On the Cause of Geomagnetic Storms

Originally published in: J. Geophys. Res., 79(7), 1105-1109, 1974.


Institute of Geophysics and Planetary Physics
University of California, Los Angeles, California 90024

Abstract. A study of the solar wind and interplanetary magnetic field during four geomagnetic storms shows that strong southward interplanetary magnetic fields are associated with the development of each main phase. Weak southward magnetic fields do not necessarily lead to an increase in the ring current even though such southward fields persist. Such behavior is consistent with the existence of a threshold for the initiation of strengthening of the ring current. These data also suggest that if the threshold is exceeded and the southward field attains a new constant level, Dst will decrease until some saturation value is reached. This apparent threshold for the interplanetary magnetic field and the saturation level for Dst can be explained by a ring current injection process, which is a function of the interplanetary magnetic (or electric) field, and by a ring current dissipation process, which is a function of the strength of the ring current.


Geomagnetic storms, as observed by ground-based magnetometers, commonly begin with an increase in the strength of the geomagnetic field. These storms are called sudden commencement storms (ssc), and the enhancements are associated with sudden increases in the dynamic pressure of the solar wind. For example, Burlaga and Ogilvie [1969] report that all seven ssc events that they studied were caused by hydromagnetic shocks in the solar wind. It is obvious, though, that the subsequent main-phase minimum is not a consequence of this shock passage. Burlaga and Ogilvie, for example, found five shocks in the interplanetary medium that were not followed by geomagnetic storms.

Such a result was anticipated by Piddington [1963], who noted that in a study of 346 ssc geomagnetic storms Sugiura and Chapman [1960] had found that the size of the sudden commencement was independent of the magnitude of the subsequent main-phase minimum. Thus Piddington concluded, 'Weak, moderate and great storms all result from solar winds of the same average intensity and duration. The different main-phase intensities result from differing degrees of frictional interaction, the difference perhaps depending on the interplanetary field strength and direction.' A corollary to this conclusion was noted in the same year by Hirshberg [1963]. From a study of the horizontal component of the surface field and the Ap index she concluded, 'There is evidence for enhancement of a ring current in the absence of sudden commencement activity.'

In an independent study Akasofu [1964] noted that 'neither SSC's nor SI's are, in all cases, followed by appreciable main-phase storms and auroral activity.' Further, he found at least two examples in which negative si's preceded a main-phase minimum. Thus he concluded in agreement with Piddington that, 'changes in the plasma pressure, sudden, irregular or negative, do not contribute in any obvious way to the major phase of magnetic storms.' However, Akasofu proposed that, instead of an increased frictional interaction at storm times, there was rather an additional component of the solar wind: neutral hydrogen.

The problem in determining the cause of geomagnetic storms is therefore to identify the additional component or source of friction in the solar wind during the development phase. Whereas the orientation and magnitude of the interplanetary magnetic field is an obvious candidate, it is not the only one. A. J. Hundhausen (personal communication, 1973) has pointed out that the shock that causes the ssc has no simple relationship to the compression structure behind the shock or to the properties of the driver gas. Thus it is possible that some other parameter such as solar wind temperature or composition is responsible for the main phase development.

In the present study we use the Dst index to identify and quantify geomagnetic storms, and then we examine simultaneous interplanetary data. In particular we will use the interplanetary magnetic field data, as obtained from the Explorer 33 and 35 Ames Research Center (ARC) magnetometers and the solar wind density and velocity, as measured by the Explorer 33 and 35 Massachusetts Institute of Technology (MIT) solar wind probe. We have computed the Dst index every 2.5 min from the 12 available digital mid-latitude magnetograms for every day during the spring of 1968 when the Dst index reached -40 γ or below at some time during that day. Whereas this is a rather small threshold for storm identification compared with classical storm studies, such a low threshold was necessary to provide a significant number of events. According to the study of the occurrence rate of geomagnetic storms by Russell and McPherron [1973] we would expect on the average nine storms with Dst minimums below -40 γ during the 3 months of our study. (Our study was restricted to March, April, and May 1968 by our desire to use, and by the availability of, data in computer compatible formats.) There were 11 such 'storms' in the data. However, some of these storms reflected only rather modest decreases (10-20 γ) from an already low Dst index, and others lacked some or all of the solar wind data necessary for the study. In the end there were only four events with clearly definable onsets, whether sudden or gradual, that had Dst minimums of less than -40 γ and that had the necessary correlative data for our study. Whereas the study of additional events is highly desirable and is currently being pursued, we feel that the results obtained from these four storms provide a clear indication of the principal cause of geomagnetic storms.


March 3, 1968. Figure I shows the solar wind number density and velocity, the interplanetary magnetic field strength and its solar magnetospheric Z component, as measured by the MIT and ARC instruments on Explorer 33, as well as the Dst and AE indices of geomagnetic activity for March 3, 1968. The Dst index has been computed from digital records of mid- and low-latitude magnetograms every 2.5 min. The AE index is that available for this period from the World Data Center and has a 2.5-min temporal resolution.

