- Email: [email protected]
Magnetospheric plasma motion during a sudden commencement
Ptanet.
Space
Sci. 1973,
Vol. 21, pp. 1713 to 1729.
Pergmon
Pms.
Printed
in Northern
Ireland
MAGNETOSPHERIC PLASMA MOTION A SUDDEN COMMENCEMENT
DURING
C.-A. LIN, D. T. YOUNG* and R. A. WOLF Department of Space Physics and Astronomy, Rice Wniversity, Houston, Texas 77001, U.S.A. (Received 2 April 1973)
Abstract-A sudden commencement occurred at 2348 UT on 15 February 1967, when the ATS-1 satellite was about 2 hr past local noon at a geocentric distance of 6.6RB. Plasma was observed by the Sup~the~al Ion Detector (SID) first to flow in the antisolar direction, as expected, but then to flow westward, for about 2min, at about SOkm/sec. Analysis of ground magnetograms suggests that the surprising westward flow, which must have involved an electric field of about 10 mV/m at 66R,, resulted from the ionosphere’s reaction to the sudden commencement. l3eginning about 2 min before the start of the westward flow at ATS-1, ground ma~etomete~ near the foot of the ATS-1 field line typicahy recorded ma~etic-field deflections of about 70 y. to the northeast. No attempt is made in this paper to explain these ground observations. However, taking the ground observations, assuming a height-integrated Hall conductivity of 1 mho, and a standoff distance of 7*2R, inferred from Explorer 33 solar-wind data, we derive a magnetospheric electric field that agrees in magnitude and direction with that required to produce the observed flow at ATM.
At the time of the magnetic storm sudden commencement of 15 February 1967, conditions in the solar wind were being monitored by Vela 3A and Explorer 33, and ATS-1 was in the dayside ma~etosphere at 13 : 31 ma~eti~ local time. The event is unique in one respect: the electric field and the density of cold plasma were both sufficiently high that the Suprathermal Ion Detector (SID) aboard ATS-1 provided vector measurements of the The aim of this paper is to present E x B-drift velocity during the sudden commen~ment. the SID data and an interpretation of the results. The data have been presented before (Young, 1969), but never published. DATA
The Rice University Suprathermal Ion Detector has been described elsewhere (Freeman, 1968). Briefly, the detector measures energy per unit charge in 20 differential steps between 0 and 50 eV and in two integral modes of E ZI== 0 eV and E > 50 eV. Energy discrimination is accomp~shed by applying a 200 c/s square wave to a positively biased planar grid. I#y using an up-down counter which counts up during the excursion of the square wave below the bias level, and down during the excursion above the bias level a differential count is obtained. The width of the di~erential passbands changes with energy in the manner shown in Table 1. Because a complete energy spectrum requires 112 set it is obvious that a true differential spectrum can be obtained only if: (1) the spread of an energy spectrum incident at a particular enera is broader than the passband at that energy, and (2) the incident flux is nearly constant over a period st 112 sec. The satellite is oriented with its spin axis antiparallei to the Earth’s spin axis so that the field of view of the SID is symmetric about the equatorial plane. Since the angular response of the detector is Gaussian in shape with cz fi: 6.5” about the axis of symmetry, the instrument only measures directional fluxes near the equatorial plane. The satellite rotates * Present address: Physikalisches Institut der Universitiit Bern, Sidlerstr.5, 3000 Bern, Switzerland. 7
2713
1714
C.-A. LIN. D. T. YOUNG and R. A. WOLF TABLE 1. ENERGY PASSBANDSOF SID DETECTOR
Energy (eV)*
Number of steps in the interval
Passband widths (eV>
OI‘?&~5 61E,< 10 15 < E, 1_<45
10 5 4
O-5 1.0 10.0
* E. is the energy associated with the center of agiven passband. through approximately 12’ between accumulation directions of arrival can be identified to within
intervals so there is minimal overlap and about 12”. Figure 1 shows the spatial
coordinates that will be used throughout this discussion. The channeltron itself is biased 3.2 kV negative with respect to ground; electrons with energies less than 3.2 keV are thus excluded. Detection efficiency reaches a relative maximum for 4-5 keV electrons and then falls off rapidly for more energetic ones. AT&I
data
At 2348 UT, the approximate time of the beginning of the sudden commencement at CoIlege, ATS-1 was located as shown in Fig. 1. magnetometer data supplied by W. D. Cummings and P. J. Coleman, Jr. {private communication) shows that the magnetic B field at ATS-1 increased steadily from 135 y to 195 y between 2348 and 2352 UT. During this same period the SID showed a factor of 3 increase in the isotropic background flux composed of ions with E > 50 eV and electrons with E > 3.2 keV, In addition a sequence of low-energy, highly anisotropic fluxes were observed during the period when B was increasing and for about 1 min afterwards. An overall view of the event is shown in Fig. 2. Figure 3 affords a more detailed description of the magnitude and direction of ion fluxes and the associated vector magnetic fieId at the ATS. The magnitude of the H component of
FIG. 1. ImxnoN OF ATS-1 AT 2348 UT, I5 FEBRUARY1967. Thef plane of the paper is the geographic equatorial plane, and the view is down on the north pole. The coordinates are solar-geographic. The Earth’s Equator is in the xy plane. The Earth-Sun line is in the xz plane, 14-l” below the xy plane. The satellite spins clo&wise. @ is the azimuthal angle describing the detector look direction.
MAGNETOSPI3EIW.Z PLASMA MOTION DURING A SUDDEN COMMENCEMENT FEB.
IS,
I967
0 ISOTROPIC WXGROIJND l
2344
2346
2352
2%
FLU
MAWJITUDE OF TOThL MA6NEtlC
ii40
1715
2400
0004
FIELD AT AT-S.
0008
0012
0016U-f
mu. 2. iI%! AT ATS-1 (UNPUBLls~ hfAGNETOMETER DATA SUPPLSED BY w. D. Cummas AND P. J. COLEMAN, JR.). Values of the isotropic background flux in the Da& we 60 set intervals with no amaging. E > 0 eV channel were taken in other than ffow directions. The two shaded regions indicate periods during which the low energy anisotropies were observed.
the field increases steadily, while that of the equatorial component remains relativeIy small throughout. Because of the relatively large geographic northward component B,, the direction of the total B field is nearly constant through the event. The geographic equatoriaf component, B,, = B, + By, changes direction appreciably, however, during the period when flow is observed in the antisolar direction. It is clear from Fig. 3 that the direction of the low energy particle fluxes divides the event into two distinct parts, viz. flow in an antisolar direction (0, M 180’) and ffow from dusk to dawn and roughly perpendicular to the Earth-Sun line (6, M 100”). The flow directions and flux levels for these flows (1 and 2) are indicated in Fig. 4. The angular distribution for all particle fluxes throughout the event is at the limit of resolution for the detector so the flow directions are correct to ~12”. Typical angular dis~ibutions for fluxes throu~out this period are shown in Fig. 5. Comparison of the integral channel differences in the antisolar and dusk-dawn flows indicates that about 80 per cent of the incident flux was composed of positive ions with E < 50 eV for the former case with the fraction changing to 70 per cent for the latter. About the only generalization that can be made concerning the energy in the two spectra is that the dusk to dawn flow is slightly ‘harder’ than the other. This conclusion is due mainly to percentage increases in the E > 50 eV and the E = 15 + 5 eV channels. No attempt has been made to present differential spectra using this data despite the directional consistency of the fluxes over sufficiently long periods. This is due to evidence to be discussed later that temporal changes in ion number density and velocity might well have been large enough to invalidate the usual interpretation of the differential data. Further, a large portion of the fluxes occur in the higher energy channels where the passbands are 10 eV wide. Angular distribution data shows that the spread in thermal energies
C.-A. LIN, D. T, YOUNG and R. A. WOLF
ENERGY
PER UNIT CtiARGE f&‘)
I90 a0 170 g IBQ
=”
150
40
140 ~ 30 ;= % 130 20 m” 120
2348
2349
2350
Frci. 3. flETAft_ED
2351
2352
PLASMA AND ~A~N~&-~
2353
IO
2 $4 UT
DATA.
The plots of particie-ffux ~gn~tud~ represent the fIux2 the direction of maximum counting rate, minus the isotropic flux; error bars indicate 3: 1/N Poisson counting error, The magnitudes of the northward field & and equatorial field Bep are in y. The angle 8, represents the direction of the equatorial component of B, measured ciackwise from the Sun as seen by an observer north of the equatorial plane. The angle 0, gives the direction of the equatorial component of the flow velocity, measured ciocks~ise from the Sun. Error bars indicate angular spread away from the flow direction of fluxes more than 1 a above the mean. 0, w 180” represents Bow away from the Sun, while 0, @ 90” represents flow in the westward t-y) direction, 8, = Q, _t 180”, where QIis shown in Fig. 1. was ~1 eV for 15 eV fluxes, thus destroying the differential nature of the higher energy
channels. After cessation of the ~r~ct~on~ fluxes the isotropic background rate remains elevated (Fig_ 2) as does the ma~itude of B. The differential channels show fluctuations typical of >50 eV background particles. There is no consistent dir~~~on~~ty or preferred energy in these data and the flux seldom reaches l-5 x IO’ cmW2see-l sr-I. This was pretty much the state of affairs for the Iow-energy particles until 0625 UT on 16 February. At 06 25 OS UT, the 15 3: 5 eV channel indicated a flux of about 3.5 x 10’ ions cm-2 set-l sr-’ traveling in the direction shown in Fig. 4 for flow 3. Thirty set later the integral channels indicated the same flux level in the same direction with energy E < 50 eV. Other differential channels during this short interval registered fluxes of - IO7 ions crnTa see-l srl in the same direction. AIthough this third flow observation is based on a smait sample, it is highly significant statistically and there is no reason to regard the data as spurious, ~~~~~-w~~d data The sudden ~omrnen~rne~~ of 15 February 1967 was caused by an ~nterpia~~tary shock related to a class 4B solar flare that began at 1746 UT on 13 February (Hirshberg and
MAGNETOSPHERIC
PLASMA MOTION DURING
A SUDDEN COMMENCEMENT- 1717
SUN
SYNCHRONOUS ORBIT r=6.6RE
Flow
Date (1967)
UT
Energy per unit charge (eV)
1
15 Feb.
2
15 Feb.
3
16 Feb.
23 48 58 23 49 29 23 52 44 23 53 14 06 25 08 06 25 39
15 f 5 <50 15 &5 (50 15 f 5 <50
Flux (ions/cm*srsec) 3-5 6.2 24.8 2.7 3.5 3.3
x x x x x x
10’ 10’ 10’ 10’ 10’ 10’
FIG.~. SOMEBASIC CHARACTERISTICSOP THEPLOWSOBSERVED ON 15 AND 16 FEBRUARY 1967. Quoted flux magnitudes represent the flux in the direction of maximum counting rate minus the isotropic flux.
