ' ' ' ' ' '''1 ' ' ' ' ' '''1 ' [ i ,
œstorea,
SOusingexpression
(7), the energydissipatedper Dart-Leader
Charge-per-Unit-Length
Estimates
unit length is
C•o
w/ Lleader
1 Ebreak
E
L
Estored
L
A2
"
œ/L in the range6 x 102- 2 x 106J/m. Thisrangeof the potential differenceA•btotcausedby the subtrac-
valuesextendshigherthandoesthe2x 102- 2x 104J/m tion of Qlowfrom Qtot is accountedfor in the calcu-
rangeof œ/L valuesin Figure 3, whichwasgenerated lation of A•b: this producesthe downwardbendingof
from expression(8). If the chargeper unit length Aq the bottom curve.) As can be seenin the figure, for
on the channelis estimatedas Xq = (•low/Lmain, where the reasonablevalue Qlow- 2 C, the ratio varies from
Lmainis the length of the main channel,then the (old) 8 to 58, dependingon the value of Qtot, so the (new)
QlowACtot methodof estimatingœ/L is comparedwith stored-electrostatic-energy
estimate resultsin a consid-
the (new) stored-electrostatic-energy
methodby taking erably smallervaluefor the energydissipationœ/L in
the ratio of expression(1) to expression
(8); this gives the main lightningchannelduring a singlestrokethan
did the traditionalQlowAtPtot/Lm•in
method.This factor
-2
of 8 to 58 may be indicativeof the multiplicity of the
(œ/Z)old
•. Qlow
(œ/L)new A•btot
Lmain (Lmain
Qlow
) channelsinside and below the thundercloud;however,
as pointedout by a referee,the derivationof expression
(10) is very approximate.
1 (k Ebreak
Ecloud --
1'
In Appendix C it is shownthat A•btotm 2•rErl•yor(ex- 4. Energy Conservation and the Radius
pression(C2b)), where .Y,is the areal chargedensity of a Lightning Channel
of the negative-chargelayer in the bottom of a thun-
If the energyœthat is depositedinto a lightningchan-
dercloud and where rl•yoris the horizontal radius of nel by a return strokeis depositedrapidly sothat heat-
this chargelayer. Using this expression,using Qtot m ing, moleculardissociation,and ionizationoccur prior
•rr12ayorY,
(where Qtotisthetotalamount ofcharge inthe to the radial expansionof the channel, then the de-
chargelayer in the cloud), and usingthe valuesEbreak positedenergycan be brokeninto three parts:
= 2 X 106V/m andEclo,d
-- 5 X 104V/m, expression
(9) simplifiesto œ = œdisso
-[- œth
.... 1 -Jr-œ,oniz , (11)
(œ/L)old I Qtot Lmain where
. (10)
(E/L) new 2 QlowT'layer 3
œth
.... l 2 Z (1 -[- ,•) •atomic
•B Tinit (12)
For twovaluesof Qtot,expression(10) is usedto plotthe
ratio of the old-methodœ/L value to the new-method is the initial (after energydepositionbut beforethe ra-
œ/L value in Figure 5, where the lengthLm•inof the dial expansion)thermal energyof the channel,
main channel is taken to be 5 km and the radius of
the cloud'schargelayer Tl•yor
is taken to be 3 km (see œdisso
= •r r 2 L nmole
c ([disso
init (13)
AppendixC). (Note in the figurethat the reductionof
is the energythat goesinto moleculardissociation,and
lOOO ' " ' ' ' ",,1 , , , , , ' "'1 '
œioniz
= • r 2 L natomit
init -• ([ioniz (14)
• Lmain
= 5km
lOO • rlayer
=3km = C
is the energy that goesinto ionization. In expressions
(12) - (14), rinitis the initial radiusof the channel,L is
the lengthof the channel,ks = 1.38x 10-16 erg/øK
is Boltzmann's constant, T•nit is the initial tempera-
lO
ture of the channel(after heating but beforeexpan-
sion),• is the fractionalionizationof the atomsin the
1
channel, natomi•is the number density of atoms in the
unexpandedchannel, nmolec is the number density of
