Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

'OURNAL

OFGEOPHYSICAL
RESEARCH,
SPACEPHYSICS VOL. 73, No. 23, D]C]B]R 1, 1968

Radial Diffusion Coefficient for Electrons at 1.76 L 5

L. L. NEWKIRK AND /. WALT

LockheedPalo Alto ResearchLaboratory, Palo Alto, CaliJornia94304

Radial diffusionby nonconservation


of the third adiabaticinvariantof particlemotionis
assumedin analyzingexperimentsin which electronsappearedto move acrossfield lines.Time-
dependentsolutionsof the Fokker-Planckdiffusionequationare obtainednumericallyand
fitted to the experimental resultsby adjusting the diffusioncoefcient. Values deducedfor the
diffusioncoefficientvary from 1.3 X 10- R/day at L _-- 1.76 to 0.10 Rday at L --_ 5. In
the interval 2.6 L 5, the coefi%ientvaries as L TM. Assuminga constantelectronsource
of arbitrary magnitudeat L -- 6 and the abovediffusioncoefi%ients, the equatorialequilib-
rium distribution is calculated for electronswith energiesabove 1.6 Mev. The calculation
yields an outer belt of electronswhoseradial distribution is in good agreementwith the
observedbelt. The calculateddistributionalso exhibitsan inner belt at L 1.5. However,
the calculatedintensity of the inner belt relative to the outer belt is severalordersof magni-
tude smaller than the experimental ratio.

INTRODUCTION this type of diffusion were made by several


Observationsindicate that trapped particles authors[Kellog,1959; Parker, 1960; Davis and
in the magnetospheresometimes appear to Chang,1962; Tverskoy,1964; Dungey,1965;
move, or diffuse, radially across field lines Fiilthammar, 1965, 1966; Ershkovich et al.,
[Frank et al., 1964; Mcllwain, 1965; Frank, 1967]. Other analyticstudiesof specificparticle
1965; Craven, 1966; Brown, 1966; ernov data [Nakada and Mead, 1965; Nakada et al.,
et al., 1966; 1967]. However, becauseof the 1965; Paolini et al., 1967; Newkirk and Walt,
varied and complexnature of the dynamicmag- 1968] indicated that these data were consistent
netosphere,no unique mechanismof particle with particle drift under variation of the third
displacement canbe expectedto satisfyall cases. invariant. These latter studies,therefore,im-
Kavanagh [1968a], for example,has attributed plied the existenceof the proposed diffusion
the apparent inward drift of particles detected process,although the results were not precise
by Explorer 12 duringmagneticstorm-timecon- enough to preclude changesof the second,or
ditions to the gross movement of field lines integral, invariant as well.
themselves, rather than particlemotionthrough The present paper compares the observed
the fieldlines.Cladis [1966], on the otherhand, radial motion of trapped electron distributions
has explainedthe suddenappearanceof mono- with that expectedfrom diffusionby nonconser-
energeticelectronsat L = 1.14 [Imho] and vation of the third invariant.Two specificmeas-
Smith, 1965] on the basis that the electrons urements are considered.In one case, Brown
were conveeredinward from higher field lines [1966] observed a narrow band of electrons
by the influenceof electricand magneticfluctua- produced by the Russian nuclear detonation of
tions in resonancefor a few cycleswith the azi- November 1, 1962, to broaden with time in a
muthal drift period of the electrons. manner suggestive of diffusion. In the other
Most investigationshave been based on the case, Fraik et al. [1964] observed the time
assumptionthat particle drift occurredthrough evolution of an anomalous radial distribution
nonconservation of the third, or flux, invariant that was introducedin the outer belt by a mag-
causedby field disturbances(generallyunspeci- netic storm. The distribution appeared to drift
fied) acting on a time scale comparableto the radially inward while maintaining its general
azimuthal drift period of the particle. The ac- shape. Finally, the radial diffusion coefficients
cumulativeeffect of many suchdisturbancesoc- deduced from these comparisonsare employed
curring at random will be to causeparticlesto to evaluate the role of radial diffusion as a
diffuse across field lines. Thoretical studies of source mechanism for the electron belts.