Colburn, 1969). Solar-wind data from Explorer 33, for just before the interplanetary shock and just after it, are shown in Table 2. The directions are expressed in terms of latitude and longitude in solar-ecliptic coordinates. More limited plasma data from Vela 3A yield densities that are about 0.7 times the Explorer 33 values, and flow velocities about @85 times the Explorer 33 values (Hirshberg et al., 1970). Hirshberg et al. (1970) have used the pre-shock and post-shock plasma data to estimate the direction of the outward shock normal. The result, in solar-ecliptic coordinates, is lat -59.0” f 45” li ( long 190.7” & 3.5”. In other words, the shock normal makes an angle of about 60” with the plane of the ecliptic. The shock slammed down on the magnetosphere, impacting in the region of the northern polar cusp instead of near the subsolar point. Ground magnetometer data In Tables 3 and 4, we present data from ground magnetometers located near the foot of the ATS-1 field line. Table 3 lists the positions of six stations in geomagnetic coordinates. The pre-commencement magnetic fields are also given in Table 3, where His the horizontal component, D is the declination angle (positive eastward), X (= H cos D) is the geographic horizontal northward component and Y (= H sin D) is the geographic horizontal eastward component. All these magnetometers were quite steady just prior to the sudden commencement. Table 4 lists magnetic-field deflections away from the pre-storm values, as measured from available normal and rapid-run magnetograms, where AX, A Y are the changes of the geographic horizontal northward and eastward components in y, respectively, and AX’,
1718
C.-A. LIN, D. T. YOUNG
and R. A. WOLF
.5f .25 a!/ CtiANNEt TM.YEl.EMETRY
0
35
70
AZIMUTHAL
CHANNEL
IO5 I39 174 249244 27S 314 348
ANGLE
@, t CLOCKWISE FROM SUN 1
FIG. 5. TYPICAL ANGULAR-DISTRIBWIION DATA DURING THE PERIOD OF
TABLE 2.
Density * (protonjcma) Velocity* &m/s4
Mapretiejieldf (Y)
SOLAR-WIND DATA FROM &PLGlW-t
33
Pre-shock
Post-shock
?I*=
Jlp =
1t.M
ANTISOLAR
24.0
Y( = 333 direction (solar ecliptic) coords.): lat = -5” long = 180” + 2”
0, = 450 direction (solar ecliptic): lat = -16” long = 180” + 10’
B< = 6-8 direction (solar ecliptic): lat = 45” long = 254”
B, = 17.5 direction (solar ecliptic): lat = 30” long = 245”
* Data from H. Howe, private communication. t Data from Hirshberg et al. (1970).
FLOW.
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
TABLE 3. OBSBRVATORYCOORDINATESAND PRE-STORM MAGNETIC
Position Geomagnetic coordinates 10llg lat
Observatory Mould Bay Resolute Bay Baker Lake Barrow College Meanook
FIELDS
(MLB) (RB) (BL) (BW) (CO) (ME)
Observatory
79-l” N 83.0 73.8 68.5 64.6 61.8
256.4” E 289.3 315.2 241.1 256-S 301-o
Pre-storm magnetic fields x = 1021.2 y X= 123.4~ X = 4171 y D = 25” 44 D = 28” 28.2 D = 23” 31.8’
MLB RB BL BW
co ME
Y = 2188.9 y Y = -1396y Y=27Oy
H = 9678 y H = 12868 y H = 13157.1 y
TABLE 4. GEOMAGNETICMELD CHANGES AX
UT
Mould Bay 23 47 08 48 00 49 00 49 30 5000 51 00
52 00 53 00 55 00 58 00
0 11 2: 62 42 -49 -83 -187 -231
AY
AX’
1:
18 20 21 22 -22 -56 -181 -184
Ahy’
0 18 41 156 z -47 -94 -257 -286
0 -2: -175 -38 -20 26 33 31 75
Resolute Bay 234600 48 00 49 00 50 00 51 00 52 00 53 00 55 00 58 00
9’: 85 62 0 -7 --: -92
-5: -35
0 28 59 101 118 94
0 30 35 44 20 2: 36 1
0 -100 -85 -44 20 47 76 132 131
Baker Luke 23 48 49 50 51 51 52 53 55 58
00 00 00 00 30 00 00 00 00
0 8 33 40 50 36 16 -27 8
8 6 19 30 35 -8; -84
0 8 34 44 57 45 16 -51 -18
0 -3 -4 6 13 22 -2 -70 -83
1719
C.-A.
1720
UN,
D. T. YOUNG
and R. A. WOLF
TABLE 4.-(continued) UT
AX
AY
AX’
0 -10 34 38 41 -22 -82 -85 -65 -52
0 -2 -4 -38 64 139 218 210 200 194 214 227 204 70
0 -10 27 13 68 53 43 35 41 55 56 60 127 10
0 1
2
AY’
Barrow 23 47 48 48 49 49 50 50 51 51 52 52 53 55 58
45 00 30 00 30 00 30 00 30 00 30 00 00 00
-64 -55 25 -30
0 3 -21 -52 34 130 230 224 205 193 216 223 161 75
ColIege 23 47 48 48 00 48 30 49 00 49 30 50 00 50 30 51 00 51 30 52 00 52 30 53 00 55 00 58 00 6000
Meanook 23 47 06 48 00 49 00 50 00 51 00 52 00 53 00 55 00 58 00
0 -6 -25 -6 18 24 24 -20 -30 -10 3 17 77 68 32
0
14 60 127 110
93 83 77 49
-6 -18 14 70 104 152 128 85 66 64 56 36 17
0 -1 -1 10 44
24 _y -13
-25 -13 22 51 65 45 25 27 30 42 94 76 36
0 3 5 -14 5 53 liz 129 82 59 51 19 5 2
0
0
14 58
-4 -16 -22 15 0
126 118
96 82 73 44
-11
-23 -25
A Y’ are the changes of the geomagnetic horizontal northward and eastward components in y, respectively. Note that AX and A Y increase for 2 or 3 min then decrease; also note that AX decreases after 5000 f 0100 UT. INTERPRETATION
Flow 1
The observation of antisolar flow early in the sudden-commencement event is consistent with the idea that the SC represents a general compression of the magnetosphere and thus an inward motion of the magnetopause and the plasma just inside it. The proton flow velocity corresponding to 15 eV, the center-energy of the channel showing maximum flux, is about 54 km/set, which, multiplied by a flow-duration of
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1721
3 min, suggests an average inward displacement ~15R, for plasma moving past ATS-1. The geocentric distance to the subsolar point can be estimated using the formula (Fairfield, 1971) rb = [$l”6[;;;;;:(1”“E
(I)
where m, is the proton mass, and u and n are the solar-wind number density and velocity, k is a measure of how efficiently solar particles transfer their momentum to the magnetosphere, andfrelates the geomagnetic field just inside the magnetopause to the undistorted dipole field at that point. Using the data from Table 2, we find the pre-storm standoff distance rbi and post-sudden commencement standoff distance rb, are rbi = 8-56RB,
rb, = 674R,
for
‘i = O-50;
rbi = 9*1RE,
rbf = 7*16RE for
$ = 0.72;
rbi = lO*lR,,
rbf = 8*ORB
f
for
= 1.40.
(2a)
(24
Fairfield (1971) has compared the distribution of magnetopause standoff distances for 137 boundary crossings with the distribution of standoff distances computed from Equation (1) and a set of 3146 hourly averages of nmve in the solar wind. He found that the distributions fit best for f2/k N l-40. However, the distributions did not agree for small standoff distances: 8 of the 137 observed crossings indicated rb < 8_5R,, but a negligible fraction of the hourly averages gave sufficiently large nmv2 to indicate rb < 8.5, for f2/k = l-40. This suggests that, under conditions of severe magnetospheric compressionp/k may be substantially less than 1.40. In the following calculations, we tentatively choose f2/k = O-72, keeping in mind that the actual value may be substantially larger. However, Equation (2) indicates that for 0.5
The westward flow observed in the evening sector, late in the initial phase of the storm, should presumably be interpreted as the result of enhanced convection from the tail, the convection that probably causes the main phase of a storm. Thus flow 3 does not seem particularly puzzling and will not be discussed further here. Flow 2
The 50 km/set westward flow, which persisted for about 2 min late in the sudden commencement event, cannot be explained so easily. The remainder of this paper is devoted to the analysis and interpretation of that flow. Outline of proposed interpretation
We suggest that the dramatic westward flow during the sudden commencement is a result of the reflection, from the ionosphere, of the initial earthbound sudden commencement wave. Figure 6 illustrates the suggested sequence of events. The interplanetary shock first contacts the magnetosphere at point I. Magnetospheric waves caused by that impact travel
1722
C.-A. LIN, D. T. YOUNG
and R. A. WOLF
FIG. 6. DIAGRAMILLUSTRATING SUGGESTED INTERPRETATION OF THE WESTWARD FLOW. The plane of the paper is the noon-midnight meridian plane, with the north pole up. The progress of the interplanetary shock through the magnetosheath is described by showing the shock configuration at 40 set intervals. The interplanetary shock tirst hits the magnetopause at point I.