The solar wind density and velocity and the interplanetary magnetic field strength are all relatively constant, as is the proton temperature, which is not shown. Nevertheless, the Dst index shows that a small- to moderate-sized ring current is flowing in the magnetosphere and that late in the day a small geomagnetic storm begins. Clearly, the geomagnetic activity is associated with the occurrence of a southward interplanetary magnetic field throughout most of the day. On the other hand, a storm did not develop until after 1600 UT even though the interplanetary magnetic field was significantly and almost continually southward for the preceding 16 hours. The event that appears to signal the onset of the storm is the increase of the southward component of the magnetic field to a value exceeding that attained during the preceding 16 hours. The maximum southward component earlier during the day was 5 γ, so that it appears that there existed a threshold for storm generation of 5 γ southward at this time. For clarity, values exceeding this level have been shaded in Figure 1. We note that whereas the southward component exceeds this level, Dst experiences its greatest rate of decline and only slowly changes after this time. One might be tempted to ascribe the subsequent geomagnetic storm to the rapid rate of change of the field at 1520 UT rather than to the subsequent period of strong southward field. It seems improbable to us, however, that an event lasting several minutes would lead to energy injection over several hours.

Comparison of the bottom two panels of Figure I shows the rough correlation of the occurrence of substorms, as measured by the AE index, with the occurrence of injection of energy into the ring current, as measured by the rate of change of the Dst index. Such correlation was first noted by Davis and Parthasarathy [1967] and by Pudovkin et al. [1968]. As Davis [1969] has shown, this correlation is far from perfect. Finally, we note that the immediate response of the Dst index to the southward turning of the interplanetary field was an increase in field strength even though there was little change in solar wind dynamic pressure at this time.

Fig. 1. Solar wind and geomagnetic activity data for March 3, 1968, are (from top to bottom) the solar wind number density Nsw and the solar wind velocity Vsw, as measured by the MIT plasma probe on Explorer 33; the interplanetary field strength |B| and its solar magnetospheric Z component Bz, as measured by the Ames Research Center flux-gate magnetometer on Explorer 33; and the 2.5-min evaluations of the Dst and AE indices.

Fig. 2. Solar wind and geomagnetic activity data for March 5, 1968. See Figure 1 for details.

March 5, 1968. Figure 2 shows interplanetary and geomagnetic data for March 5 in the same format as that of Figure 1. Here the solar wind velocity and the interplanetary field strength remained constant throughout the day, and the only significant change in the solar wind proton temperatures occurred late in the day after 1800 UT. However, as opposed to that on March 3, the number density decreased at the onset of the decrease in Dst. The si- associated with this pressure decrease is not separable from decreases due to the energy injection at this time. This event is another clear indication that interplanetary shocks are not necessary for the development of a moderate-sized geomagnetic storm (Δ|Dst| > 40 γ). Again, there is evidence of a 5-γ threshold for the occurrence of ring current injection. Periods of moderately large southward magnetic fields around 0100 and 0300 UT did not lead to any significant decrease in the Dst index. After the interplanetary field turned southward more than 5 γ at 0715 UT, Dst decreased rapidly. After 1000 UT, Dst does show some signs of saturation; this finding indicates that the ring current will not build up indefinitely even though the threshold is exceeded. We will return to this point in the discussion. Finally, we note again the rather imperfect correlation of Dst and AE. Here the peak activity, as measured by AE, occurs after the injection event, as measured by Dst, is complete.

May 1-2, 1968. Figure 3 shows the interplanetary and geomagnetic data for May I and 2 in the same format as that of Figures 1 and 2. During this period Explorer 33 solar wind data were not available. Only Explorer 35 solar wind measurements unaffected by the presence of the moon are shown. Both Explorer 33 and 35 magnetic field data were available. The Explorer 33 data are shown because they are more complete. This event is a small- or moderate-sized gradual commencement storm. This period does include a large increase in solar wind dynamic pressure. There is a fourfold increase in the density beginning about 2200 UT. However, this is long after the injection event has begun. Again, the AE index and Dst index changes roughly correlate, but again the quantitative agreement is poor. For example, the largest A E index (950 γ at 1500 UT) is correlated with a small increase in Dst. Finally, we note that the solar wind protons gradually became cooler during the development phase of this event.

Fig. 3. Solar wind and geomagnetic activity data for May I and 2, 1968. See Figure 1 for details.

