Inward Motion of the Magnetopause before a Substorm
Michael P. Aubry,2 Christopher T. Russell1, and Margaret G. Kivelson1
1Institute of Geophysics and Planetary Physics,
University of California, Los Angeles, 90024
2ESRO-NASA University Research Associate on leave from Groupe de Recherches Ionospheriques du CNRS, Paris
Abstract. On March 27, 1968, the UCLA magnetometer on board the inbound Ogo 5 satellite recorded an inward motion of the magnetopause by about 2 RE in two hours. It is shown that this inward motion was associated with a reversal of the vertical component of the interplanetary field from northward to southward, the solar wind momentum flux remaining constant. The inward shift did not produce any compression of the magnetospheric cavity, which implies a transfer of magnetic flux from the dayside magnetosphere to the tail: the Imp 4 satellite saw the magnetic tail field increasing at the end of this interval and later the substorm-associated collapse of this field. The substorm was also recorded on the ground. It is emphasized that the position of the magnetopause after the inward shift cannot be explained in terms of the available numerical models.
Studies of the relation between the orientation of the interplanetary magnetic field and the geomagnetic activity recorded by ground observatories have indicated that a southward-oriented interplanetary field seems to be more effective for triggering magnetic substorms and micropulsations and, in general, is correlated with a high Kp index [Fairfield and Cahill, 1966; Rostoker and Fälthammer, 1967; Schatten and Wilcox, 1967; Wilcox et al., 1967; Zelwer et al., 1967; Nishida, 1968]. There is rather general agreement that, owing to tangential stress between the magnetosphere and the solar wind, either a steady or transient transfer of magnetic flux into the tail precedes a substorm. This magnetic flux is stored, and the corresponding energy is released sporadically, producing substorms [Axford, 1965; Atkinson, 1966; Siscoe and Cummings, 1969]. If the tail magnetic flux increases before substorms, the magnetic flux on the dayside magnetosphere should decrease, and consequently the dayside magnetopause should be closer to the earth. This has been proved theoretically by Unti and Atkinson ; however, from their two-dimensional Chapman-Ferraro model, they concluded that the inward displacement of the nose associated with an increase in the tail flux was rather small. From an experimental point of view, Patel and Dessler  studied the relation between the magnetopause radius and the three-hour ap index. The results were too scattered to allow any conclusion to be drawn, but the three-hour ap index is certainly not the relevant parameter to be considered if one wants to delineate the time sequence just before a substorm. An improvement on this study was made by Meng , who used the hourly AE index as a measure of substorm activity [Davis and Sugiura, 1966]. He showed that the magnetopause was encountered closer to the earth when the AE index was large.
It seems logical to expect some change in the dayside magnetopause associated with the substorm sequence, e.g., inward motion before the substorm, and outward motion produced by the increased ring current after the substorm. Further, if the triggering of substorms is related both to the orientation of the interplanetary field and to the position of the dayside magnetopause, we expect that the magnetopause position must bear some relation to the interplanetary field orientation.
From a theoretical point of view, the study of the shape and position of the magnetopause is based on the assumption of equilibrium of the pressure on the two sides of the boundary, [Mead and Beard, 1964; Lees, 1964; Spreiter et al., 1966; Spreiter and Alksne, 1969]. The external pressure is obtained from various models of the interplanetary magnetic field and of the solar wind plasma flow, and may include elastic or inelastic collisions of the particles with the boundary. The internal pressure is supplied by the magnetic field alone; the magnetic field is the sum of the earth's field plus the field of the magnetopause surface currents, which can be calculated self-consistently, or from an image dipole. Shield  has added the effect of a ring current.
The orientation of the interplanetary field is not considered as a relevant parameter of the boundary position in these studies except in the papers of Lees  and Shield . These calculations predict an increase of the nose distance when the orientation of the magnetic field changes from horizontal (parallel to the main plasma flow) to vertical southward. This prediction disagrees with the phenomenological argument we have presented above, which, as we shall demonstrate, is borne out by our observations.
The aim of this paper is to present observations of the following time sequence: a reversal of the vertical component of the interplanetary magnetic field from northward to southward is immediately followed by a significant inward motion of the magnetopause and later by a substorm. The data presented cover an interval of less than three hours on March 27, 1968, during a sequence of multiple crossings of the magnetopause. Ogo 5 was inbound, and this allowed us to detect an important inward motion of the magnetopause. We use primarily data from the UCLA triaxial fluxgate magnetometer aboard the Ogo 5 satellite. The necessary information about the experiment is provided in the next section. The solar wind parameters measured from the MIT and Ames experiments aboard Explorer 35 were kindly made available by J. H. Binsack of MIT and C. P. Sonett and D. S. Colburn at Ames. Together these experiments demonstrate that the inward shift was not caused by an increase of the solar wind momentum flux but can be clearly associated with a reversal of the vertical component of the solar wind magnetic field. Tail magnetic-field data from Imp 4 have been kindly provided by D. H. Fairfield and N. F. Ness of GSFC. This experiment conclusively demonstrates that the configuration of the tail was altered both during the inward motion of the magnetopause and at the time of the subsequent substorm recorded at Sodankyla. The position of the boundary after the reversal, as stated, does not fit with the theoretical prediction of Lees . This suggests that corrections should be made to the models of the laminar flow interaction of the solar wind with the geomagnetic cavity.