moleculesin the unexpanded channel, ([disco is the en-
ergyrequiredto dissociate onemoleculeof air, and
O. 1 10
is the energyrequiredto ionizeone atom of dissociated
QIow [C] air. In expression(12), the I is the contributionof
Figure 5. The two methods(old and new) for calculating atoms and ions and the • is the contribution of free
per unit lengthœ/L ona channelare electrons. As can be seenin Figure 6, the N2 molecules
the energydissipated
compared by takingthe ratio of the two œ/L values.The of air are fully dissociatedfor temperatures>• 15,000
old valueis obtainedfrom expression(1), and the new value øK, which are the temperaturesof interestfor the light-
is obtained from expression(8). The ratio is plotted as a
functionof the amountof chargeQlowloweredby the light- ning channels,so nmolec = 0.5natomic will be taken. In
ning stroke, for two values of the total amount of charge accounting for the depositedenergy in expression (11),
Qtot in the thundercloud. the energylost throughoptical radiation and through
BOROVSKY' LIGHTNING ENERGETICS 11,543
10
Nitrogen(sea level) .T' = 1.27 x 10-6 eT/13øOøK
• 4.3X 1019atoms/cm
3
o N+ for 10,000øK __<T __<13,000øK
Z I ß
• - 3.09 x 10-4 eT/2880øK
o
.--'--
z
0.1 for 13,000øK _< T _< 20,000øK
o
• O.Ol ß • = 1.16
log15,200
oK
for 20,000øK _• T _• 29,000øK
o.ool
•oooo •oooo •oooo •oooo •oooo
• - 0.839
log12,000øK
T [OK]
for 29,000øK _• T _• 40,000øK
Figure 6. •he composi[io•
•c[•o• of the •[•o•e•
log(19,900
• = 1.48 oK)
for 40,000øK _• T _• 50,000øK. (16)
RF emissionis ignored; this is justified as follows. Ac- (Note that • goesaboveI for temperatures• 39,000
cording
to Figure3, forAq= I x 10-4 - 5 x 10-4 C/m, øK; this is becausesecondionization becomesimpor-
the energydepositedper unit length in a singlestroke tant, as can be seenfrom Figure 6.) For the energy
is in the rangeof 102- 104J/m. The energylostvia of dissociationof air, the energyfor dissociatingan N•.
optical radiation per singlestroke is estimated from the moleculewill be taken: eais•o
= 9.8 eV = !.57 x 10-11 erg
measurements of Barasch[1970]and Guo and Krider [e.g.,Rees,1989]. For the energyof ionization,the en-
[1982]to be in the rangeof 8 - 400 J/m, whichis a ergy for ionizing atomic nitrogen will be taken' eio.,•.=
small fraction of the depositedenergy (as was previ- 14.5eV = 2.33x 10-11erg[e.g.,Condon, 1967].Taking
ouslyconcluded by Krider et al. [1968]and by Conner the initial temperature of the channelto be the peak
[1968]). The energylost from the channelby means spectroscopictemperature of lightning, which can be
of RF emission is estimated from the measurements of
25,000- 35,000øK [Prueitt, 1963; Uman,1964;Orville,
Krider and Guo [1983]to be-•10 J/m, assumingL - 1977, 1968;Salanave,1980],T•.,t = 30,000øK = 2.6 eV
5 km. Hence, neglect of the optical and RF emission is used.For this valueof T•.•, expression (16) yieldsthe
is justified when consideringthe energeticsof lightning fractional ionization of the fully dissociatedair to be •r
channels.
= 0.77. Comparingthe three terms in expression(11)
Insertingexpressions
(12)- (14) into expression (11), for theseparameters(e.g.,for T - 30,000øK) onefinds
thenusingexpression(11) in expression(8), takingnatomitthat 30% of the depositedenergygoesinto thermal en-
-- 2nmolec
for fully dissociatedair, and solving for ri.it ergy, 21% goesinto dissociationof air moleculesin the
yields channel,and 49% goesinto ionizationof the channelair.