down along field lines into the north-polar ionosphere and give rise to a system of currents that spreads out over the northern ionosphere. We have not gained any new insight into the reason for these currents having the magnitudes and directions they do. We just consider the implications of the observed currents. The existence of the currents implies the existence of electric fields in the ionosphere, electric fields that propagate, by means of AlfvCn waves, along field lines into the outer magnetosphere, causing rapid E x B drifts there. The current pattern generated in the northern ionosphere involves electric fields, which, in the early afternoon sector, have large northward components. Thus flux tubes in the topside ionosphere tend to drift westward. This westward motion of the ionospheric ends of the field lines near ATS-1 requires a corresponding westward motion in the equatorial plane. Plasma density at ATS-I To estimate wave travel times in an effort to clarify the sequence of events, we need information on the density of plasma in the outer magnetosphere. One might expect that the plasmapause would be at L > 6.6 at the time of the SC, because magnetic activity had been low for several days. The maximum value of the 3 hr Kp index for 15 February 1967 was I+, excluding the last period of the day, which included the SC. For 12, 13 and 14 February the maximum Kp indices were l+, 2 and 2- respectively. On the other hand, the fluxes measured by the SID detector indicate number densities below usual plasmasphere levels. A peak flux of 6 x 10’ ion/ems set sr, with a streaming
MAGNETOSPHEFUC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1723
velocity of 54 km/set in the equatorial plane, corresponds to a number density of the order of 2 cm-s. However, this calculation of number density from flux at ATS-1 was based on the assumption that the ions were streaming in the equatorial plane. This may not be true: for one thing, the magnetic field is about 9’ away from being normal to the equatorial plane; also, the ions may be streaming supersonically southward along the field lines, since the shock came down on top of the north-polar region of the magnetosphere. As a result, the SID detector, with viewing angle limited to within about 6.5” of the geographic equatorial plane, may never see the peak ion flux, in which case the density of 2 cm4 quoted above may be much too small. Computed wave travel times
We need to make theoretical estimates of several times: tlB, the time required for a wave to travel from the impact point I to point B, at the northern ionospheric end of the polar cusp; tIA, the time required for the compressional wave to reach ATS-1, once the wave has impacted at I; tBA, the time required for an Alfven wave to travel from the aurora1 ionosphere to ATS-1; and tAD, the time required for a compressional wave to propagate from ATS-1 to the equatorial ionosphere. We assume that B ds tlB
%
-
s2
vA
where ds is an element of length over the minimum time path from I to B, and VAis the local AlfvBn velocity. The plasma density along the path from I to B is hard to estimate. We suspect that the minimum time path is one that moves along field lines that are not part of either the plasmasphere or the polar cusp. The polar wind may not reach large geocentric distances on these field lines while they are on the dayside of the Earth, so the number density is likely to be small in the outer parts of these lines, where the wave velocity is low due to the small magnetic field. For an order-of-magnitude estimate, we therefore replace SI”d+‘, by %/VA ( near point I) and obtain, taking the magnetic field at point I to be 56 y, as suggested by Equation (10) of Mead (1964),
tlB-
16 set
nl 1I/2
[ lcmm8
’
(3)
The first compressional-wave ray to reach ATS-1 crossed the magnetopause at some point C. The time delay tIA is equal to t,, + tCA, where tCA is the travel time from C to A, and tic = t, - tI is the difference between the time the shock passes C and the time the shock passes I. To estimate tIC, we have analyzed the motion of the interplanetary shock through the magnetosheath. Figure 6 displays shock contours for the plane containing the initial shock normal and the solar-wind flow direction. The shock contours were constructed graphically, for 40 set intervals, from sound and flow velocities plotted by Spreiter et al. (1966), for y = +, M, = 8, and upstream flow velocity v, = 400 km/set. We estimate tcA to be the distance from C to A, divided by the average AlfvCn speed; in the latter, the average magnetic field was estimated to 120 y. Choosing the point C on the magnetopause in such a way as to minimize tIA, we obtain 20 set tIA
=
50 set
110 xc
fern, = 1 cmm3 for noA = 10 cm-’ for ncA = 100 cms3
(4)
1724
C.-A. LIN, D. T. YOUNG
and R. A. WOLF
where flcB. = [(n1’e>]2is the square of the average of zF2 over the ray path from C to A. Comparison of the ATS-1 data with rapid-run magneto&rams suggest that the SC arrived at ATS-1 either approximately the same time as, or less than a minute later than, its arrival at the northern aurora1 zone. This is consistent with theoretical relations (3) and (4). To estimate tBd, we assume the Alfven wave travels along an L = 6.6 dipole field line with constant number density nG.s; we obtain tBA = (8.2 sec)(ns.$2
(51 where ~t~.~is in cm-a. Assuming n*.* < 120 cm--a, the value measured at L = 6=6 by Chappell et al. (1971) on a pass that followed four days with Kp < I+, and ns+ > 2 crnd, the value suggested by ATS-1 data, we find that 12sec g tBa g 9Osec. The time required for a compressional wave to move in the equatorial ATS-I to the equatorial ionosphere is given by 6*6~?#
(61 plane from
&
tAD =
s l&t vA(eq. PI-) * We assume that the plasmasphere is completely full, and take ns
(250 cm-3)(5[L)2’”
representative of a completely filled plasmasphere. (See the dayside density profife measured on 24 October 1968 by Chappell et al., 1971.) We assume a dipole magnetic field and neglect l/va outside the plasmasphere, because of the low density there, and find that tar, m (73 set> [%]z”5
(3
where &,, = 6.6 if ATS-1 is inside the plasmasphere, and L,,, is the L-value of the plasmapause if ATS-1 is outside the plasmasphere. If the plasmapause lies at L > 5, the time required for a compressional wave to travel from 6*6R, to the Earth and reflect back to 6*6R, again is about IO0 set (within about a factor of 15 either way). This travel time is about the same as the duration of antisolar flow at ATS-1. We suggest that the flow persisted until stopped by the compressional wave reflected back from the Earth. MAPPING BETWEEN !t’HE EQUATOBIAL PLANE AND THE NORTH-POLAR IONOSPHERE
Magnetic-field-line mapping calculations were performed using a three-dipole model developed by T. W. Hill (personal communication). One dipole is that of the Earth, and the other two dipoles are parallel to that one; one has a dipole moment 125 times the Earth’s and is placed at X,, = 37~8R,; the other has a dipole moment -2 times (the minus sign means antiparallel) the Earth’s and is placed at X,, = -3OR,. With these values of the parameters, the magnetopause standoff distance is 7*16R,. This simple image-dipole model has the advantage of giving a realistically shaped ideal magnetopause with no magnetic field normal to it. Figure 7 shows the ATS-1 mapping point computed using this model (point marked ‘A7’). Also plotted for comparison is the point ‘A,’ which maps to ATS-I in a model that neglects magnetopause and tail currents. The motion of the magnetopause almost into ATS-1 causes the AT&l mapping point to move north and west, close to the point ‘P7’, which maps to the magnetopause.
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1725
FIG. 7. DIAGRAMSHOWINGTHE NORTH-POLAR IONOSPHERE AT 2348 UT ON 15 FEBRUARY. The dashed line marked ‘Sunset line 1’ is the line on which the solar-zenith angle is 90”. The solid line 2 shows the line on which the solar-zenith angle is 95-5”. The line of sight between the Sun and an observer, standing on the solid line at I20 km altitude, just grazes the surface of the Earth. The points marked A8, A7 and A6 represent the points that map to ATS-I, for standoff distances of 8*ORx, 7*16R, and 674&, respectively. ‘Am’ represents the ATS-1 mapping point, computed neglecting magnetopause currents. P8, P7 and P6 represent the points that map to the magnetopause for standoff distance 8*OR,, 7_16R, and 6*74R8 respectively. They are not on the Earth-Sun line because the solar wind is not flowing along the Earth-Sun line.
The mapping calculations have also been done for standoff distances of 6*74R, and
S-OR,.The results are indicated in Fig. 7 by points A6, P6, A8 and P8. ~~pping of the ~ono~p~criceIec~r~c~e~dto the eq~a~orialplane We interpret the westward flow in the equatorial plane as being part of a secondary event caused by the ionospheric reaction to the SC. This secondary event, unlike the primary compression, does not cause a significant net change in the magnetic field in the magnetosphere: it consists of a rearrangement of flux tubes. Consequently, we use the same magnetic-field model to analyze the entire secondary event. We can label a given field line by either of two pairs of numbers: +,, and tJi, the geomagnetic longitude and colatitude of the field line at some arbitrary Axed height in the ionosphere; or r, and +,, the geocentric distance and geomagnetic longitude of the field line, where it crosses the geomagnetic equatorial plane. In the secondary event, the ionospheric coordinates of the field line change by amounts hOi and A&., while the equatorial coordinates change as follows:
1726
C.-A.