7231
7232 NEWKIRK AND WALT
In this work, attention is limited to trapped as a function of r are described in the next sec-
electronsmirroring in the equatorialplane. The tion and must preservethe constancyof/ and J.
one-dimensional,Fokker-Planck equationin the The time-dependentdiffusionequation 1 was
form given by Fiilthammar [1966] is used to integrated by explicit, finite difference tech-
describe the radial diffusion of the electrons in niques [Forsythe and Wasow,1960]. Given (a)
a dipolefield. an initial distributionfunctionn(r, t, J ----O,
t = 0) determinedfrom experimentaldata, (b)

On
Ot O\2r
(D_
-- Or O(r'n)
rr ) _n__
r (1)fied valuesfor Ds(r) and r(r),
appropriateboundaryconditions, and (c) speci-
the distribution
functionn(r,.t , J = 0, t) at later timescan be
In the equation,n(r, t, J = O,t) is the particle
calculated.If Ds and r are functionsonly of r
distribution function at time t, r is the radius
and if the initial energy distributionat all r is
in the plane of the geomagneticequator, t is
cons}stentwith the transport of electronsacross
the magneticmoment, and J is the integral in-
r at constantt and J, then the energydistribu-
variant, set equal to zero for equatorialparticles.
tion at each r will not changewith time. Under
Ds is the diffusion coefficient, defined as the
these assumptionsthe diffusionequation needs
mean square radial displacementper unit time,
to be solved only once at each r for an arbi-
and is assumedto be a function of r only. In the
region of space under considerationthe loss
trarily chosenvalue of t; solutionsfor other t
values can be obtainedfrom the known energy
processeshave not been identified. However,
dependenceat each r. For comparisonwith ex-
sinceexperimentsindicate that the decay of the
perimental measurements,the results are con-
trapped electronflux is approximatedexponen-
verted to integral omnidirectional fluxes by
tial, the loss term will be set equal to --n/r,
means of equation 2, used together with the
where r(r) is the observedexponentialdecay:
particle energy dependenceand the conversion
time. Values for r(r) were taken from the com-
factors obtained from considerations
of pitch-
pilation by Walt [1966]. For energeticelectrons
angledistributions.
r risesto a peak-alueof 470daysat L = 1.5
Since the locations where experimental data
and decreasesto a constant value of 20 days
were obtained are given in terms of the mag-
beyondL -- 2.1.
netic shell parameter L, the symbolsL and r
The experimentaldata consistof omnidirec-
will be usedinterchangeablyin the remainderof
tional fluxes of electronsat or near the geomag-
the paper.
netic equator having energiesabove a fixed
value. A useful relation between the measured RESULTS AND DISCUSSION
flux and the distribution function used in the
The experimental results of Brown [1966]
calculationis givenby the proportionality[Na-
will be consideredfirst. He observed,near the
kada and Mead, 1965]
equator, a broadening with time of a narrow
band of electronsproducedat L -- 1.765by the
2 = o, 0 Russian high-altitude detonation of November
where F is the differentialflux (em- see- Mev- 1, 1962. During a period of about 17 days im-
ster-') of particles with energy E mirroring at mediatelyfollowingthe detonation,Brown meas-
the equator. With equation 2 the experimental ured the omnidirectionalflux of electrons(> 1.9
data can be convertedto n(r, t, J = O, t), the Mev) as a function of L. The L-shell profiles
flux in invariant space,if the pitch-angle distri- of the flux showedan isolatedpeak whosefull
bution and the energy spectrumare known as a width at half maximum (FWI-IM) derived after
function of r. The distribution in pitch angles correctionsfor backgroundincreasedfrom 0.040
was taken from measured distributions where Rs to 0.053 Rs (Rs ------
earth radius) during the
availableand from interpolationselsewhere.Be- time period. The FWI-IM valueswere basedon
causeof the small solid angle in the lossconein a Gaussian shape that the profiles closely re-
the L regionconsideredhere, the uncertaintyin sembled.In the present calculationthe initial
pitch-angledistributionsintroducesan estimated flux distributionwas taken to be Gaussian,cen-
error of lessthan 10% in n(r, t, J = O, t). The tered at L ---- 1.765 and having a FWHM =
proceduresfor choosingthe energy distribution 0.040 Rs. As boundaryconditionsfor this prob-
RADIAL ELECTRON DIFFUSION 7233
i i i i i