TakingAq- 4 x 10-4 C/m - 1.2x 104statC/cm(cgs
Winit
__ )•q [•__
_•_
log
(Ebreal•
1/2
Ecloua/]
(7rn•tømic)-l/2 units)andn•omi•-- 5.0 x 1019cm-3 (appropriate
fora
x [(1•1_.•')•kBTmit
3 •1_•(•disso
1 •1_•'•ioniz
1-1/2
(15) 10 .... I ....
Nitrogen(sea level)
4.3 X 1019 atoms/cm3
I .... I .... I .... I .... I ' ' ' ' I ....
ß
....,.....""'1111111-
..,-" ..,-
=5xl
018
cm-3 natomic
= 5x1019
cm-3
1
• • • natomic
kB L(rfin•,-
Tatmos 2 T2 init)(18a)
sity. The radial component of the Poynting vector $ electrical conductivity achof a channelwith a temper-
- (c/4•) E x B is $r = (c/4•)EzBo, wherethe z di- ature Tch • 15,000øK can be written (in cgsunits)
rection is along the channel axis and the 0 direction is
1.2
azimuthal around the channel axis. Since Er >> Ez for
lightning [e.g., Borovsky,1995, Figure 8], the denomi- eYch
-- 7.6X1088-1
(o--•) ; (31)
natorof expression
(24) canbewrittenasE•/8•. Thus
expression(24) becomes usingthis and usingexpression
(22) for r .... gy,expres-
sion (30) becomes
E.B•
v....gy= 2cE? ' (25)
The radial electricfield Er (outsidethe coronaradius
Theat
= Vreturn
2 7r
2 (r-•n
rchlog
2•q e-1/4
(EcloudEbreak)l/2
expression to =
azimuthal magnetic induction Bo outsidethe channelis
given by Ampere's law to be
x 7.6x10ss-1 (•K)
1.2
. (32)
If the risetime Theftis lessthan the channel-expansion
21
= _ _ , time Texp•n,then rch -- rinit should be used in expres-
sion (32); if the risetimeis greaterthan the channel-
where I is the current flowing in the lightning channel. expansion time, then rch - r•in•l should be used in ex-
Usingtheseexpressions for Er and Bo in expression(25) pression(32). Computersimulationsof channelevolu-
and then usingexpressions (25) and (22) in expression tion indicate that channel expansionbegins a few mi-
(23) yieldsthe energy-flowtime croseconds after heating starts [Plooster,1971; Hill,
1971;Paxtonet al., 1986],soTexp•n is typicallya fewmi-
croseconds.In the two panelsof Figure 11 the channel-
Theat rcn2 I dr . (27)
• d 7'energy AqIE.r heating risetime Theatfor a return stroke is plotted as a
function of the linear chargedensity Aqon the channel
prior to the return stroke. The lower limit to Theftis
Accordingto expression (27), the largerE, is,the shorter appropriate for a channelthat has not begun its radial
the risetime Theatwill be. Since E, near a channel is the
expansion(Theft((Texp•n)and the upperlimit to Theftis
sameas Ez within the channel[Borovsky, 1995,Figure appropriatefor a channelthat is fully expanded(Theat
8], the value of E, can be estimatedby usingOhm's >> Texp•n); hencethe top panelusesexpression (15) for
law j,. =achE,, where j,. is the current density flowing
rchand the bottom panel usesexpression(20) for rch.
in the channel and ach is the electrical conductivity of
Also,the top panelusesn•tom•c- 5 x 10TMcm-3, whichis
the channel.Usingj,. = I/•rrc•h,whererchis the radius appropriate for a stepped-leader-loadedchannel in cool
of the conductingchannel, Ohm's law yields
air, andthe bottompaneluses7t•tomic ----5 X 10ls cm-3,
I 1 which is appropriate for a dart-leader-loaded channelin
= .