LIN,
D. T. YOUNG
and R. A. WOLF
Attributing the displacements A$,, he,, A@,, A+, to E x B drifts, and putting E B = 0, we can use Equations (Sa) and (8b) to relate the time integrals of the secondary-event electric fields in the ionosphere and equatorial plane. Let e,, be fEre dt, the time-integr~ over the secondary event of the radial component of the electric field in the equatorial plane. Similarly, let .sde= SE,+ d t, set = jEef dt, edi = SE+, dt, where the subscripts e and i refer to the equatorial plane and ionosphere. Equations @a) and (8b) then become l
where RE is the radius of the Earth at ionospheric altitude. The square-bracketed coefficients depend on the rna~~tic-fiend model. dipole model, we have (Mozer, 1970)
For a simple
which becomes, for L = 6.6, Ezs = -00316,*
(1 Ia)
and
To get an idea of how the magnetopause affects mapping of electric fields just inside it, consider a ~ylindri~lly s~metr~c magnetosphere with a ~eld-~~~ed boundary that has equatorial radius to. Assume that the region just inside the magnetopause has magnetic field B, in the equatorial plane; this region maps to the north pole region of the ionosphere, where the magnetic field is B,,, The region within a small colatitude angle Oi of the north pole maps into the equatorial region between re and r,. If 0, and 1 - r,fr, are both much less than unity, conservation of magnetic flux then implies that W+,,RE2 ei2 = 3, r. PO - r,>, or
which, substituted into Equation (9) along with the relation dr = +*, gives err w
-+[$-J1ia[&J1’a
(1%
and
Far from the ma~etopa~~, the east-west component of the electric field is attenuated a little less in mapping to the equatorial plane than the north-south component is. But the
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1727
situation is very different for field lines that go close to the magnetopause: the equatorial electric field normal to the magnetopause gets very large as rs + r,,, while the tangential component goes to zero. We have used the 3-image dipole model with standoff distance 7*16RE to map from 8 points in the equatorial plane near ATS-1 to the ionosphere, to get estimates of the coefficients in Equation (9). The results are &78M Eei(-o*12) + &ci(-o*0159)
(13a)
ede M e,,(O*OO38)+ ~(0.033).
(13b)
Comparison of these results with Equation (11) indicates that the radial electric field is considerably enhanced over the dipole result, while the east-west field is reduced. Both results are in qualitative agreement with the effects indicated in Equation (12). We have also derived coefficients for a standoff distance of 6*74R,; the results are in good agreement with the idea suggested by Equation (12a) that, near the magnetopause, Ebbis proportional to the inverse square root of the distance to the magnetopause. Rewriting Equation (13), neglecting the small off-diagonal terms, but taking approximate account of variations in the distance to the magnetopause, yields 0*60RE & ,d 5% %i
)I l/2 )I l/Z
d magnetopause
(14b)
is the distance from ATS-1 to the magnetopause and 0*60R, is the wheredmegnetopauae approximate distance from ATS-1 to the magnetopause for standoff distance 7*16R,. Note that the E x B drifts induced just inside the magnetopause by AlfvCn waves from the ionosphere are nearly tangential to the magnetopause. We have also derived coefficients for a standoff distance of 8*OR,. The results are &r# % &,,(--O-035) + &,(-0@031)
(15a)
E+ M e,,(0*0013) + a+~(o*O43).
(15b)
Ionospheric conductivity
As indicated in Fig. 7, the ATS-1 mapping point lies approximately on the sunset line, i.e. the line on which the solar-zenith angle is 90”. Sunlight can reach the region of the E-layer that maps to ATS-1, but only by reflection or scattering, or by traversing hundreds of kilometers path length in the R-layer. We thus expect the photoionization flux to be low. Field lines in the region are closed: heating by polar-cusp plasma (Kennel and Rees, 1972) should not be important in the immediate vicinity of the ATS-1 mapping point. Most authors assume that, at local noon at mid-latitudes, the height-integrated Hall conductivity is about 20 mho. (See Gottlieb and Fejer, 1967.) The electron number density in the dayside E-layer is roughly proportional to cos112a, where a = solar-zenith angle; we expect the height-integrated Hall conductivity to vary in a similar way, for a less than 90” and not too close to 90”. Estimates of the nightside height-integrated Hall conductivity, outside the zone of aurora1 precipitation, are typically -0.15 mho. Having no accurate way to estimate a,, the height-integrated Hall conductivity around the ATS-1 mapping point, we assume oH - 1 mho.
C.-A. LIN, D. T. YOUNG
1728
and R. A. WOLF
Ionospheric electric fields
Having no real knowledge of the 3-dimensional current system during the sudden commencement and no knowledge of the relevant inhomogeneities in Hall conductivity, it seems most reasonable (Vasyliunas, 1970) to assume that the ground magnetic variations are due to overhead Hall currents, which means that, near the north pole, the horizontal components of the ground magnetic variations are related to the horizontal ionospheric electric field by
or -2
= -1.6 mV/m
Egi = -AX’
(AX’lly) (oH/l mho)
POOH
E,i = 2AY’.
(16b)
POOH
The strong westward flow at ATS-1 began between 23 50 30 and 23 51 30 and ended about 23 53 20 UT. Taking the time-delay to be 60 set i 30 set, consistent with inequality 6, the corresponding ionospheric times should be, roughly, 23 50 00-23 52 20 UT. It is clear from Equation (13a) that the east-west ionospheric electric field has little effect on the east-west motion of plasma in the equatorial plane, so we need not worry about A Y’. Between 23 50 00 and 23 52 20 UT, AX’ is consistently positive at nearly all the nearby stations, and is of the order of 50 y (see Table 4). ElectricJelds
at AT&l
Assuming that the ionospheric and magnetospheric have, substituting Equation (16a) in Equation (14a), mV E'eN9'6~
electric fields act over 2 min, we
1
l/2
(AX’/5Oy) ' (~a/l mho)
Recall that &cg =
E,, dt s
and
aei =
(17) '
Egi dt. s
The observed equatorial electric field, i.e. the electric field required to produce the observed E x B drift velocity of about 54 km/set west in a northward magnetic field of about 190 y, is about 10 mV/m, radially outward. The electric field derived from the ionospheric currents is in the correct direction to explain the westward motion at ATS-I, and its magnitude is sufficient to produce the observed speeds. Note, however, that the derived electric field is sensitive to the assumed standoff distance, both through the factor (0.6/d)l12 and through the conductivity. (Moving the magnetopause away from ATS-I moves the north-ionospheric end of the ATS-1 field line southward, toward the sunlight.) Beginning and end of the westwardjow
Different magnetometers near the ATS-1 mapping point developed their large northward deflections at various times between 23 48 00 and 23 50 00 UT. The fact that the westward
MAGNE~SF~E~C
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1729
at ATS-1 did not develop before about 23 50 30 UT can be explained either as a result of the AIfvCn wave travel time or by the fact that for given Eet, Er, increases as the magnetopause approaches ATS-1. Thus, we would expect the westward flow at ATS-1 to become strong as the inward motion of the magnetopause neared its end, and that is what the observations indicate. Detectable directional fluxes at ATS-1 terminated abruptly about 23 53 20 UT. Within our picture, this could be due to one or a combination of the following four effects. (1) Termination of the northward electric field at the northern-ionospheric end of the ATS-1 field line. The time of the termination of the strong northward electric field varies from one auroral-zone station to another, but in a few cases it ceases between 23 51 00 and 23 53 04 UT, due to the decreasing of X, Y, H and D components. (2) Reflection from the southern ionosphere of the Alfvkn wave caused by the northward electric field in the northern aurora1 zone. Since the southern-ionosphere end of the ATS-I field line is in sunlight, the conductivity should be substantially higher there. Because the southern ionosphere will not accept large electric fields, it should reflect the AIfvCn wave. The electric field of the reflected wave will oppose that of the initial wave and will greatly retard the flow at ATS-1, once it reaches the satellite. The northward electric field at the ATS-1 mapping point should drop about 60 & 30 see after the cessation of flow at ATS-1. This model would suggest that the duration of the rapid westward flow at ATS-1 should be about 2t,,, or 24-180 set, which is consistent with the observations. (3) ATS-1 may have moved into a region where the average thermal velocity was high enough to make the streaming unobservable. (4) As a result somehow of the sudden commencement, the Hall conductivity at the north-ionospheric end of the ATS-1 field line may have increased. flow
Acknowledgemenfs-We are grateful to J. W. Freeman, Jr. for originally suggestingthe problem and encouragingthe work, to W. D. Cummings, P. J. Coleman, Jr. and H. Howe for communicatingunpublished data, to G. Rostoker for pointing out an error (now corrected), to William Paulishek of World Data Center A, NOAA, for supplying the ground magnetograms, to T. W. Hill for letting us use his magnetic field program, and to W. J. Burke, A. J. Dessler and V. M. Vasyliunas for illuminating conversations. The work was supported by the National Aeronautics and Space Administration under grants NGR-44-006-100, NGL-44-006-012 and NGR-44-006-033. REFEXENCES CHAPPELL,C. R., HARRIS, K. H. and SHARP, G. W. (1971). J.geophys. Res. 76,7632. FAIRFIELD,D. H. (1971). J.geqQs. Res. 76,670O. FREEMAN,J. W., JR. (1968). J.geophys. Res. 73,415l. GCXTLIIZB, B. and FFJER, J. A. (1967). J.geophys. Res. 72,239. HXRSHBERG, J., ALKSNE,A., COLBURN,D. S., BANE, S. J. and HUNDHAUSEN,A. J. (1970). J.geophys. Res. 75,l. HIRSHBERG,J. and COLBTJRN,D. S. (1969). P&et. Space Sci. 17,1183. KENNEL, C. F. and REES, M. H. (1972). J. geophys. Res. 77,2294. MEAD, G. D. (1964). J.geophys. Res. 69,1181. MOZER, F. S. (1970). Planet. Space Sci. 18,259. SP~IIXR, J. R., SUMMERS,A. L, and ALKSNE,A. Y. (1966). Planer. Space Sci. 14,223. VASYLIUNAS,V. M. (1970). Particles and Fields in the Magnefosphere (Ed. B. M. McCormac), p. 60. Reidel, New York. YOUNG, D. T. (1969). Presented at the Rice-AGU Conference on Electric Fields in the Magnetosphere, Houston, Texas, U.S.A.