Ogo 5 was launched on March 4, 1968, into a highly elliptic orbit with an apogee of 24.4 RE geocentric and perigee at an altitude of 300 km. The height of perigee, however, has increased at a rate of >1 RE/year, and the apogee has correspondingly decreased. Apogee initially was at 0900 LT, and, owing to the earth's orbital motion; occurred at successively earlier local times for succeeding orbits. The orbital plane was so inclined that outbound passes crossed the magnetopause well above the magnetospheric equator, whereas inbound passes crossed the magnetopause close to the magnetospheric equator.
The satellite carried an extensive set of energetic particle and magnetic- and electric-field experiments. In this paper, we will be concerned mainly with the identification of the magnetopause traversals. For this purpose, we have used principally data from the UCLA triaxial flux-gate magnetometer. This instrument is described in detail in a later article on the structure of the magnetopause during this same period [Aubry et al., 1971]. We have also examined data from the UCLA energetic electron spectrometer, the JPL solar wind experiment, and the Lockheed ion mass spectrometer to corroborate our identifications. Descriptions of these experiments are given by Kivelson et al. , Neugebauer , and Harris and Sharp .
On March 27, 1968, the inbound Ogo 5 satellite recorded multiple crossings of the magnetopause during an interval of more than two hours, from 1700 to 1915 UT. The 1-min averages of the magnetic field in the geocentric solar magnetospheric (GSM) coordinate system from 1500-2100 UT are shown in Figure 1.
|Fig. 1. Variation of the magnetic field versus universal time on March 27, 1968, as measured by the UCLA flux-gate magnetometer on Ogo 5. The GSM reference system is used; BT refers to the total field.|
In Figure 2, 4.6-sec averages of the same data, but in the reference system of the satellite, are presented for the time interval from 1719 to 1919 UT. The figure consists of three panels. Each panel contains forty minutes of data. The three components Bxs, Bys and Bzs of the magnetic field in the reference system of the satellite, as well as the total field Br appear in each panel. In this paper we will be concerned mainly with Bxs, which is the component of the magnetic field along the XS axis of the satellite reference system. For the period considered, this axis is nearly antiparallel to the ZGSM axis (the angle varies between 170o and 180o), and crossings of the magnetopause appear as reversals of Bx, which is positive in the magnetosheath and negative in the magnetosphere. In order to avoid any ambiguity, these two regions are labeled in Figure 2 for the first crossings.
Figure 3 gives the position of the satellite at the time of the observation. The first clear encounter of the boundary took place at 1700 UT (Figure 1), when the geocentric distance of the satellite was 12.81 RE (point A in Figure 3). The field amplitude was about the same on both sides of the boundary; only the horizontal component varied. The data from the UCLA energetic electron spectrometer the JPL solar wind experiment, and the Lockheed ion mass spectrometer confirm that Ogo 5 entered the magnetosphere at 1700.
|Fig. 2. Variation of the magnetic field as measured in the reference system of the satellite versus universal time from 1719 to 1919 UT on March 27, 1968. The letters B, C, and D refer to the position of the satellite as seen in Figure 3.|
|Fig. 3.Projection of the orbit of Ogo 5 in the GSM equatorial plane at the time of the observations; the extreme extrapolated positions of the boundary are shown.|
At 1730, a sequence of multiple crossings began, lasting until 1915. The structure and the oscillations of the boundary during this sequence are discussed in a later paper [Aubry et al., 1971]. We are interested here only in the average inward shift of the magnetopause. This sequence of multiple crossings can be clearly divided into two parts: before 1840 and after 1840. From 1730 to 1840 (points B and C in Figures 2 and 3), the appearance of the magnetopause crossings is consistent with a constant mean magnetopause position about which the boundary oscillated with a period of from 3.5 to 6 min. As the satellite proceeded radially inward toward this mean position, it spent successively less time in the magnetosheath and more time in the magnetosphere during each oscillation. Around 1800 UT, the times spent in each region were about equal. This we take as the mean location of the magnetopause during the interval: 11.6 RE geocentric radial distance. From 1800 to 1830, the pattern of crossings continued as Ogo 5 moved inward away from the mean position, successively spending a larger fraction of each oscillation within the magnetosphere.
From 1840 to 1916 (points C and D in Figures 2 and 3), a new pattern of crossings is evident. The general aspect of the data changes, owing to the presence of short-period oscillations (1 min) with nearly the same amplitude as the long-period ones (5 to 7 min). Moreover, the particle flux detected from 1840 to 1905 UT was extremely variable even on the magnetospheric side of the boundary. During this period, the satellite was repeatedly inside the magnetosheath. This is proven by the very high amplitude reached by the southward magnetic field: 40 nT, or even 50 nT (between 1847 UT and 1848 UT, for instance). In contrast, before 1840 the magnetic field did not completely reverse during the boundary crossings (see 1827 UT and 1838 UT, for instance), and thus it appears that from 1825 to 1840 the satellite moved back and forth between the magnetosphere and the sheet of current, but did not fully reach the magnetosheath. Consequently, this sequence of data after 1840 corresponds to a new inward motion of the average magnetopause. The last crossing of the boundary occurred at a distance of about 9.9 RE.