E. •rr•2h
ach rarefiedair. Using the top panel of Figure 11 with X,
= 4 x 10-4 C/m yieldsa rangeof heatingrisetimes of
2 x 10-6 - 3 x 10-4 s for a firstreturnstroke.Likewise,
In the electrodynamicdescriptionof lightning,the cur-
usingthe bottompanelwith X, = I x 10-4 C/m yields
rent I flowingin the channelis relatedto the chargeper
where a rangeof heatingrisetimesof I x 10-c - 2 x 10-4 sec
unit length Xq on the channelby I = )•qVcharge,
Vcharg
e is the velocity at which chargemovesalong the for a subsequentreturn stroke. As will be discussedin
channel. Borovsky[1995]showedV•ha•eto be equalto section 6, the observedvalues of the heating risetime
the real part of the complexphasevelocityof the elec- fall into the lowerend (unexpanded-channelend)of the
tric wave moving along the channel: this quantity is range of heating-risetimevalues predicted by the elec-
approximately equal to the velocity of the return stroke trodynamic model.
Vret•rn.Thus I = XqVret•rn
for a return stroke. Using ex-
pression(28) in expression (27) and usingI = XqVret•n 6. Summary and Comparison With the
gives Literature
Theat= rch7rV2rch
• 7.energy
2Crch
• dr
return
, (29) In section 2 of this report the flow of energy from
a thundercloud to a lightning channel was discussed.
which is integrated to yield The amount of energy that is depositedalong a light-
ning channel was found to be related directly to the
•r •'ch O'ch •'energy linear chargedensityAqthat is loadedonto the channel
Theat• 2
V return
log\ rch
. (30) by a stepped or dart leader. By estimating the linear
chargedensity A• on lightning channels(in Appendix
Accordingto expression
(Ald) of Borovsky[1995],the A), the energydissipatedper unit lengthœ/L in light-
BOROVSKY: LIGHTNING ENERGETICS 11,547
10-2
First ReturnS{i0kes I oooO'• monlytakenin lightningreviews[e.g.,$chonland,1956;
Berger,1977; Uman and Krider, 1982; Uman,1987].
10-3 natorni
c=5X10
TM
cm-3 ......-' Valuesfor the energydissipationper unit lengthœ/L
10'4
upper
limit
torisetime
-_ / .. of 2 x 102- 1 x 104J/m wereobtained
forreturnstrokes
10-s
-(channel
fully
expanded,••
• ..""
....-" .
ß
ß
ß
o
.
.
o
in section 3 of this report. These values agree with
earlier theoretical calculations that were obtained with
computer simulations of the Ohmic heating of chan-
10-•
nels based on observed temporal profiles of the chan-
o
ß
ß
ß
ß
o
ß
10'7
,ß
...
øl'•
.
ß
...-"
I
ß
r i , i I
..o
i i i i
r limit to risetime
(channelnotexpanded)
i i i JI i i
(taken from an analogybetweenlightningand labora-
tory arcs), and the lowerœ/L valuesof this report im-
ply that the thunder-production efficienciesof lightning
10-s 10'4 10'• may be higher than the 0.18% estimated by Holmes et
Zq [C/m] al. [1971](whichwereobtainedby using2.3x 105J/m
Figure 11. The return-strokeheatingrisetime•'heat
is plot- for E/L).
ted asa functionof the linearchargedensityAqthat is on a The cloud-to-groundelectrostatic-potentialdifference
channel prior to the return stroke. Specific values used for A½•o•was calculated in Appendix C for use in compar-
the figureareTch-- 30,000øK, Vr.turn
-- 1.5 X 1010cm/s, ing the œ/L valuescalculatedin this report with ElL
Zbr.ak= 2 X 106V/m = 66.7statV/cm,Zc•oud = 5 X 104 valuesestimated by meansof a traditional method. Val-
V/m = 1.7statV/cm,œioni•. = 14.5eV -- 2.3 X 10-11 erg, uesof 3 x 10• - 9 x l0 s V wereobtainedfor A½•o•.These
œdisso= 9.8eV = 1.6x 10-11 erg,and• = 0.77.Theupper values are in agreement with typical theoretical values
limit (obtainedfrom expressions
(33) and (15)) is appropri- in the literature [e.g.,Bruce,1944;Malan, 1963; Tzur
ate for a channel that has fully expanded and the lower limit
and Roble,1985].