Space
Sci. 1973,
Vol. 21, pp. 1713 to 1729.
Pergmon
Pms.
Printed
in Northern
Ireland
MAGNETOSPHERIC PLASMA MOTION A SUDDEN COMMENCEMENT
DURING
C.-A. LIN, D. T. YOUNG* and R. A. WOLF Department of Space Physics and Astronomy, Rice Wniversity, Houston, Texas 77001, U.S.A. (Received 2 April 1973)
Abstract-A sudden commencement occurred at 2348 UT on 15 February 1967, when the ATS-1 satellite was about 2 hr past local noon at a geocentric distance of 6.6RB. Plasma was observed by the Sup~the~al Ion Detector (SID) first to flow in the antisolar direction, as expected, but then to flow westward, for about 2min, at about SOkm/sec. Analysis of ground magnetograms suggests that the surprising westward flow, which must have involved an electric field of about 10 mV/m at 66R,, resulted from the ionosphere’s reaction to the sudden commencement. l3eginning about 2 min before the start of the westward flow at ATS-1, ground ma~etomete~ near the foot of the ATS-1 field line typicahy recorded ma~etic-field deflections of about 70 y. to the northeast. No attempt is made in this paper to explain these ground observations. However, taking the ground observations, assuming a height-integrated Hall conductivity of 1 mho, and a standoff distance of 7*2R, inferred from Explorer 33 solar-wind data, we derive a magnetospheric electric field that agrees in magnitude and direction with that required to produce the observed flow at ATM.
At the time of the magnetic storm sudden commencement of 15 February 1967, conditions in the solar wind were being monitored by Vela 3A and Explorer 33, and ATS-1 was in the dayside ma~etosphere at 13 : 31 ma~eti~ local time. The event is unique in one respect: the electric field and the density of cold plasma were both sufficiently high that the Suprathermal Ion Detector (SID) aboard ATS-1 provided vector measurements of the The aim of this paper is to present E x B-drift velocity during the sudden commen~ment. the SID data and an interpretation of the results. The data have been presented before (Young, 1969), but never published. DATA
The Rice University Suprathermal Ion Detector has been described elsewhere (Freeman, 1968). Briefly, the detector measures energy per unit charge in 20 differential steps between 0 and 50 eV and in two integral modes of E ZI== 0 eV and E > 50 eV. Energy discrimination is accomp~shed by applying a 200 c/s square wave to a positively biased planar grid. I#y using an up-down counter which counts up during the excursion of the square wave below the bias level, and down during the excursion above the bias level a differential count is obtained. The width of the di~erential passbands changes with energy in the manner shown in Table 1. Because a complete energy spectrum requires 112 set it is obvious that a true differential spectrum can be obtained only if: (1) the spread of an energy spectrum incident at a particular enera is broader than the passband at that energy, and (2) the incident flux is nearly constant over a period st 112 sec. The satellite is oriented with its spin axis antiparallei to the Earth’s spin axis so that the field of view of the SID is symmetric about the equatorial plane. Since the angular response of the detector is Gaussian in shape with cz fi: 6.5” about the axis of symmetry, the instrument only measures directional fluxes near the equatorial plane. The satellite rotates * Present address: Physikalisches Institut der Universitiit Bern, Sidlerstr.5, 3000 Bern, Switzerland. 7
2713
1714
C.-A. LIN. D. T. YOUNG and R. A. WOLF TABLE 1. ENERGY PASSBANDSOF SID DETECTOR
Energy (eV)*
Number of steps in the interval
Passband widths (eV>
OI‘?&~5 61E,< 10 15 < E, 1_<45
10 5 4
O-5 1.0 10.0
* E. is the energy associated with the center of agiven passband. through approximately 12’ between accumulation directions of arrival can be identified to within
intervals so there is minimal overlap and about 12”. Figure 1 shows the spatial
coordinates that will be used throughout this discussion. The channeltron itself is biased 3.2 kV negative with respect to ground; electrons with energies less than 3.2 keV are thus excluded. Detection efficiency reaches a relative maximum for 4-5 keV electrons and then falls off rapidly for more energetic ones. AT&I
data
At 2348 UT, the approximate time of the beginning of the sudden commencement at CoIlege, ATS-1 was located as shown in Fig. 1. magnetometer data supplied by W. D. Cummings and P. J. Coleman, Jr. {private communication) shows that the magnetic B field at ATS-1 increased steadily from 135 y to 195 y between 2348 and 2352 UT. During this same period the SID showed a factor of 3 increase in the isotropic background flux composed of ions with E > 50 eV and electrons with E > 3.2 keV, In addition a sequence of low-energy, highly anisotropic fluxes were observed during the period when B was increasing and for about 1 min afterwards. An overall view of the event is shown in Fig. 2. Figure 3 affords a more detailed description of the magnitude and direction of ion fluxes and the associated vector magnetic fieId at the ATS. The magnitude of the H component of
FIG. 1. ImxnoN OF ATS-1 AT 2348 UT, I5 FEBRUARY1967. Thef plane of the paper is the geographic equatorial plane, and the view is down on the north pole. The coordinates are solar-geographic. The Earth’s Equator is in the xy plane. The Earth-Sun line is in the xz plane, 14-l” below the xy plane. The satellite spins clo&wise. @ is the azimuthal angle describing the detector look direction.
MAGNETOSPI3EIW.Z PLASMA MOTION DURING A SUDDEN COMMENCEMENT FEB.
IS,
I967
0 ISOTROPIC WXGROIJND l
2344
2346
2352
2%
FLU
MAWJITUDE OF TOThL MA6NEtlC
ii40
1715
2400
0004
FIELD AT AT-S.
0008
0012
0016U-f
mu. 2. iI%! AT ATS-1 (UNPUBLls~ hfAGNETOMETER DATA SUPPLSED BY w. D. Cummas AND P. J. COLEMAN, JR.). Values of the isotropic background flux in the Da& we 60 set intervals with no amaging. E > 0 eV channel were taken in other than ffow directions. The two shaded regions indicate periods during which the low energy anisotropies were observed.
the field increases steadily, while that of the equatorial component remains relativeIy small throughout. Because of the relatively large geographic northward component B,, the direction of the total B field is nearly constant through the event. The geographic equatoriaf component, B,, = B, + By, changes direction appreciably, however, during the period when flow is observed in the antisolar direction. It is clear from Fig. 3 that the direction of the low energy particle fluxes divides the event into two distinct parts, viz. flow in an antisolar direction (0, M 180’) and ffow from dusk to dawn and roughly perpendicular to the Earth-Sun line (6, M 100”). The flow directions and flux levels for these flows (1 and 2) are indicated in Fig. 4. The angular distribution for all particle fluxes throughout the event is at the limit of resolution for the detector so the flow directions are correct to ~12”. Typical angular dis~ibutions for fluxes throu~out this period are shown in Fig. 5. Comparison of the integral channel differences in the antisolar and dusk-dawn flows indicates that about 80 per cent of the incident flux was composed of positive ions with E < 50 eV for the former case with the fraction changing to 70 per cent for the latter. About the only generalization that can be made concerning the energy in the two spectra is that the dusk to dawn flow is slightly ‘harder’ than the other. This conclusion is due mainly to percentage increases in the E > 50 eV and the E = 15 + 5 eV channels. No attempt has been made to present differential spectra using this data despite the directional consistency of the fluxes over sufficiently long periods. This is due to evidence to be discussed later that temporal changes in ion number density and velocity might well have been large enough to invalidate the usual interpretation of the differential data. Further, a large portion of the fluxes occur in the higher energy channels where the passbands are 10 eV wide. Angular distribution data shows that the spread in thermal energies
C.-A. LIN, D. T, YOUNG and R. A. WOLF
ENERGY
PER UNIT CtiARGE f&‘)
I90 a0 170 g IBQ
=”
150
40
140 ~ 30 ;= % 130 20 m” 120
2348
2349
2350
Frci. 3. flETAft_ED
2351
2352
PLASMA AND ~A~N~&-~
2353
IO
2 $4 UT
DATA.
The plots of particie-ffux ~gn~tud~ represent the fIux2 the direction of maximum counting rate, minus the isotropic flux; error bars indicate 3: 1/N Poisson counting error, The magnitudes of the northward field & and equatorial field Bep are in y. The angle 8, represents the direction of the equatorial component of B, measured ciackwise from the Sun as seen by an observer north of the equatorial plane. The angle 0, gives the direction of the equatorial component of the flow velocity, measured ciocks~ise from the Sun. Error bars indicate angular spread away from the flow direction of fluxes more than 1 a above the mean. 0, w 180” represents Bow away from the Sun, while 0, @ 90” represents flow in the westward t-y) direction, 8, = Q, _t 180”, where QIis shown in Fig. 1. was ~1 eV for 15 eV fluxes, thus destroying the differential nature of the higher energy
channels. After cessation of the ~r~ct~on~ fluxes the isotropic background rate remains elevated (Fig_ 2) as does the ma~itude of B. The differential channels show fluctuations typical of >50 eV background particles. There is no consistent dir~~~on~~ty or preferred energy in these data and the flux seldom reaches l-5 x IO’ cmW2see-l sr-I. This was pretty much the state of affairs for the Iow-energy particles until 0625 UT on 16 February. At 06 25 OS UT, the 15 3: 5 eV channel indicated a flux of about 3.5 x 10’ ions cm-2 set-l sr-’ traveling in the direction shown in Fig. 4 for flow 3. Thirty set later the integral channels indicated the same flux level in the same direction with energy E < 50 eV. Other differential channels during this short interval registered fluxes of - IO7 ions crnTa see-l srl in the same direction. AIthough this third flow observation is based on a smait sample, it is highly significant statistically and there is no reason to regard the data as spurious, ~~~~~-w~~d data The sudden ~omrnen~rne~~ of 15 February 1967 was caused by an ~nterpia~~tary shock related to a class 4B solar flare that began at 1746 UT on 13 February (Hirshberg and
MAGNETOSPHERIC
PLASMA MOTION DURING
A SUDDEN COMMENCEMENT- 1717
SUN
SYNCHRONOUS ORBIT r=6.6RE
Flow
Date (1967)
UT
Energy per unit charge (eV)
1
15 Feb.
2
15 Feb.
3
16 Feb.
23 48 58 23 49 29 23 52 44 23 53 14 06 25 08 06 25 39
15 f 5 <50 15 &5 (50 15 f 5 <50
Flux (ions/cm*srsec) 3-5 6.2 24.8 2.7 3.5 3.3
x x x x x x
10’ 10’ 10’ 10’ 10’ 10’
FIG.~. SOMEBASIC CHARACTERISTICSOP THEPLOWSOBSERVED ON 15 AND 16 FEBRUARY 1967. Quoted flux magnitudes represent the flux in the direction of maximum counting rate minus the isotropic flux.