So far we have discussed only the position of the magnetopause; however, there was a very important change in the magnetosheath field that can be seen in Figure 1. The field in the magnetosheath was northward before 1700 UT (GSM Z component positive in Figure 1) and became southward after 1730 UT. Indeed, at each crossing after this time the field in the magnetosheath appears to have been southward (GSM Z component negative).
The data from the Ames magnetometer aboard the lunar orbiter Explorer 35 satellite were kindly made available by C. P. Sonett and D. S. Colburn. The variation of the orientation of the interplanetary field is shown in Figure 4; also shown in this figure are the positions of the four satellites from which the data used in this study were obtained: Explorer 35 at the moon, Explorer 33 in the afternoon magnetosheath, Ogo 5 at the morning magnetopause, and Imp 4 in the tail. The direction moon-earth makes an angle less than 12o with the GSM X axis, and we neglect this angle in the following argument. The velocity of the solar wind at the time of observation was 470 km/sec. Owing to the presence of fluctuations in the direction of the interplanetary field, it is difficult to give the precise time of the reversal of this field defined by the change of q from positive to negative values, but one can reasonably claim that it occurred between 1710 and 1715 UT. This reversal, convected by the solar wind, had to travel to the magnetopause. We know the average direction of the projection of the interplanetary magnetic field in the equatorial plane (Figure 4). If we use the model of Spreiter and Alksne  showing the deformation of magnetic field lines between the bow shock and the magnetopause, it appears that the reversal had to travel less than 60 RE, in the interplanetary medium before any perturbation reached the magnetopause near 0900 LT. That gives an upper limit of 14 min for the travel time, and so the field reversal should have reached the magnetopause at Ogo 5 between 1724 and 1729 UT.
On Explorer 33, at this time in the afternoon magnetosheath, the Ames magnetometer detected the reversal of the magnetosheath field between 1721 and 1727 UT. This is shown in the middle panel of Figure 5, which is the angle as measured by Explorer 33. At 1731 UT (4 to 10 min later), Ogo 5 recorded the magnetopause moving inward and measured a mainly horizontal magnetic field just outside the magnetopause. We cannot determine, however, whether the spacecraft actually entered the magnetosheath at this time or penetrated only the current sheet of the magnetopause. During the next pass into the magnetosheath, 10 min later, Ogo 5 measured a southward field. The observation of the reversal by Ogo 5 is delayed from the observation of the same event at Explorer 33 by about 7 min.
There appears undoubtedly to be a relation between the reversal of the field and the inward motion of the magnetopause. However, owing to the inaccuracy in the reversal time (blurred by magnetic-field fluctuations) and to the uncertainty about a complete crossing at 1731 UT, it is difficult to compute accurately the time constant involved in this relationship. Let us emphasize that the MIT experiment aboard Explorer 35 detected no change in the solar wind momentum flux associated with the change in the magnetic field orientation. This point will be discussed later. After 1812 UT, Explorer 35 passed behind the moon, and so we cannot check the orientation of the interplanetary magnetic field after this time, but the magnetosheath field was monitored continuously by the NASA-Ames magnetometer aboard Explorer 33. The events labeled B, C, and C1, on the curve for the angle in Figure 5 (Explorer 33 data) consistently occurred about 7 min before the events with the same labels in Figure 2 (Ogo 5 data). If we propose that the first inward motion of the boundary is a consequence of the first field reversal, we must suggest that the second inward motion and the highly varying boundary (between C and C1 in Figure 2) is a consequence of the nearly 90o southward magnetosheath field (between C and C1 in Figure 5). Thus, the observations imply that the more southward the magnetosheath field, the more inward and the more variable the magnetopause.
Let us analyze more carefully the inward shift of the boundary. The projection of the orbit of Ogo 5 in the GSM equatorial plane is shown in Figure 3. From 1700 to 1915 UT (points A and D on the graph), the distance between the spacecraft and this GSM equatorial plane varied from 1.7 to 0.04 RE. At 1700 UT the satellite detected the magnetopause at A, at a geocentric distance of 12.8 RE; the last crossing took place at D, at about 10 RE. If we assume that the general shape of the magnetopause did not change during this period, this implies an inward motion of the nose from A' to D'.
To compute A' and D', we use the relation
This is the equation of an ellipse with one focus at the earth and with an eccentricity . The angle is the sun-earth-satellite angle. Using magnetopause crossings from the orbits surrounding this orbit, we find = 0.35 gives the best fit to our data. We can then take this equation and use it to find the positions of the nose, A' and D', when the magnetopause was encountered at A and D. Doing this, we obtain distances of 11.7 and 9.5 RE. The corresponding equatorial cross sections of the boundary are drawn in Figure 3. Position D, however, does not appear to be an average position of the boundary. Taking the average position over the last sequence of crossings, CD, we get a geocentric distance of 10.4 RE and a corresponding nose distance of 9.8 RE.
Summarizing the above observations, it appears that, during about 2 hours after the reversal of the interplanetary field (from northward to southward), the magnetopause almost continuously but not, uniformly moved radially inward. The extrapolated position of the nose changed from 11.7 to 9.8 RE. The velocity of the Ogo 5 satellite normal to the boundary, owing to a fortunate coincidence, roughly matched the change of the magnetopause location until 1915.