(obtainedfrom expressions
(32) and (20)) is appropriatefor
a channel that has not begun its expansion. Initial channel radii ?'initof 0.08- 1.5 cm and final
channel radii rfinalof 1.2 - 20 cm were obtained in section
4 of this report, with ?'init• 0.2 - 0.4 cm and Trina
l ----3 -
ning channelswascalculated(in section3). From these 6 cm being the most likely values. The values obtained
valuesof œ/L the radii of lightningchannelswerecalcu- for ?'initare in agreement with computer simulations of
lated throughthe useof energyconservation (in section channel evolution, which obtain initial radii of 0.002-
4). Both the initial radii ?'init
prior to channelexpansion 0.5 cm [Plooster,1971;Hill, 1971;Paxtonet al., 1986].
and the final radii Vfina
I after channel expansionwere cal- Burn marks indicate channel radii in the ranges of 0.2
culated. Finally, the channel-heatingrisetimes rh.at for - 0.5 cm and 2 - 3.5 cm [Urnan, 1964; Taylor, 1965;
return strokes were calculated from estimates of where Jones,1968], which agreewith the valuesof Viaitand
energy is stored along lightning channelsand estimates •'finalobtained in section 4. The •'finalvalues obtained in
of the radial energy-flowvelocitiestoward the channels, section4 agreewith high-resolutionphotographs,which
with both of these quantities depending on the linear indicatechannelradii of 3 - 12 cm [Evansand Walker,
chargedensity Xq residingalong the channelprior to 1963; Orville et al., 1974;Salanave,1980;Idone,1992].
the returnstroke. Heating-risetime values rh.at of 1 - 200 /•S for return
The values obtained for the linear charge densities strokes were obtained in section 5 of this report. The
loadedonto lightningchannels are Aq• 1 X 10--4 - spectroscopic measurements of Orville [1968]indicate
5 x 10-4 C/m (seeFigure4), withthe upperendofthis that the heating times for return strokes are of the or-
rangepertainingto stepped-leader-fed channelsand the der of 5 /•s, and photometric risetimes, which should
lower end pertaining to dart-leader-fedchannels.These be equivalent to heating risetimes, are measuredto be
Xq values agree with values in much of the literature in the range of 1 - 60 /•s for return strokes[Krider
(see referencesin Appendix A) but are considerably et al., 1968; Ganesh et al., 1984; Jordan and Uman,
lowerthanthe Aqvalueof 1 x 10-3 C/m (ormore)com- 1983;Mach and Rust, 1989;Baker et al., 1990].Hence
11,548 BOROVSKY' LIGHTNING ENERGETICS
the heating-risetime values predicted by the electro- to ground by a first return stroke, since these return
dynamic model span the range of observed channel- strokesdrain steppedleaders[e.g., Bruce and Golde,
heating times, which are 1 - 50/•s, and they also span 1941; Wagner and McCann, 1942; Price and Pierce,
the range of energy-depositiontimes predicted by com- 1977; Linet al., 1980]. Electric-field-change
measure-
puter simulations of channel evolution, which are 5 - ments and time integrals of measured currents yield
10 /•s [Plooster,1971; Hill, 1971; Paxton et al., 1986]. typical charge-transfer values for first return strokes
Note that computer simulations predict that the heat- of 2- 20 C [Pierce,1955; Brook et al., 1962; Berger,
ing risetime rheatcan be much longer than the current 1967; Golde,1977;Krehbielet al., 1979;Proctor,1981].