Colburn, 1969). Solar-wind data from Explorer 33, for just before the interplanetary shock and just after it, are shown in Table 2. The directions are expressed in terms of latitude and longitude in solar-ecliptic coordinates. More limited plasma data from Vela 3A yield densities that are about 0.7 times the Explorer 33 values, and flow velocities about @85 times the Explorer 33 values (Hirshberg et al., 1970). Hirshberg et al. (1970) have used the pre-shock and post-shock plasma data to estimate the direction of the outward shock normal. The result, in solar-ecliptic coordinates, is lat -59.0” f 45” li ( long 190.7” & 3.5”. In other words, the shock normal makes an angle of about 60” with the plane of the ecliptic. The shock slammed down on the magnetosphere, impacting in the region of the northern polar cusp instead of near the subsolar point. Ground magnetometer data In Tables 3 and 4, we present data from ground magnetometers located near the foot of the ATS-1 field line. Table 3 lists the positions of six stations in geomagnetic coordinates. The pre-commencement magnetic fields are also given in Table 3, where His the horizontal component, D is the declination angle (positive eastward), X (= H cos D) is the geographic horizontal northward component and Y (= H sin D) is the geographic horizontal eastward component. All these magnetometers were quite steady just prior to the sudden commencement. Table 4 lists magnetic-field deflections away from the pre-storm values, as measured from available normal and rapid-run magnetograms, where AX, A Y are the changes of the geographic horizontal northward and eastward components in y, respectively, and AX’,
1718
C.-A. LIN, D. T. YOUNG
and R. A. WOLF
.5f .25 a!/ CtiANNEt TM.YEl.EMETRY
0
35
70
AZIMUTHAL
CHANNEL
IO5 I39 174 249244 27S 314 348
ANGLE
@, t CLOCKWISE FROM SUN 1
FIG. 5. TYPICAL ANGULAR-DISTRIBWIION DATA DURING THE PERIOD OF
TABLE 2.
Density * (protonjcma) Velocity* &m/s4
Mapretiejieldf (Y)
SOLAR-WIND DATA FROM &PLGlW-t
33
Pre-shock
Post-shock
?I*=
Jlp =
1t.M
ANTISOLAR
24.0
Y( = 333 direction (solar ecliptic) coords.): lat = -5” long = 180” + 2”
0, = 450 direction (solar ecliptic): lat = -16” long = 180” + 10’
B< = 6-8 direction (solar ecliptic): lat = 45” long = 254”
B, = 17.5 direction (solar ecliptic): lat = 30” long = 245”
* Data from H. Howe, private communication. t Data from Hirshberg et al. (1970).
FLOW.
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
TABLE 3. OBSBRVATORYCOORDINATESAND PRE-STORM MAGNETIC
Position Geomagnetic coordinates 10llg lat
Observatory Mould Bay Resolute Bay Baker Lake Barrow College Meanook
FIELDS
(MLB) (RB) (BL) (BW) (CO) (ME)
Observatory
79-l” N 83.0 73.8 68.5 64.6 61.8
256.4” E 289.3 315.2 241.1 256-S 301-o
Pre-storm magnetic fields x = 1021.2 y X= 123.4~ X = 4171 y D = 25” 44 D = 28” 28.2 D = 23” 31.8’
MLB RB BL BW
co ME
Y = 2188.9 y Y = -1396y Y=27Oy
H = 9678 y H = 12868 y H = 13157.1 y
TABLE 4. GEOMAGNETICMELD CHANGES AX
UT
Mould Bay 23 47 08 48 00 49 00 49 30 5000 51 00
52 00 53 00 55 00 58 00
0 11 2: 62 42 -49 -83 -187 -231
AY
AX’
1:
18 20 21 22 -22 -56 -181 -184
Ahy’
0 18 41 156 z -47 -94 -257 -286
0 -2: -175 -38 -20 26 33 31 75
Resolute Bay 234600 48 00 49 00 50 00 51 00 52 00 53 00 55 00 58 00
9’: 85 62 0 -7 --: -92
-5: -35
0 28 59 101 118 94
0 30 35 44 20 2: 36 1
0 -100 -85 -44 20 47 76 132 131
Baker Luke 23 48 49 50 51 51 52 53 55 58
00 00 00 00 30 00 00 00 00
0 8 33 40 50 36 16 -27 8
8 6 19 30 35 -8; -84
0 8 34 44 57 45 16 -51 -18
0 -3 -4 6 13 22 -2 -70 -83
1719
C.-A.
1720
UN,
D. T. YOUNG
and R. A. WOLF
TABLE 4.-(continued) UT
AX
AY
AX’
0 -10 34 38 41 -22 -82 -85 -65 -52
0 -2 -4 -38 64 139 218 210 200 194 214 227 204 70
0 -10 27 13 68 53 43 35 41 55 56 60 127 10
0 1
2
AY’
Barrow 23 47 48 48 49 49 50 50 51 51 52 52 53 55 58
45 00 30 00 30 00 30 00 30 00 30 00 00 00
-64 -55 25 -30
0 3 -21 -52 34 130 230 224 205 193 216 223 161 75
ColIege 23 47 48 48 00 48 30 49 00 49 30 50 00 50 30 51 00 51 30 52 00 52 30 53 00 55 00 58 00 6000
Meanook 23 47 06 48 00 49 00 50 00 51 00 52 00 53 00 55 00 58 00
0 -6 -25 -6 18 24 24 -20 -30 -10 3 17 77 68 32
0
14 60 127 110
93 83 77 49
-6 -18 14 70 104 152 128 85 66 64 56 36 17
0 -1 -1 10 44
24 _y -13
-25 -13 22 51 65 45 25 27 30 42 94 76 36
0 3 5 -14 5 53 liz 129 82 59 51 19 5 2
0
0
14 58
-4 -16 -22 15 0
126 118
96 82 73 44
-11
-23 -25
A Y’ are the changes of the geomagnetic horizontal northward and eastward components in y, respectively. Note that AX and A Y increase for 2 or 3 min then decrease; also note that AX decreases after 5000 f 0100 UT. INTERPRETATION
Flow 1
The observation of antisolar flow early in the sudden-commencement event is consistent with the idea that the SC represents a general compression of the magnetosphere and thus an inward motion of the magnetopause and the plasma just inside it. The proton flow velocity corresponding to 15 eV, the center-energy of the channel showing maximum flux, is about 54 km/set, which, multiplied by a flow-duration of
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1721
3 min, suggests an average inward displacement ~15R, for plasma moving past ATS-1. The geocentric distance to the subsolar point can be estimated using the formula (Fairfield, 1971) rb = [$l”6[;;;;;:(1”“E
(I)
where m, is the proton mass, and u and n are the solar-wind number density and velocity, k is a measure of how efficiently solar particles transfer their momentum to the magnetosphere, andfrelates the geomagnetic field just inside the magnetopause to the undistorted dipole field at that point. Using the data from Table 2, we find the pre-storm standoff distance rbi and post-sudden commencement standoff distance rb, are rbi = 8-56RB,
rb, = 674R,
for
‘i = O-50;
rbi = 9*1RE,
rbf = 7*16RE for
$ = 0.72;
rbi = lO*lR,,
rbf = 8*ORB
f
for
= 1.40.
(2a)
(24
Fairfield (1971) has compared the distribution of magnetopause standoff distances for 137 boundary crossings with the distribution of standoff distances computed from Equation (1) and a set of 3146 hourly averages of nmve in the solar wind. He found that the distributions fit best for f2/k N l-40. However, the distributions did not agree for small standoff distances: 8 of the 137 observed crossings indicated rb < 8_5R,, but a negligible fraction of the hourly averages gave sufficiently large nmv2 to indicate rb < 8.5, for f2/k = l-40. This suggests that, under conditions of severe magnetospheric compressionp/k may be substantially less than 1.40. In the following calculations, we tentatively choose f2/k = O-72, keeping in mind that the actual value may be substantially larger. However, Equation (2) indicates that for 0.5
The westward flow observed in the evening sector, late in the initial phase of the storm, should presumably be interpreted as the result of enhanced convection from the tail, the convection that probably causes the main phase of a storm. Thus flow 3 does not seem particularly puzzling and will not be discussed further here. Flow 2
The 50 km/set westward flow, which persisted for about 2 min late in the sudden commencement event, cannot be explained so easily. The remainder of this paper is devoted to the analysis and interpretation of that flow. Outline of proposed interpretation
We suggest that the dramatic westward flow during the sudden commencement is a result of the reflection, from the ionosphere, of the initial earthbound sudden commencement wave. Figure 6 illustrates the suggested sequence of events. The interplanetary shock first contacts the magnetosphere at point I. Magnetospheric waves caused by that impact travel
1722
C.-A. LIN, D. T. YOUNG
and R. A. WOLF
FIG. 6. DIAGRAMILLUSTRATING SUGGESTED INTERPRETATION OF THE WESTWARD FLOW. The plane of the paper is the noon-midnight meridian plane, with the north pole up. The progress of the interplanetary shock through the magnetosheath is described by showing the shock configuration at 40 set intervals. The interplanetary shock tirst hits the magnetopause at point I.