lem, the flux valuesat L -- 1.60 and 1.90 were


(a)
maintainedat zero. These end points were suf- _

ficiently far from the peak of the profile that


any error introducedby the assumedboundary
conditions was negligible. The initial energy lO5
spectrumwas assumedto be a fissionbeta spec- ___
trum at L -- 1.765; it was transformedunder
(/, J) conservationto give the energydepend-
enceof the electronsat other L values.A single
'%/
/ D,=3.9
ANALYSIS
x10-SL
9
I I I I I I
value of Ds was assumedto apply over the small i I I i i _

range of L involved. It was found that a value (b)


of 1.3 X 10-s R/day producedcalculatedpro- _

files whoseFWHM values were in good agree-


w
-

ment with the actual profiles for the 17-day


period. 105
A similar procedurewas applied to the data
obtainedby Frank et al. [1964]. They observed
an equatorial flux profile of electrons (>1.6
Mev energy)that appeare
d to moveradiallyin- I
FRANK
I
ETAL.(1I

ward while maintainingits generalshapeduring - i

a period lastingseveraldaysfrom December20, (c)


1962, to January 8, 1963. Figure lb showstheir
_

publishedresults,omitting an additional profile


that was measuredat a time correspondingto
t -- 19 days. This latter profile was not con- 105
//3/0DAYS
sideredin the presentwork sinceit exhibiteda
significant depletion of electrons at high L
--"", // / PRESENT
/ / ANALYSIS
values,leadingto an uncertaintyin the boundary D2:1.0x10-SL
10
conditionsto be applied. I I I I I I
2.5 3.0 3.5 4.0 4.5 5.0
The profile specifiedby t -- 0 in Figure lb L
was extrapolatedsmoothlyto zero at boundaries
(L -- I and L -- 8) far from the regionof in- Fig. 1. Equatorialprofilesshowingthe syste-
terest and was then used to determine an initial matic inward drift of electrons during a period of
9 days. Values of the omnidirectional fluxes of
distributionfunction.The zero boundary values electrons above 1.6 Mev are given. The time t -- 0
were maintained throughout the calculation.It correspondsto December 20, 1962. Experimental
was found that in the regiorof L under con- results are given by (b). Calculated results are
given by (a) and (c) for different representations
sideration(2.5 < L < 5) the calculatedresults of D..
were very insensitiveto different plausibleex-
trapolations in the boundary region at low L
because of the small values of the diffusion changes(15-50%) in calculatedfluxesonly at L
coefficientin this region.Various extrapolations values above the leading edge of the profile
to the right-handboundaryproducedsignificant (L , 3.5), but did not alter the over-all time
modificationsto calculated fluxes only in the evolution of the shapeof the flux distribution.
region 4 ( L 5, where changesof 15-50% D2 was representedby a power law variation,
occurred. L " [Fgilthammar, 1966], as would be expected
For this problem, the initial energy depend- for magnetic field fluctuationswhose spectral
ence was specifiedby an exponentialspectrum power density dependson a power of the fre-
exp (--E/0.6 Mev) at L -- 4, transformedun- quency. Calculations were made for several
der (, J) conservation to give the spectrumat valuesof m, and the resultsfor m -- 9 and m --
other L values. The substitution of either 0.4 10 are shown in Figures l a and c. In both cases
Mev or 0.8 Mev instead of 0.6 Mev in the de- the theoretical curves comparefavorably with
nominator of the exponent caused significant ,the generalshape and displacementof the ex-
7234 NEWKIRK AND WALT