risetime in the channel. Judging from the heights of the charge regions that
The electrodynamic picture of lightning has provided are drained by first return strokes,the lengths of first-
a straightforward and natural method for calculating return-strokechannelsare about 5 - 10 km [e.g.,Kre-
the energy dissipatedper unit length in a channelby a hbielet al., 1979]. Because the stepped-leader-produced
return stroke and has yielded estimatesof channelradii channelis highly branchedwith blind channels(seeFig-
and channel-heatingrisetimes. As discussedin this sec- ure 2), theselengthvalueswill be multipliedby 5 to pro-
tion, the values calculated seem reasonable. Therefore vide an estimate of the total length of stepped-leader-
these tests of the electrodynamic description are suc- produced channels. The factor of 5 is arrived at in Ap-
cessful. pendix B. Multiplying the abovementioned5- 10 km
Sincethe energydissipatedper unit lengthœ/L in a channel-length estimates by 5 yields Lleader-- 25 - 50
lightning channelis intimately related to the chargeper km. Taking (•leade•-- 2- 20 C and Lleade•-- 25- 50 km
unit length Xq loaded onto the channel,in the future it yieldsA, • 4 x 10-5 - 8 x 10-4 C/m. Thisrangeof
will be very interestingto compareanticipatedXqheight valuesis indicated in the top panel of Figure 4.
profiles with observed return-stroke-luminosity height The secondand third methodsfor estimatingAqcome
profiles. For instance, it is anticipated that stepped- from assumingthat the current in a leader is causedby a
leader-loaded channels will have larger linear charge simple spatial translation of chargedensity as the leader
densitiesnear the ground [e.g., Little, 1978, Figures4 expands(sort of a slidingstick of charge);in this case,
and5]; accordingto the electrodynamicmodelthe chan- Aq- I, eade•/Vleade•,
whereIleaaer is the currentcarriedby
nels shouldtherefore be brightest nearestthe ground. A
the leader and Vleade•is the velocity of the leader tip
decreasein luminosity with height has been reported for
along the channel. The stepped leader propagatesin-
return strokes[Schonland
et al., 1935;$chonland,1956; termittently and the current is impulsive with each step
Mach andRust,1989]. Similarly,it will beveryinterest- [e.g.,Schonlandet al., 1938;Kitagawaand Kobayashi,
ing to apply the œ/L predictionsof the electrodynamic 1958;Beasleyet al., 1983];accordingly,
to obtainAq-
modelto the bidirectional-leader model[Kasemir,1960; Ileade•/Vleade•,
time-averagedvaluesof Ileade•and Vleader
Kawasakeand Mazur, 1992;Mazur andRuhnke,1993], can be used or instantaneous impulse values of Ileade•
which predicts a null in the linear chargedensity along and Vleade•can be used, which will result in two esti-
the lightning channel. mates of Aq. Time-averaged values of stepped-leader
currents have been inferred from electric-field-change
Appendix A: Charge per Unit Length on measurementstaken far from lightning channelsto be
Channels {Iaeader
) -- 345- 2300 A [Thomsonet al., 1985];these
measurementspick up the vertical currents in the blind
In this appendixthe linear chargedensities•q loaded branch channels in addition to the current in the main
onto lightning channelsby stepped leadersand by dart channel,so realisticvaluesfor the main channelsmay be
leaders are estimated. For each type of leader, several •1/5 of thesevalues,giving{Ileade•)
-- 70 - 460 A. This
different methods are used to obtain the estimates. The
range of values agrees with values of leader currents
Xq values from the various methods are collectedinto near the ground that can be derived from the measure-
Figure 4. mentsof Krider et al. [1977],whereinit wasfoundthat
singlestepsdeliver1 - 4 x 10-a C of chargeand that
Channels Loaded by Stepped Leaders
single steps are separated by 16 /•s on average: these
The linear charge density •q on a channel that is valuesgive (/leader)
-- AQ/At - 62- 250 A. Typicalav-
charged by a stepped leader is estimated by five dif- eragevaluesfor the stepped-leader
velocityare (Vleade•)
ferent methods, and the rangesof Xq values obtained -- 1 X 107- 3.6 X 10s cm/s [Schonland,
1953;Orville
from these methods are plotted in the top panel of and Idone, 1982; Beasleyet al., 1983]. Thesecombine
Figure 4. The first is the method used by $chonland to yield Aq-- < /leader
> / < Vleader
> ----2.0 X 10--5
whereQleader- 4.6 x 10-4 C/m. Typicalimpulse-current
[1953],whichtakesAqto be Qleadev/Lleader, valuesfor
is the total amount of charge lowered by the stepped steppedleadersare /leader
-- 2- 8 kA [Krider et al.,
leader and Lleaderis the cloud-to-groundlength of the 1977], where the current corresponds to a singlestep
channel made by the leader. The charge on a stepped in a single channel, and typical impulse-velocityvalues
leader Qlowshould be equal to the charge transferred areVleader-- 2 X 109- 7 X 109cm/s [Krideret al., 1977;
BOROVSKY: LIGHTNING ENERGETICS 11,$49
tions of thunderclouds accumulatesinto horizontal lay- muchlessthan the value4.6 x 109V estimatedby Meek
ers. The height of sucha layer will be denotedhere as [1939].