down along field lines into the north-polar ionosphere and give rise to a system of currents that spreads out over the northern ionosphere. We have not gained any new insight into the reason for these currents having the magnitudes and directions they do. We just consider the implications of the observed currents. The existence of the currents implies the existence of electric fields in the ionosphere, electric fields that propagate, by means of AlfvCn waves, along field lines into the outer magnetosphere, causing rapid E x B drifts there. The current pattern generated in the northern ionosphere involves electric fields, which, in the early afternoon sector, have large northward components. Thus flux tubes in the topside ionosphere tend to drift westward. This westward motion of the ionospheric ends of the field lines near ATS-1 requires a corresponding westward motion in the equatorial plane. Plasma density at ATS-I To estimate wave travel times in an effort to clarify the sequence of events, we need information on the density of plasma in the outer magnetosphere. One might expect that the plasmapause would be at L > 6.6 at the time of the SC, because magnetic activity had been low for several days. The maximum value of the 3 hr Kp index for 15 February 1967 was I+, excluding the last period of the day, which included the SC. For 12, 13 and 14 February the maximum Kp indices were l+, 2 and 2- respectively. On the other hand, the fluxes measured by the SID detector indicate number densities below usual plasmasphere levels. A peak flux of 6 x 10’ ion/ems set sr, with a streaming
MAGNETOSPHEFUC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1723
velocity of 54 km/set in the equatorial plane, corresponds to a number density of the order of 2 cm-s. However, this calculation of number density from flux at ATS-1 was based on the assumption that the ions were streaming in the equatorial plane. This may not be true: for one thing, the magnetic field is about 9’ away from being normal to the equatorial plane; also, the ions may be streaming supersonically southward along the field lines, since the shock came down on top of the north-polar region of the magnetosphere. As a result, the SID detector, with viewing angle limited to within about 6.5” of the geographic equatorial plane, may never see the peak ion flux, in which case the density of 2 cm4 quoted above may be much too small. Computed wave travel times
We need to make theoretical estimates of several times: tlB, the time required for a wave to travel from the impact point I to point B, at the northern ionospheric end of the polar cusp; tIA, the time required for the compressional wave to reach ATS-1, once the wave has impacted at I; tBA, the time required for an Alfven wave to travel from the aurora1 ionosphere to ATS-1; and tAD, the time required for a compressional wave to propagate from ATS-1 to the equatorial ionosphere. We assume that B ds tlB
%
-
s2
vA
where ds is an element of length over the minimum time path from I to B, and VAis the local AlfvBn velocity. The plasma density along the path from I to B is hard to estimate. We suspect that the minimum time path is one that moves along field lines that are not part of either the plasmasphere or the polar cusp. The polar wind may not reach large geocentric distances on these field lines while they are on the dayside of the Earth, so the number density is likely to be small in the outer parts of these lines, where the wave velocity is low due to the small magnetic field. For an order-of-magnitude estimate, we therefore replace SI”d+‘, by %/VA ( near point I) and obtain, taking the magnetic field at point I to be 56 y, as suggested by Equation (10) of Mead (1964),
tlB-
16 set
nl 1I/2
[ lcmm8
’
(3)
The first compressional-wave ray to reach ATS-1 crossed the magnetopause at some point C. The time delay tIA is equal to t,, + tCA, where tCA is the travel time from C to A, and tic = t, - tI is the difference between the time the shock passes C and the time the shock passes I. To estimate tIC, we have analyzed the motion of the interplanetary shock through the magnetosheath. Figure 6 displays shock contours for the plane containing the initial shock normal and the solar-wind flow direction. The shock contours were constructed graphically, for 40 set intervals, from sound and flow velocities plotted by Spreiter et al. (1966), for y = +, M, = 8, and upstream flow velocity v, = 400 km/set. We estimate tcA to be the distance from C to A, divided by the average AlfvCn speed; in the latter, the average magnetic field was estimated to 120 y. Choosing the point C on the magnetopause in such a way as to minimize tIA, we obtain 20 set tIA
=
50 set
110 xc
fern, = 1 cmm3 for noA = 10 cm-’ for ncA = 100 cms3
(4)
1724
C.-A. LIN, D. T. YOUNG
and R. A. WOLF
where flcB. = [(n1’e>]2is the square of the average of zF2 over the ray path from C to A. Comparison of the ATS-1 data with rapid-run magneto&rams suggest that the SC arrived at ATS-1 either approximately the same time as, or less than a minute later than, its arrival at the northern aurora1 zone. This is consistent with theoretical relations (3) and (4). To estimate tBd, we assume the Alfven wave travels along an L = 6.6 dipole field line with constant number density nG.s; we obtain tBA = (8.2 sec)(ns.$2
(51 where ~t~.~is in cm-a. Assuming n*.* < 120 cm--a, the value measured at L = 6=6 by Chappell et al. (1971) on a pass that followed four days with Kp < I+, and ns+ > 2 crnd, the value suggested by ATS-1 data, we find that 12sec g tBa g 9Osec. The time required for a compressional wave to move in the equatorial ATS-I to the equatorial ionosphere is given by 6*6~?#
(61 plane from
&
tAD =
s l&t vA(eq. PI-) * We assume that the plasmasphere is completely full, and take ns
(250 cm-3)(5[L)2’”
representative of a completely filled plasmasphere. (See the dayside density profife measured on 24 October 1968 by Chappell et al., 1971.) We assume a dipole magnetic field and neglect l/va outside the plasmasphere, because of the low density there, and find that tar, m (73 set> [%]z”5
(3
where &,, = 6.6 if ATS-1 is inside the plasmasphere, and L,,, is the L-value of the plasmapause if ATS-1 is outside the plasmasphere. If the plasmapause lies at L > 5, the time required for a compressional wave to travel from 6*6R, to the Earth and reflect back to 6*6R, again is about IO0 set (within about a factor of 15 either way). This travel time is about the same as the duration of antisolar flow at ATS-1. We suggest that the flow persisted until stopped by the compressional wave reflected back from the Earth. MAPPING BETWEEN !t’HE EQUATOBIAL PLANE AND THE NORTH-POLAR IONOSPHERE
Magnetic-field-line mapping calculations were performed using a three-dipole model developed by T. W. Hill (personal communication). One dipole is that of the Earth, and the other two dipoles are parallel to that one; one has a dipole moment 125 times the Earth’s and is placed at X,, = 37~8R,; the other has a dipole moment -2 times (the minus sign means antiparallel) the Earth’s and is placed at X,, = -3OR,. With these values of the parameters, the magnetopause standoff distance is 7*16R,. This simple image-dipole model has the advantage of giving a realistically shaped ideal magnetopause with no magnetic field normal to it. Figure 7 shows the ATS-1 mapping point computed using this model (point marked ‘A7’). Also plotted for comparison is the point ‘A,’ which maps to ATS-I in a model that neglects magnetopause and tail currents. The motion of the magnetopause almost into ATS-1 causes the AT&l mapping point to move north and west, close to the point ‘P7’, which maps to the magnetopause.
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1725
FIG. 7. DIAGRAMSHOWINGTHE NORTH-POLAR IONOSPHERE AT 2348 UT ON 15 FEBRUARY. The dashed line marked ‘Sunset line 1’ is the line on which the solar-zenith angle is 90”. The solid line 2 shows the line on which the solar-zenith angle is 95-5”. The line of sight between the Sun and an observer, standing on the solid line at I20 km altitude, just grazes the surface of the Earth. The points marked A8, A7 and A6 represent the points that map to ATS-I, for standoff distances of 8*ORx, 7*16R, and 674&, respectively. ‘Am’ represents the ATS-1 mapping point, computed neglecting magnetopause currents. P8, P7 and P6 represent the points that map to the magnetopause for standoff distance 8*OR,, 7_16R, and 6*74R8 respectively. They are not on the Earth-Sun line because the solar wind is not flowing along the Earth-Sun line.
The mapping calculations have also been done for standoff distances of 6*74R, and
S-OR,.The results are indicated in Fig. 7 by points A6, P6, A8 and P8. ~~pping of the ~ono~p~criceIec~r~c~e~dto the eq~a~orialplane We interpret the westward flow in the equatorial plane as being part of a secondary event caused by the ionospheric reaction to the SC. This secondary event, unlike the primary compression, does not cause a significant net change in the magnetic field in the magnetosphere: it consists of a rearrangement of flux tubes. Consequently, we use the same magnetic-field model to analyze the entire secondary event. We can label a given field line by either of two pairs of numbers: +,, and tJi, the geomagnetic longitude and colatitude of the field line at some arbitrary Axed height in the ionosphere; or r, and +,, the geocentric distance and geomagnetic longitude of the field line, where it crosses the geomagnetic equatorial plane. In the secondary event, the ionospheric coordinates of the field line change by amounts hOi and A&., while the equatorial coordinates change as follows:
1726
C.-A.