perimentalprofilesgivenin Figure lb. However, sonableagreementwith the values5-15 X 10-'


the casem -- 10 is preferablesincethe slopeof R'/sec that wereestimatedby Tvers/coy[1965]
the leading edge agreesbetter with experiment. from magneticdata on suddenimpulses.It also
Lower values of m give a slopethat is signifi- agreesfavorably with the resultsof McDiarmid
cantly lessthan observedand highervaluesof m andBurrows[1967] whointerpretedtheir meas-
result in large discrepancies
with the measured urementsof the positionof the outer-beltpeak
flux at high L. The successof radial diffusionin in terms of radial diffusiontheory to obtain the
reproducingthe observedflux profilesof Figure values6-9 X 10-' R'/sec.
lb is a significantfindingsincethe apparentin- Recently Kavanagh [1968b] has calculated
ward motion of structure is more suggestiveof values of D, for low-energyelectrons (50-100
convection than diffusion. key) in the interval 4 < L < 7. In effect,he
Figures la and c illustrate that the calculation compiled equilibrium electron fluxes from Ex-
cannot determine precise values for D. How- plorer 12 data, convertedtheseto n(r, 1,J -
ever, if the diffusion coefficientis a smoothly 0), and integratedequation1 to obtain
varying function of L, it should lie within the
shadedband shownin Figure 2. Also displayed
in Figure 2 are the singlevalue of D obtained -= (11r)(alar)(rn)()
from the analysisof Brown'sexperimentand the whereK is a constantof integrationwhichwas
values of D deducedpreviously by Newkirk set equal to zero. He obtained values of D that
and Walt [1968] at low L values. have a complicated radial dependence.Also,his
The solidline in the shadedregionrepresents value of D at L = 5 is two ordersof magni-
D. = DoLi,where Do = 1.0 X 10-8 R/day -- tude higher than that shownin Figure 2. Al-
12 X 10-' R'/see. This value of Do is in rea- thoughhis analysisand the presentone utilize
the sameequation,there are significantdiffer-
eneesin the approach.The presentwork is
based on the observedtime dependenceof
structuralchangesand is only slightlydepend-
ent on the decaytime r, whereasKavanagh's
resultsare inverselyproportionalto r as shown
in equation 3. Sincethe values of r that he used
were based on 40-key electrons and are about
two ordersof magnitudelowerthan the 20 days
appropriatefor higherenergies,
the largervalue
of D obtainedin hisworkis expected.
It is of interest to utilize the diffusion coef-
ficients of Figure 2 to test whether radial dif-
fusion can lead to the observed electron belts.
This point was investigatedby applying an
arbitrary initial distributionand realisticbound-
ary conditions
to equation1 and then integrat-
ing it until equilibrium was reached. At the
upperboundaryL -- 6 an arbitrary flux having
an energyspectrumproportionalto exp (--E/0.3
Mev) was assumedto exist. At the lower bound-
ary L -- 1.15the fluxwasset equalto zero.
The equatorial equilibrium flux distribution
Fig. 2. Radial diffusion coefficientD. versusL. calculatedfor electronsgreater than 1.6 Mev is
The results of the present analysis are given by shownby the dashedline in Figure 3. The cal-
the circle at L -- 1.76 and by the shaded zone culation yielded an outer electron belt whose
(2.6 < L < 15)whose center line representsD = radial distributioncomparesfavorably with the
1.0 X 10- L . The solid line in the range 1.115<
L < 1.20 shows results previously deduced by typical electron profile compiled from experi-
Newkirk and Walt [1068]. mental data by ette et al. [1966], shownby
RADIAL ELECTRON DIFFUSION 7235