h. Near a fiat layer of chargethat has an areal charge
densityE (= chargeper unit area), the electric-field
strength E is Acknowledgments. The author wishesto thank Charlie
Rhodes for useful conversations. This work was supported
= by the U.S. Department of Energy.
[e.g.,Jackson,1975,expression
1.22]. (Note that the
symbol E is chosenhere to representthe areal charge References
density to avoid confusionwith the electrical conduc-
Baker, L., R. L. Gardner, A. H. Paxton, C. E. Baum, and W.
tivity or.) The electricfield pointsupward belowthe Rison, Simultaneous measurement of current, electromag-
layerand pointsdownwardabovethe layer;withinthe netic fields, and optical emission from a lightning stroke,
layer, the electric field goesto zero. In situ measure- in Lightning Electromagnetics,editedby R. L. Gardner,
ments of the vertical-electric-fieldstructure yield values pp. 365-374, Hemisphere,New York, 1990.
of E of 0.2- 5 C/km2 - 0.06- 1.5statC/cm
2 [Winnet Barasch, G. E., Spectral intensities emitted by lightning dis-
charges,J. Geophys. Res., 75, 1049, 1970.
al., 1978, 1981; Byrne et al., 1983; Marshall and Rust, Beasley,W. H., M. A. Uman, D. M. Jordan,and C. Ganesh,
1991; Marshall and Lin, 1992]. These E valuescorre- Simultaneous pulses in light and electric field from stepped
spondto verticalelectricfieldsof 11 - 280 kV/m, which leadersnear groundlevel, J. Geophys. Res., 88, 8617,
are comparableto or strongerthan the tens-of-kV/m 1983.
Proctor, D. E., Lightning and precipitation in a small multi-derstormwith its global electricalenvironment,J. Geo-
cellularthunderstorm,J. Geophys. Res., 88, 5421, 1983. phys. Res., 90, 5989, 1985.
Prueitt, M. L., The excitation temperature of lightning, J. Uman, M. A., The diameterof lightning,J. Geophys. Res.,
Geophys. Res., 68, 803, 1963. 69, 583, 1964.
Rees,M. H., Physics and Chemistry of the Upper Atmo- Uman, M. A., Lightning, sec. 7.3.3, Dover,New York, 1984.
sphere, table 2.2.1, CambridgeUniv. Press, New York, Uman, M. A., The Lightning Discharge, secs. 5.4 and
1989. A.1.6, Academic, San Diego, Calif., 1987.
Rhodes, C. T., X. M. Shao, P. R. Krehbiel, R. J. Thomas, Uman, M. A., and E. P. Krider, A review of natural light-
and C. O. Hayenga, Observations of lightning phenomena ning: Experimental data and modeling, IEEE Trans.
usingradio interferometry,•. (•½ophys.•½s., gg, 13,059, Electromagn. Cornpat., 2•, 79, 1982.