LIN,
D. T. YOUNG
and R. A. WOLF
Attributing the displacements A$,, he,, A@,, A+, to E x B drifts, and putting E B = 0, we can use Equations (Sa) and (8b) to relate the time integrals of the secondary-event electric fields in the ionosphere and equatorial plane. Let e,, be fEre dt, the time-integr~ over the secondary event of the radial component of the electric field in the equatorial plane. Similarly, let .sde= SE,+ d t, set = jEef dt, edi = SE+, dt, where the subscripts e and i refer to the equatorial plane and ionosphere. Equations @a) and (8b) then become l
where RE is the radius of the Earth at ionospheric altitude. The square-bracketed coefficients depend on the rna~~tic-fiend model. dipole model, we have (Mozer, 1970)
For a simple
which becomes, for L = 6.6, Ezs = -00316,*
(1 Ia)
and
To get an idea of how the magnetopause affects mapping of electric fields just inside it, consider a ~ylindri~lly s~metr~c magnetosphere with a ~eld-~~~ed boundary that has equatorial radius to. Assume that the region just inside the magnetopause has magnetic field B, in the equatorial plane; this region maps to the north pole region of the ionosphere, where the magnetic field is B,,, The region within a small colatitude angle Oi of the north pole maps into the equatorial region between re and r,. If 0, and 1 - r,fr, are both much less than unity, conservation of magnetic flux then implies that W+,,RE2 ei2 = 3, r. PO - r,>, or
which, substituted into Equation (9) along with the relation dr = +*, gives err w
-+[$-J1ia[&J1’a
(1%
and
Far from the ma~etopa~~, the east-west component of the electric field is attenuated a little less in mapping to the equatorial plane than the north-south component is. But the
MAGNETOSPHERIC
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1727
situation is very different for field lines that go close to the magnetopause: the equatorial electric field normal to the magnetopause gets very large as rs + r,,, while the tangential component goes to zero. We have used the 3-image dipole model with standoff distance 7*16RE to map from 8 points in the equatorial plane near ATS-1 to the ionosphere, to get estimates of the coefficients in Equation (9). The results are &78M Eei(-o*12) + &ci(-o*0159)
(13a)
ede M e,,(O*OO38)+ ~(0.033).
(13b)
Comparison of these results with Equation (11) indicates that the radial electric field is considerably enhanced over the dipole result, while the east-west field is reduced. Both results are in qualitative agreement with the effects indicated in Equation (12). We have also derived coefficients for a standoff distance of 6*74R,; the results are in good agreement with the idea suggested by Equation (12a) that, near the magnetopause, Ebbis proportional to the inverse square root of the distance to the magnetopause. Rewriting Equation (13), neglecting the small off-diagonal terms, but taking approximate account of variations in the distance to the magnetopause, yields 0*60RE & ,d 5% %i
)I l/2 )I l/Z
d magnetopause
(14b)
is the distance from ATS-1 to the magnetopause and 0*60R, is the wheredmegnetopauae approximate distance from ATS-1 to the magnetopause for standoff distance 7*16R,. Note that the E x B drifts induced just inside the magnetopause by AlfvCn waves from the ionosphere are nearly tangential to the magnetopause. We have also derived coefficients for a standoff distance of 8*OR,. The results are &r# % &,,(--O-035) + &,(-0@031)
(15a)
E+ M e,,(0*0013) + a+~(o*O43).
(15b)
Ionospheric conductivity
As indicated in Fig. 7, the ATS-1 mapping point lies approximately on the sunset line, i.e. the line on which the solar-zenith angle is 90”. Sunlight can reach the region of the E-layer that maps to ATS-1, but only by reflection or scattering, or by traversing hundreds of kilometers path length in the R-layer. We thus expect the photoionization flux to be low. Field lines in the region are closed: heating by polar-cusp plasma (Kennel and Rees, 1972) should not be important in the immediate vicinity of the ATS-1 mapping point. Most authors assume that, at local noon at mid-latitudes, the height-integrated Hall conductivity is about 20 mho. (See Gottlieb and Fejer, 1967.) The electron number density in the dayside E-layer is roughly proportional to cos112a, where a = solar-zenith angle; we expect the height-integrated Hall conductivity to vary in a similar way, for a less than 90” and not too close to 90”. Estimates of the nightside height-integrated Hall conductivity, outside the zone of aurora1 precipitation, are typically -0.15 mho. Having no accurate way to estimate a,, the height-integrated Hall conductivity around the ATS-1 mapping point, we assume oH - 1 mho.
C.-A. LIN, D. T. YOUNG
1728
and R. A. WOLF
Ionospheric electric fields
Having no real knowledge of the 3-dimensional current system during the sudden commencement and no knowledge of the relevant inhomogeneities in Hall conductivity, it seems most reasonable (Vasyliunas, 1970) to assume that the ground magnetic variations are due to overhead Hall currents, which means that, near the north pole, the horizontal components of the ground magnetic variations are related to the horizontal ionospheric electric field by
or -2
= -1.6 mV/m
Egi = -AX’
(AX’lly) (oH/l mho)
POOH
E,i = 2AY’.
(16b)
POOH
The strong westward flow at ATS-1 began between 23 50 30 and 23 51 30 and ended about 23 53 20 UT. Taking the time-delay to be 60 set i 30 set, consistent with inequality 6, the corresponding ionospheric times should be, roughly, 23 50 00-23 52 20 UT. It is clear from Equation (13a) that the east-west ionospheric electric field has little effect on the east-west motion of plasma in the equatorial plane, so we need not worry about A Y’. Between 23 50 00 and 23 52 20 UT, AX’ is consistently positive at nearly all the nearby stations, and is of the order of 50 y (see Table 4). ElectricJelds
at AT&l
Assuming that the ionospheric and magnetospheric have, substituting Equation (16a) in Equation (14a), mV E'eN9'6~
electric fields act over 2 min, we
1
l/2
(AX’/5Oy) ' (~a/l mho)
Recall that &cg =
E,, dt s
and
aei =
(17) '
Egi dt. s
The observed equatorial electric field, i.e. the electric field required to produce the observed E x B drift velocity of about 54 km/set west in a northward magnetic field of about 190 y, is about 10 mV/m, radially outward. The electric field derived from the ionospheric currents is in the correct direction to explain the westward motion at ATS-I, and its magnitude is sufficient to produce the observed speeds. Note, however, that the derived electric field is sensitive to the assumed standoff distance, both through the factor (0.6/d)l12 and through the conductivity. (Moving the magnetopause away from ATS-I moves the north-ionospheric end of the ATS-1 field line southward, toward the sunlight.) Beginning and end of the westwardjow
Different magnetometers near the ATS-1 mapping point developed their large northward deflections at various times between 23 48 00 and 23 50 00 UT. The fact that the westward
MAGNE~SF~E~C
PLASMA
MOTION
DURING
A SUDDEN
COMMENCEMENT
1729
at ATS-1 did not develop before about 23 50 30 UT can be explained either as a result of the AIfvCn wave travel time or by the fact that for given Eet, Er, increases as the magnetopause approaches ATS-1. Thus, we would expect the westward flow at ATS-1 to become strong as the inward motion of the magnetopause neared its end, and that is what the observations indicate. Detectable directional fluxes at ATS-1 terminated abruptly about 23 53 20 UT. Within our picture, this could be due to one or a combination of the following four effects. (1) Termination of the northward electric field at the northern-ionospheric end of the ATS-1 field line. The time of the termination of the strong northward electric field varies from one auroral-zone station to another, but in a few cases it ceases between 23 51 00 and 23 53 04 UT, due to the decreasing of X, Y, H and D components. (2) Reflection from the southern ionosphere of the Alfvkn wave caused by the northward electric field in the northern aurora1 zone. Since the southern-ionosphere end of the ATS-I field line is in sunlight, the conductivity should be substantially higher there. Because the southern ionosphere will not accept large electric fields, it should reflect the AIfvCn wave. The electric field of the reflected wave will oppose that of the initial wave and will greatly retard the flow at ATS-1, once it reaches the satellite. The northward electric field at the ATS-1 mapping point should drop about 60 & 30 see after the cessation of flow at ATS-1. This model would suggest that the duration of the rapid westward flow at ATS-1 should be about 2t,,, or 24-180 set, which is consistent with the observations. (3) ATS-1 may have moved into a region where the average thermal velocity was high enough to make the streaming unobservable. (4) As a result somehow of the sudden commencement, the Hall conductivity at the north-ionospheric end of the ATS-1 field line may have increased. flow
Acknowledgemenfs-We are grateful to J. W. Freeman, Jr. for originally suggestingthe problem and encouragingthe work, to W. D. Cummings, P. J. Coleman, Jr. and H. Howe for communicatingunpublished data, to G. Rostoker for pointing out an error (now corrected), to William Paulishek of World Data Center A, NOAA, for supplying the ground magnetograms, to T. W. Hill for letting us use his magnetic field program, and to W. J. Burke, A. J. Dessler and V. M. Vasyliunas for illuminating conversations. The work was supported by the National Aeronautics and Space Administration under grants NGR-44-006-100, NGL-44-006-012 and NGR-44-006-033. REFEXENCES CHAPPELL,C. R., HARRIS, K. H. and SHARP, G. W. (1971). J.geophys. Res. 76,7632. FAIRFIELD,D. H. (1971). J.geqQs. Res. 76,670O. FREEMAN,J. W., JR. (1968). J.geophys. Res. 73,415l. GCXTLIIZB, B. and FFJER, J. A. (1967). J.geophys. Res. 72,239. HXRSHBERG, J., ALKSNE,A., COLBURN,D. S., BANE, S. J. and HUNDHAUSEN,A. J. (1970). J.geophys. Res. 75,l. HIRSHBERG,J. and COLBTJRN,D. S. (1969). P&et. Space Sci. 17,1183. KENNEL, C. F. and REES, M. H. (1972). J. geophys. Res. 77,2294. MEAD, G. D. (1964). J.geophys. Res. 69,1181. MOZER, F. S. (1970). Planet. Space Sci. 18,259. SP~IIXR, J. R., SUMMERS,A. L, and ALKSNE,A. Y. (1966). Planer. Space Sci. 14,223. VASYLIUNAS,V. M. (1970). Particles and Fields in the Magnefosphere (Ed. B. M. McCormac), p. 60. Reidel, New York. YOUNG, D. T. (1969). Presented at the Rice-AGU Conference on Electric Fields in the Magnetosphere, Houston, Texas, U.S.A.