POSSIBLE CAUSES OF THE INWARD MOTION
Almost all instruments capable of resolving the magnetopause have observed multiple magnetopause crossings. Most studies have centered on the oscillatory motion of the boundary [Heppner et al., 1967; Smith and Davis, 1970], Cummings and Coleman  have studied the nonperiodic motion, but at a disturbed time. During our interval (1700 to 1900 UT), DST was moderate (-22 to -28 ) and Kp was 3. What, then, was the cause or combination of causes of the inward motion? The solar wind momentum flux might have changed, the ring current might have decreased, and the interaction between the solar wind and the cavity might have changed, or, in other words, the effective viscosity at the boundary might have been altered. We shall consider these possible causes one at a time, assuming that they did not act simultaneously.
Change in the solar wind momentum flux. The data from the MIT experiment aboard Explorer 35 for the time of our observations were kindly made available to us by J. Binsack. The following values of the parameters of the solar wind plasma flow were measured:
|Velocity 470 km/sec|
Proton density 3 cm-3
Temperature 9.104 oK
The fluctuations of the hourly averages around these values were less than about 10% between 1200 UT and 2300 UT on March 27.
Now we can apply the formula given by Shield  relating the distance of the nose to the parameters of the solar wind.
where n is the proton density, v is the solar wind velocity, RN is the geocentric distance of the nose in earth radii, and f 2/K is the parameter of the interaction between the solar wind and the magnetospheric cavity. We obtain with f 2 = 1 (image dipole) and K = 1 (inelastic collision) Equation 1 allows us to draw the corresponding boundary (thick line in Figure 3). We note that the Mead and Beard  model would give a nose distance of 10 RE.
RN = 10.55 RE
|Fig. 4. Variation of the orientation and amplitude of the solar wind magnetic field as measured by the Ames experiment aboard Explorer 35. The meaning of the and f angles in the XYZ solar equatorial reference system is explained at the bottom right of the figure. The position in the GSM equatorial plane of the four satellites used in this study is shown.|
|Fig. 5. Variation versus universal time of the AE index, the angle of the magnetosheath field drawn from the Explorer 33 data, the total field in the tail, and the component of this field on the GSM, Z axis as measured on Imp 4. The data from Explorer 33 has been shifted by 7 min in order to take account of the solar wind convection time between Explorer 33 and Imp 4. The events labeled B, C, and C1 are to be associated with the same labels in Figures 2 and 3. E refers to the beginning of the tail contraction and F refers to the field collapse. Except for the Explorer 33 data, these curves were made available by D. H. Fairfield.|
If we want to explain the shift in the position of the boundary by a change in the momentum of the solar wind, we should find either a variation in velocity V2 - V1 such that
V2/V1=(11.7/9.8)3 = 1.7
which is ruled out by the Explorer 35 measurement, or a variation of density by a factor of 3, which could be caused by an increase of the number of protons or by change in the composition of the solar wind leading to approximately 70% protons and 30% particles. The increase in proton density is ruled out by the measurements of Explorer 35, and the alternate hypothesis is highly improbable [Robbins et al., 1970]. Thus, we can reasonably consider that this shift of about 2 RE of the magnetopause position is not produced by an increase of the solar wind momentum.
|Fig. 6. Variation versus universal time of the magnetic field at Sodankyla, the 1-hour averaged DST, and the geomagnetic planetary three-hour-range Kp indices. The vertical dashed lines correspond to the time of the E and F events in Figure 5. The quiet-day curves for the H and D components at Sodankyla are shown.|
Decrease in the ring current. The shift could have been caused by a decrease of the ring current; such a decrease should correspond to an algebraic increase of the DST index This DST is plotted in Figure 6, and a small increase of 7 nT is observed after 1700 UT. Such a small variation in DST may not be significant, because it is based on only a few well-spaced equatorial stations. However, we shall show in the next section that if this increase is physically meaningful it must correspond to a decrease in the effective ring current; in any case, we must determine whether such a change in the ring current could cause any significant variation in the magnetopause position. Shield  computed the equatorial magnetic field produced by the quiet-time ring current and obtained 40 nT at the ground, and so the 7 nT change corresponds to a less than 20% decrease of the ring current. Also, from the values given by Shield [1969, Table 2] it is possible to determine that the total disappearance of the ring current would produce a relative decrease in the nose distance of 10%. The 20% decrease in the ring current can therefore only account for a 2% relative decrease of this nose distance, i.e., about 0.2 RE and cannot explain the observed 1.9 RE inward shift.
Change in the interaction between the solar wind and the magnetospheric cavity. We can try to relate the boundary displacement to a change of normal pressure associated with a change of the parameter of interaction f 2/K in (2). From this relation, knowing the density and velocity of the solar wind as well as the distance of the nose, one can compute the value of f 2/K. This gives
2/K = 1.9 for RN = 11.7 RE
f 2/K = 0.6 for RN = 9.8 RE
We cannot discuss the first value of 1.9, because at 1700 UT the interplanetary magnetic field was northward and so had a component parallel to the geomagnetic field; no theoretical values of f 2/K are available in this case. However, from the numbers given by Shield [1969, Table 2] for other cases, such a value of f 2/K is not unreasonable. On the contrary, the other extreme value, f 2/K = 0.6, obtained in the anti-parallel case, is definitely outside the range of theoretical expectation, 1.5 to 3.5 [Shield, 1969]. Therefore, it seems unlikely that the inward shift could be related to a change in the laminar flow pressure of the solar wind on the boundary.