thereby reducingthe flux at L <( 4.5-)1. The


small inner-belt maximum (below the scale of
Figure 3) occursbecauseof the maximumin
at L -- 1.5. Thus, the resultsshow that the
EXPERIMENTAL observationsof Brown [1966] and Frank et al.
u '1o5 .... CALCULATED [1964], if interpretedas radial diffusion,imply
a diffusion coefficientthat is adequate in mag-
nitude and has the proper radial dependenceto
producethe outer radiation belt, providing an
adequatesourceof electronsis presentat L --
.io4 6.
In conclusion, it shouldbe stressedthat the
apparent successof radial diffusiontheory in
explainingthe observedradial motion doesnot
t0 3_ prove that radial diffusionconserving/, and J
is the applicableprocess;convectionor other
o
kinds of diffusioncouldalsobe compatiblewith
the data. Nevertheless,the present resultslend
strong support to the type of diffusion con-
10 - sidered, and the diffusion coefficientsdeduced
from the analysiswill be useful for future com-
parison with coefficients derived from a con-
I
i sideration of specific disturbance mechanisms
i
I that move particlesacrossfield lines.
101 I I Ii I I I
1 2 3 4 5 6 Acknowledgments. This work was carried out
L under the Defense Atomic Support Agency, con-
tract I)A-49-146-XZ-570.
Fig. 3. Equilibrium distribution of omnidirec-
riohal fluxes of electrons (> 1.6 Mev) at the equa- lEFERENCES
tor. The dashedline givescalculatedresultsbased
on a constant flux at L -- 6 and radial diffusion to Brown, W. L., Observations of the transient
lower L values. The solid line is the typical pro- behavior of electrons in the artificial radi-
file based on various experimental data [Vette ation belts, in Radiation Trapped in the
et al., 1966].The normalizationof the two curves Earth's Magnetic Field, edited by Billy M.
is arbitrary. McCormac, D. Reidel Publishing Company,
Dordrecht, Holland, 1966.
the solidline in Figure3. The experimentaldata Cladis, J. B., Acceleration of geomagnetically
were obtained after Starfish so that the inner trapped electrons by variations of ionospheric
currents,J. Geophys.Res., 71, 5019, 1966.
belt of the typicalprofileincludes
bothartificial Craven, J. D., Temporal variations of electron
and natural electrons. The calculation also intensities at low altitudes in the outer ra-
producedan innerbelt whoseintensityliesbe- diation zone as observed with satellite Injun
lowthe scaleof Figure3 andis thereforeseveral 3, J. Geophys.Res., 71, 5643, 1966.
Davis, L., Jr., and D. B. Chang, On the effect
ordersof magnitudelessthan the experimental of geomagnetic fluctuations on trapped par-
values.In the calculation,the outer-beltmaxi- ticles, J. Geophys.Res., 67, 2169, 1962.
mum at L 4.5-)1 results from the combined
Dungey, J. W., Effects of electromagneticper-
effectsof the acceleration
of particlesduringthe turbations on particles trapped in the radia-
inward drift and the lossof particlesas repre- tion belts, Space $ci. Rev., 4, 199, 1965.
Ershkovich, A. I., V. D. Pletnev, G. A. Skuri-
sentedby . At highL, the acceleration accom- din, L. S. Chesalin, and V. P. Shalimov,
panyinginwarddrift dominates, and the more Motion of a charged particle in the magnet-
numerouslowerenergyparticlesare accelerated osphere under the action of a sudden geo-
abovethe fixedthresholdenergy,causingan in- magnetic impulse, J. Atmospheric Terrest.
Phys., 29, 459, 1967.
creasein flux with decreasing
L for L > 4.5--1. Fiilthammer, C.-G., Effects of time-dependent