1994. Wagner, C. F., and A. R. Hileman, The lightning stroke, II,
Rustan, P. L., M. A. Uman, D. G. Childers, and W. H. Power Appar. $yst., 53, 622, 1961.
Beasley, Lightning source locations from VHF radiation Wagner, C. F., and G. D. McCann, Induced voltages on
data for a flash at KennedySpaceCenter, J. Geophy3. transmission
lines, Trans. Am. Inst. Electr. Eng., 61,
Res., 85, 4893, 1980. 916, 1942.
Salanave,L. E., Lightning and Its Spectrum,chaps.4 and Waldteufel, P., P. Metzger, J.-L. Boulay, P. Laroche, and
6, Univ. of Ariz. Press, Tucson, 1980. P. Hubert, Triggered lightning strokes originating in clear
Schonland,B. F. J., Progressivelightning IV- The discharge air, J. Geophys. Res., 85, 2861, 1980.
mechanism,Proc. R. $oc. London A, 16•, 132, 1938. Williams, E. R., C. M. Cooke, and K. A. Wright, Electrical
Schonland, B. F. J., The pilot streamer in lightning and the discharge propagation in and around space charge clouds,
longspark, Proc. R. $oc. London A, 220, 25, 1953. J. Geophys. Res., 90, 6059, 1985.
Schonland,B. F. J., The lightningdischarge,Encycl. Phys., Wilson, C. T. R., Somethundercloudproblems,J. Franklin
22, 576, (sects.20, 22, and 55), 1956. Inst., 208, 1, 1929.
Schonland,B. F. J., D. J. Malan, and H. Collens, Progressive Winn, W. P., and L. G. Byerley, Electric field growth in
lightning,II, Proc. R. $oc. London A, 152, 595, 1935. thunderclouds,Q. J. R. Meteorol. $oc., 101, 979, 1975.
Schonland, B. F. J., D. B. Hodges, and H. Collens, Pro- Winn, W. P., G. W. Schwede, and C. B. Moore, Measure-
gressiveLightning V. A comparisonof photographic and ments of electric fields in thunderclouds,J. Geophys.
electricalstudiesof the dischargeprocess,Proc. R. Soc. Res., 79, 1761, 1974.
London A, 166, 56, 1938. Winn, W. P., C. P. Moore, C. R. Holmes, and L. G. By-
Soula, S., and S. Chauzy, Multilevel measurement of the erley III, Thunderstorm on July 16, 1975, over Langmuir
electric field underneath a thundercloud, 2, Dynamical
Laboratory: A casestudy, J. Geophys. Res., 83, 3079,
evolutionof a ground spacechargelayer, J. Geophys. 1978.
Res., 96, 22,327, 1991. Winn, W. P., C. B. Moore, and C. R. Holmes, Electric field
Taylor, A. R., Diameter of lightning as indicated by tree structure in an active part of a small, isolated thunder-
scars,J. Geophys. Res., 70, 5693, 1965. cloud, J. Geophys. Res., 86, 1187, 1981.
Thomson, E. M., M. A. Uman, and W. H. Beasley, Speed Zeleny, J., The discharge of electricity from pointed conduc-
and current for lightning stepped leaders near ground as tors differingin size, Phys. Rev., 25, 305, 1907.
determinedfrom electricfield records,J. Geophys. Res.,
90, 8136, 1985.
Thottappillil, R., D. K. McLain, M. A. Uman, and G. Di-
endorfer, Extension of the Diendorfer-Uman lightning re- J. E. Borovsky, Mail Stop D466, Los Alamos National Labo-
turn stroke model to the case of a variable upward return ratory, LosAlamos, NM 87545. (eraall: jborovsky@lanl.gov)
stroke speed and a variable downward discharge current
speed,J. Geophys. Res., 96, 17,143, 1991. (ReceivedOctober 15, 1997;
Tzur, I., and R. G. Roble, The interaction of a dipolaf thun- acceptedOctober30, 1997)