The reversal of the magnetosheath field just before 1730 UT could have another consequence: it could produce a reconnection of the interplanetary and geomagnetic fields and consequently an increase in the drag due to a normal component of the magnetic field to the boundary [Levy et al., 1964; Sonnerup and Cahill, 1967]. We shall show in the forthcoming paper [Aubry et al., 1971] that, although no steady reconnection occurred at the boundary near the satellite, the extremely variable and nonsteady tangential discontinuity with transient normal components was observed. The observation above that the more southward the magnetosheath field, the more inward and the more variable the boundary gives support to explanation of the inward motion in terms of increased drag.
CONSEQUENCES OF THE INWARD MOTION
We saw above that the magnetopause moved radially inward from 1700 UT until at least 1915 UT. This amounted to a change of nose position of 11.7 to 9.8 RE. From the preceding section, it appears that the most probable explanation of this was a change in the nature of the interaction between the solar wind and the magnetosphere, presumably from a condition of laminar flow to some other state. The cause of this change was the appearance of a southward component in the external field. In this section we will investigate the effects of this motion on the magnetosphere.
We can determine first whether the shift produced any compression of the magnetic field inside the magnetopause. Such a compression should be associated with an increase in the surface currents on the magnetopause, and so, if the other magnetospheric parameters remain constant, they should produce an increase in the ground magnetic field and a larger increase in the magnetic field just inside the boundary.
To calculate the ground magnetic field, we use the formula given by Mead  relating the variation B1 of the equatorial ground magnetic field to the variation of the geocentric distance RE of the nose between 11.7 and 9.8 RE
If the shift produced a compression of the magnetospheric cavity, the horizontal component of the magnetic field at the equatorial stations should have increased by +11 nT between 1700 and 1900 UT; this would appear as a variation of 11 nT in the DST during this period of time. Figure 5 does show an increase of 7 nT in the hourly average after 1700 UT, as was discussed above.
To calculate the magnetic field B2, just inside the magnetopause due to the surface currents, we again use Mead's model [Mead, 1964, equation 10']. For 0900 UT in the equatorial plane just inside the boundary, assuming that this boundary is defined by equation 1, B2, can be written
B2 (nT) = 41,250/RN3
For RN = 11.7, we obtain 26 nT, and, for RN = 9.8, we obtain 44 nT, i.e., a variation of 18 nT between 1700 and 1900 UT Figure 1 shows that the difference between the total field inside the magnetosphere and the dipole field between 1700 and 1900 UT remains roughly constant with a value between 20 and 30 nT, and B2 is about zero.
Observing B1 (7 nT) on the ground not associated with B2 at the magnetopause implies that the variation of the ground magnetic field was not due to a compression of the magnetospheric cavity. Thus, if this change is real, a decrease of the ring current must have been responsible for it. This justifies a posteriori our discussion of the decrease in the ring current.
If there is no compression of the magnetic field inside the cavity, the inward shift of the boundary implies a transport of magnetic flux from the front part of the magnetosphere to the tail. A very crude estimate of this transport can be made. Assuming that the equatorial sections of the boundary are semicircular with initial and final radii of 11.7 and 9.8 RE and that the magnetic field is vertical with a 50 nT amplitude, this represents a flux of about 108 Webers transported in two hours, which implies a flux rate of the order of 104 Wb/s. That is comparable to the flux rates assumed by Atkinson , namely, 103 to 105 Wb/s carried into the tail before substorms. The flux in a tail of 20 RE radius with a 20 nT magnetic field is about 109 Wb. Therefore the shift of the boundary should produce an increase of the tail magnetic flux of about 15%.
To determine whether this increase had any consequences in the tail magnetic flux, we looked at the data from the NASA Goddard magnetometer aboard Imp 4 (Figure 5) and at ground magnetograms. Imp 4 at this time was approximately 24 RE behind the earth, 10 RE toward dusk from the noon-midnight meridian, and within 3 RE of the GSM equator. On the night side of the earth, Sodankyla was the only nearly auroral station available (Figure 6). Its geomagnetic latitude is 63.8o N, and, between 1700 UT and 2000 UT, the corresponding local time interval was 1900 to 2200 LT.