At lower L the reducedvalue of D, causesslower electric fields on geomagneticallytrapped radia-
radial transfer so that the losses dominate, tion, J. Geophys.Res., 70, 2503, 1965.
7236 NEWKIRK AND WALT
Filthammar, C.-G., On the transportof trapped protons in the outer radiation belt, J. Geo-
particles in the outer magnetosphere,J. Geo- phys.Res.,70, 4777,1965.
phys.Res.,71, 1487,1966. Newkirk, L. L., and M. Walt, Radial diffusion
Forsythe, G. E., and W. R. Wasow, Finite- coefficientfor electronsat low L values, J.
Difference Methods /or Partial Dierentia Geophys.Res.,73, 1013,1968.
Equations, Section 14, .John Wiley & Sons, Paolini, F. R., G. C. Theodoridis,and S. Frank-
New York, 1960. enthal, Space and time variations in outer-
Frank, L. A., Inward radial diffusion of elec- belt electron spectra, J. Geophys. Res., 72,
trons greater than 1.6 million electron volts 4590,1967.
in the outer radiation zone, J. Geophys.Res., Parker, E. N., Geomagnetic fluctuations and
70,3533,1965. the form of the outer zone of the Van Allen
Frank, L. A., J. A. Van Allen, and It. K. ItilIs, radiation belt, J. Geophys. Res., 65, 3117,
A study of charged particles in the earth's 1960.
outer radiation zone with Explorer 14, J. Tverskoy, B. A., Dynamics of the radiation
Geophys.Res., 69, 2171,1964. belts of the earth, 2, Geomagnetismand
Imhof, W. L., and R. L. Smith, Observations Aeronomy,4, 351, 1964.
of nearly monoenergetihigh-energy electrons Tverskoy, B. A., Transport and accelerationof
in the inner radiation belt, Phys. Rev. Let- chargedparticlesin the earth's magnetosphere,
ters,14,885,1965. Geomagnetism and Aeronomy,5, 617, 1965.
Kavanagh, L. I)., Jr., A revised (B, L) coor- Vernov, S. N., S. N. Kuznetsov,Yu. I. Logat-
dinate system to allow for distention of the chev, G. B. Lopatine, E. N. Sosnovets,and
magnetosphereby a symmetric ring current, V.G. Stolpovsky,Radial diffusion of _ 100
J. Geophys.Res.,73, 185,1968a. key electronsin the outer radiation belt, CO-
Kavanagh, L. D., Jr., An empirical evaluation SPAR SpaceSci. Symp.,London,1967.
of radial diffusion coefficients for electrons Vernov, S. N., I. A. Savenko, M. V. Tel'tsov,
of 50-100 key from L -- 4 to L -- 7, J. and P. I. Shavrin, Observationof a diffusion
Geophys. Res.,73,2959,1968b. wave of relativistic electronsin the outer
Kellogg, P. J., Van Allen radiation of solar radiationbelt, Geomagnetism and Aeronomy,
origin,Nature, 183, 1295,1959. 6, 503, 1966.
McDiarmid,I. B., and J. R. Burrows,Depend- Vette, J. I., A. B. Lucero,and J. A. Wright,
ence of the positionof the outer radiation Modelsof the trappedradiationenvironment;
zoneintensitymaximaon electronenergy.and vol. 2, Inner and outer zoneelectrons,
NASA
magneticactivity, Can. J. Phys., 45, 2873, SP-3024,NationalAeronautics and SpaceAd-
1967. ministration, Washington,D.C., 1966.
McIlwain,C. E., Long-termchanges in the dis- Walt, M., Lossrates of trappedelectronsby
tribution of the 40- to 110-Mevtrappedpro- atmosphericcollisions,Radiation Trappedin
tons (abstract). Trans. Am. Geophys.Union, the Earth's Magnetic Field, edited by Billy
46,141,1965. M. McCormac,p. 337, D. Reidel Publishing
Nakada, M.P., J. W. Dungey,and W.N. Company,Dordrecht,Holland,1966.
Hess..On the origin of the outer-belt protons,
1,J. Geophys.
Res.,70,3529,1965. (ReceivedJune 21, 1968;
Nakada, M.P., and G. D. Mead, Diffusionof revisedAugust19, 1968.)

You might also like