From the Imp 4 data in Figure 5, the typical signature of a substorm near the center of the tail as reported by Fairfield and Ness  is apparent: an increase of the total field BT associated with a decrease of the Bz component (starting after 1840 UT, point E) and then a sudden collapse of BT associated with a recovery of Bz (at 1937 UT, point F). The substorm is seen on the Sodankyla magnetogram also. Both the magnetogram from Sodankyla and the DST in Figure 6 suggest that there was a small substorm at about 1600 UT, which caused a positive bay at Sodankyla and an increase in the ring current. (The net effect of substorms to increase the ring current has been shown by Davis and Parthasarathy  and Davis ). After this substorm, D and H returned to their quiet-time curves from about 1700 to 1800 UT. At about 1800 UT, both D and H began to deviate again from the quiet-day curve, and at 1940 UT the expansion phase of a new substorm began. Such a perturbation beginning at 1800 UT and preceding this isolated new substorm has been termed the 'growth phase' by McPherron , who has suggested that it corresponds to enhanced magnetospheric convection. From only one magnetogram, we cannot unambiguously identify the phenomenon observed at 1800 UT as the growth phase of the new substorm, but we note that there is only about a 30-min delay between the first inward shift of the day-side magnetopause and the first perturbation of the ground magnetogram. The tail data neither contradict nor confirm the identification of a growth phase on the Sodankyla measurement. However, we can say that the tail data, the Sodankyla magnetogram, and the DST index agree that a substorm expansion phase occurred. We also note that the AE index, which did not include the Sodankyla data, increased slightly after the reversal of the magnetosheath field but was not strongly enhanced. We could not have identified this substorm by using the AE index alone.
DISCUSSION AND CONCLUSION
Much work has justifiably been done on correlating interplanetary parameters with magnetospheric indices of geomagnetic activity. Unfortunately, the geomagnetic indices generally used (Kp or ap) respond to several kinds of geomagnetic phenomena, worldwide events or bay events, for instance, and the various parameters characterizing the solar wind are intercorrelated themselves, so it is very difficult to determine the specific consequence in the magnetosphere of the change of only one solar wind parameter. Recently Hirshberg and Colburn  have examined this question in depth, using both new observations and previous results. They confirmed the high correlation of a southward interplanetary field with geomagnetic disturbances, and found that the highest correlation occurred when the GSM coordinate system was used to describe the interplanetary field. As a hypothesis regarding the relation between the solar wind parameter and the magnetosphere, they suggested, first, that worldwide geomagnetic fluctuations would be associated with changes in the solar wind momentum, this association being independent of the field orientation and, second, that bay events would be associated with southward magnetic fields.
Our results in this one event confirm their second point: that the existence of a southward component of the interplanetary field leads to a substorm, or bay, through the erosion of dayside magnetospheric flux resulting in increased total magnetic flux in the tail. It is presumed that the later release of this flux, which represents a storage of energy, provides the energy subsequently deposited in the magnetosphere during the substorm. Thus the correlation between Kp and the southward component of the interplanetary field arises through the magnetic effects of substorms. This, of course, agrees with the study of Rostoker and Fälthammar . On the other hand, this in no way implies that other solar wind parameters such as transverse fluctuations [Ballif et al., 1967] do not directly affect Kp, and in fact our results say nothing about worldwide fluctuations.
In regard to the time constant involved, Hirshberg and Colburn  have shown for a particular geomagnetic storm that the main phase followed the occurrence of a southward component of the interplanetary field within less than an hour. This main phase corresponds to the increase in the ring current after substorms [Davis and Parthasarathy, 1967], and the only difference between a classical substorm as observed by us and the sequence of substorms that lead to the main phase as observed by Hirshberg and Colburn  may be simply the state of the solar wind at the time of the occurrence of the southward component. In our example, the increase in the ring current as measured by the DST occurred 1.5 hours after the reversal.
Finally, regarding the absence of change in the solar wind momentum, our results agree with those of Gosling et al. , who have shown that there was no change in solar wind momentum flux in association with the development of the April 17-18, 1965, storm. Unfortunately they had no magnetic field data. The observations of Meng  show that the dayside magnetopause is systematically closer to earth when a substorm is in progress. Thus he states that substorms occur when the magnetosphere is in a 'compressed condition. However, it has been shown by Fairfield and Cahill  that the high-latitude disturbances are associated with a southward magnetosheath field. The magnetic field orientation does not influence the solar wind momentum, and so cannot compress the dayside magnetosphere. Consequently, we suggest that, although compression of the magnetospheric cavity produced by change of the solar wind momentum is probably responsible for worldwide events, as has been discussed by Hirshberg and Colburn , it is erosion of the dayside magnetosphere associated with southward interplanetary field as observed on March 27, 1968, that often accompanies substorms. The fact that erosion produced by tangential stresses can have such a large effect on the position of the magnetopause is important if one attempts to model the boundary currents to predict magnetospheric fields. Available numerical models involve only normal stresses on the boundary.
Since the flux eroded from the day-side magnetosphere must appear in the tail, the question arises as to how this result agrees with previous measurements of the tail field. Feldman et al.  have shown that, far down the tail, the field strength is determined by the thermal pressure of the solar wind. Thus changing the total flux in the tail would merely change the tail radius at these distances. However, if we do increase the radius of the tail, while the solar wind remains constant, we expect an increase in the tail field strength near the earth, since we produce more flaring of the boundary in this region. Such a change in the tail field strength near the earth has been observed [Lazarus et al., 1968; Fairfield and Ness, 1970].
The data from Explorer 33 and Imp 4 can be examined for evidence of a relation between southward reversals of the magnetosheath field and the increase and subsequent collapse of the tail field. In Figure 5, the Explorer 33 data has been shifted by 7 min to take account of the convection time of any irregularity at the solar wind speed (470 km/sec) between Explorer 33 and Imp 4 (the distance is roughly 30 RE, as can be seen in Figure 4). The fact that the data from Explorer 33 require the same time displacement to be compared with Ogo 5 and with Imp 4 data is obviously a coincidence. It appears that the beginning of the field increase (point E in Figure 5) and the field collapse (point F in Figure 5) coincide in the shifted time reference system with sharp southward reversals of the magnetosheath field. We are aware that these coincidences may be fortuitous. Furthermore, the compression of the field in the tail observed by Imp 4 beginning about 1840 may not be due to the effect of increased flaring as flux is added to the tail. Rather, since the Z GSM position of Imp 4 is about - 1.0 RE at this time, it is probable that it was imbedded in the plasma sheet before 1840 and that this field increase represents a thinning of the plasma sheet. Let us note that sharp reversals of the magnetosheath field have at least two consequences: first, they increase the variability of the boundary and so the interaction of the solar wind with the magnetopause (as observed in Figure 2 after 1842 UT), and second, they trigger strong electric impulses (arising from the sudden change in sign of VxB) through the tail. We suggest that such electric field impulses can be very efficient at triggering changes (a field collapse, for instance) in a nonequilibrium tail field configuration.
Our measurements of the substorm sequence are in agreement with previous observations. What we have added to the picture of the solar wind-magnetosphere interaction is the observation of the erosion of flux from the dayside magnetosphere after the reversal of the interplanetary field, and this allowed us to present the first complete observation of the pre-substorm sequence leading from the field reversal to the ground-observed expansion phase through the erosion of the dayside magnetosphere, the increase and then the collapse of the tail field.
Some specific observations are worth emphasizing. The variability of the boundary and the associated erosion of the dayside magnetosphere increased when the magnetosheath field turned more southward: a sharp reversal of the magnetosheath field could be one of the triggering mechanisms of the tail field collapse. All these observations should be repeated in order to determine whether this behavior is typical. However our results definitely show that one cannot simply infer the instantaneous position of the magnetopause from a knowledge of the solar wind momentum flux, and that within two hours the position of the magnetopause can vary significantly under apparently quiet solar wind conditions.
Acknowledgments. The co-investigators responsible for the UCLA fluxgate magnetometer are P. J. Coleman, Jr., T. A. Farley, and D. L. Judge. This study used the results of seven other experiments on board four satellites; we are very grateful to Drs. C. P. Sonett and D. S. Colburn for providing us with the Explorer 33 and 35 magnetic field observations, to Drs. D. H. Fairfield and N. F. Ness for the Imp 4 magnetic field data, which were spontaneously provided to us when they read a preliminary form of this paper, and to Dr. J. Binsack for the Explorer 35 plasma data. Also, we wish to thank Dr. S. L. Ossakow and G. Sharp for the data from the Lockheed ion spectrometer and Dr. M. Neugebauer for the data from the JPL solar wind experiment on board Ogo 5.
Helpful discussions with P. J. Coleman, Jr., F. Coroniti, R. Holzer, C. Kennel, R. McPherron, and G. Siscoe are gratefully acknowledged.
This work was supported by the National Aeronautics and Space Administration under NASA contract NAS5-9098. One of us (M.P.A.) also received support from ESRO and NASA. The Editor wishes to thank D. H. Fairfield and G. Rostoker for their assistance in evaluating this paper.
Akasofu, S. L, Polar and Magnetospheric Substorms, D. Reidel, New York, 1968.
Atkinson, G., A theory of polar substorm, J. Geophys. Res., 71 (21), 5157-5164, 1966.
Aubry, M. P., M. G. Kivelson, and C. T. Russell, The motion and structure of the magnetopause, to be published in J. Geophys. Res., 1971.
Axford, W. I., Magnetic storm effects associated with the tail of the magnetosphere, paper presented at the ESRO Conference on the Magnetosphere, Stockholm, November 15-18,1965.
Ballif, J. R., D. E. Jones, P. J. Coleman, Jr., L. Davis, Jr., and E. J. Smith, Transverse fluctuations in the interplanetary magnetic field: A requisite for geomagnetic variability, J. Geophys. Res., 72(17), 4357-4364, 1967.
Cummings, W. D., and P. J. Coleman, Jr., Magnetic fields in the magnetopause and vicinity at synchronous altitude, J. Geophys. Res., 73(17), 5699-5718, 1968.
Davis, T. N., Temporal behavior of energy injected into the geomagnetic ring burrent, J. Geophys. Res., 74, 6266, 1969.
Davis, T. N., and M. Sugiura, Auroral electrojet activity index AE and its universal time variations, J. Geophys. Res., 71(3), 785-801, 1966.
Davis, T. N., and R. Parthasarathy, The relationship between polar magnetic activity DP and growth of the geomagnetic ring current, J. Geophys. Res., 72(23), 5825-5836, 1967.
Fairfield, D. H., and L. J. Cahill,, Transition region magnetic field and polar magnetic disturbances, J. Geophys. Res., 71(l), 155-169,1966.
Fairfield, D. H., and N. F. Ness, Configuration of the geomagnetic tail during substorms, preprint X-692-70164, NASA, Goddard Space Flight Center, Greenbelt, Md., 1970.
Feldman, W. C., J. D. Mihalov, C. P. Sonett, and J. H. Wolfe, Pressure balance across the distant magnetopause, preprint, NASA, Ames Research Center, Moffett Field, Calif., 1970.
Gosling, J. T., J. R. Asbridge, S. J. Bame, A. J. Hundhausen, and I. B. Strong, Measurements, the interplanetary solar wind during the large geomagnetic storm of April 17-18, 1965, J. Geophys, Res., 72(7), 1813-1821, 1967.
Harris, K. K., and G. W. Sharp, Ogo 5 ion spectrometer, IEEE Tram. Geosci. Electron., GE-7 (2), 93-98, 1969.
Heppner, J. P., M. Sugiura, T. L. Skillman, B. G. Ledley, and M. Campbell, Ogo A magnetic field observations, J. Geophys. Res., 72(21), 5417-5471, 1967.
Hirshberg, J., and D. S. Colburn, Interplanetary field and geomagnetic variations: A unified view, Planet. Space Sci., 17, 1183-1206, 1969.
Kivelson, M. G., M. P. Aubry, and T. A. Farley, Electron flux asymmetries observed by back to back detectors on Ogo 5, to be submitted to J. Geophys. Res., 1971.
Lazarus, A. J., G. L. Siscoe, and N. F. Ness, Plasma and magnetic field observations during the magnetosphere passage of Pioneer 7, J. Geophys. Res., 73(7), 2399-2410, 1968.
Lees, L., Interaction between the solar plasma wind and the geomagnetic cavity, AIAA J., 2(9), 1576-1582, 1964.
Levy, R. H.,.H. E. Petschek, and G. L. Siscoe, Aerodynamic aspects of the magnetospheric flow, AIAA J., 2, 2065, 1964.
McPherron, R. L., The growth phase of magnetospheric substorms, publication 742, Inst. Geophys. Planet. Phys., University of California, Los Angeles, 1969.
Mead, G. D., and D. B. Beard, Shape of the geomagnetic field solar wind boundary, J. Geophys. Res., 69(7), 1169-1179,1964.
Mead, G. D., Deformation of the geomagnetic field by the solar wind, J. Geophys. Res., 69(7), 1181-1195, 1964.
Meng, C. I., Variation of the magnetopause position with substorm activity, J. Geophys. Res., 75(16), 3252-3254, 1970.
Neugebauer, M., Initial deceleration of solar wind positive ions in the earth's bow shock, J. Geophys. Res., 75 (4), 717-733, 1970.
Nishida, A., Coherence of geomagnetic DP-2 fluctuations with interplanetary magnetic variations, J. Geophys. Res., 73 (17), 5549, 1968.
Patel, V. L., and A. J. Dessler, Geomagnetic activity and size of magnetospheric cavity, J. Geophys. Res., 71 (7), 1940-1942, 1966.
Patel, V. L., L. J. Cahill, Jr., and A. J. Dessler, Magnetosheath field, geomagnetic index a, and stability of magnetopause, J. Geophys. Res., 72(l),426-430,1967.
Robbins, D. V, A. J. Hundhausen, and S. J. Bame, Helium in the solar wind, J. Geophys. Res., 76(7), 1178-1187, 1070.
Rostoker, G., and C.-G. Falthammar, Relationship between changes in the interplanetary magnetic field and variations in the magnetic field at the Earth surface, J. Geophys. Res., 72(23), 5853-5864, 1967.
Schatten, K. H., and J. M. Wilcox, Response of the geomagnetic activity index to the interplanetary magnetic field, J. Geophys. Res., 72(21), 5185-5192; 1967.
Shield, M. A., Pressure balance between solar wind and magnetosphere, J. Geophys. Res., 76(5), 1275-1286,1969.
Siscoe, G. L., and W. D. Cummings, On the cause of geomagnetic bays, Planet. Space Sci., 17, 1795,1969.
Smith, E. J., and Leverett Davis, Jr., Magnetic measurements in the earth's magnetosphere and magnetosheath: Mariner 5, J. Geophys. Res., 75(7), 1233-1245, 1970.
Sonnerup, B. V. 0., and L. J. Cahill, Jr., Magnetopause structure and attitude from Explorer 12 observations, J. Geophys. Res., 72(l), 171-183, 1967.
Spreiter, J. R., and A. Y. Alksne, Plasma flow around the magnetosphere, Rev. Geophys., 7 (1 and 2), 11-50, 1969.
Spreiter, J. R., A. L. Summers, and A. Y. Alksne, Hydromagnetic flow around the magnetosphere, Planet. Space Sci., 14, 223-253, 1966.
Unti, T., and G. Atkinson, Two dimensional Chapman-Ferraro problem with neutral sheet, 1, The boundary, J. Geophys. Res., 73(23), 7319-7327, 1968.
Wilcox, J. M., K. H. Schatten, and N. F. Ness, Influence of interplanetary magnetic field and plasma on geomagnetic activity during quiet sun conditions, J. Geophys. Res., 72(l), 19-26, 1967.
Zelwer, R., P. C. B. Fernando, and S. H. Ward, Interplanetary magnetic field data and corresponding geomagnetic effects for the storm of October 7, 1962, J. Geophys. Res., 72(13), 3471-3482, 1967.
(Received June 12, 1970; revised September 4, 1970)
Copyright 1970 by the American Geophysical Union.