The correlation between the Bz component and Dst again suggests a threshold effect. However, the 5 γ level, which was appropriate before and which we have shaded as we did before, is not exceeded until 2000 UT (solid vertical line). Clearly, Dst has begun to decrease before this time, and a 5 γ threshold is clearly inappropriate here. If a threshold does apply on this day, it must be about 3 γ. Such a value, indicated by the dashed horizontal line, is first crossed at 1300 UT. The reason for this change in apparent threshold is probably twofold. First, the solar wind velocity is somewhat higher on this day than on March 3 and 5, and, second, the initial ring current strength prior to the storm is less here than on March 3 and 5. In the discussion section we will examine the reasons why such differences would affect the apparent threshold.

April 5-6, 1968. Figure 4 shows the interplanetary and geomagnetic data for April 5 and 6, 1968. The format is similar to that of Figures 1-3 with the exception that the solar magnetospheric X and Y components of the interplanetary field are included. Explorer 35 solar wind data and Explorer 33 magnetic field data are used for the same reasons given for Figure 3. This example shows an almost classical ssc geomagnetic storm. The sudden commencement is accompanied by an interplanetary shock wave at 1330 UT, as evidenced by the sharp interplanetary magnetic field jump. This shock wave occurred during an Explorer 35 data gap, but the data after the gap are consistent with a compression of the number density and a rise in velocity at a shock. The solar wind then contains a high-density plug followed by a slowly rising solar wind velocity as the density decreases. This condition is a common signature of high-velocity streams [cf. Davis, 1972; Montgomery et al., 1972]. We note that the temperature of the solar wind protons remained quite constant until the end of the development phase.

Fig. 4. Solar wind and geomagnetic activity data for April 5 and 6, 1968. See Figure 1 for details. The X and Y components of the interplanetary magnetic field are also given.

The storm again illustrates the threshold effect. The Bz component is southward at about -5 γ for over 5 hours early on April 5. However, the ring current remains quite steady. It is not until after the southward field exceeds about 5 γ at 1730 UT that the injection into the ring current commences. The X and Y components of the interplanetary field show no obvious correlation with the injection event. The AE index again shows only a weak correlation with the injection process. Although this is a much larger storm, the maximum AE index (~ 1000 γ) is similar to that observed during the three events discussed previously. Furthermore, the behavior of the AE index around 0800 UT on April 6 is quite similar to the behavior during the injection event, but the Dst signature is quite different at the two times. Finally, we note that the durations of the southward field and the injection period are not significantly different from those of the storms discussed previously. Thus the increase in intensity of this storm must be related to the larger southward interplanetary magnetic field reaching 20 γ.


The development of the main phase of each of the four geomagnetic storms studied here was clearly associated with the southward turning of the interplanetary magnetic field whether or not there was an accompanying sudden increase in the solar wind dynamic pressure. Whereas we have not examined compositional data for these events, the absence of velocity, density, or temperature changes during the March 3 and May 1-2 events implies that the composition was constant when the interplanetary field turned southward. Furthermore, during the period under study there were no large (> 10 γ) Dst decreases that were not associated with southward fields. Thus these data demonstrate that a strong southward interplanetary magnetic field is sufficient for the intensification of the ring current.

The data for the March 3, March 5, and April 5-6 storms show fairly convincing evidence that a threshold of 5 γ southward in solar magnetospheric coordinates must be exceeded before energy is injected into the ring current. However, the data for May 1 and 2 show equally convincing evidence for a lower threshold. One obvious answer to this dilemma is that the threshold is not a magnetic field threshold but an electric field threshold and that the interplanetary electric field must exceed about 2 mV m-1 westward before injection occurs. This hypothesis is reasonable, since the solar wind velocity was higher on May 1 and 2 than on the other days and because we expect the merging rate to be proportional to the flux transport rate to the magnetopause and not just to the field strength at the magnetopause. However, the difference in velocity (~ 15%) is not enough for this to account for the entire difference in threshold.

There is one further difference between the May 1-2 storm and the other three storms. This difference is in the preexisting ring current strength before the storm. We note that whereas Dst was positive at the onset of the main phase decrease on April 5, this was due at least in part to the ssc compression of the main field. Thus we believe that the Dst index before the increase in the solar wind dynamic pressure is a better indication of ring current strength. Furthermore, our observation of threshold behavior on this day is based mainly on data before the ssc. On May I the Dst before the storm was roughly - 10 γ. Before the other storms it was roughly -20 γ. Thus the data suggest that the threshold is also a function of ring current strength. Not only is this a plausible hypothesis, but it also suggests the reason for the threshold.

Davis and Parthasarathy [1967] showed that the rate of change of Dst was proportional to Dst or, equivalently, the rate of energy loss from the ring current depended on the size of the ring current. The ring current obviously cannot build up unless the rate of energy injection exceeds the loss of energy. Thus we expect a threshold dependent on ring current strength. Equivalently, we expect a saturation Dst for each value of the southward magnetic field (or more correctly, the westward electric field). If the interplanetary field remained southward long enough at some constant value, we would expect the ring current to increase in strength until the normal loss mechanisms balanced the injection rate. This may have occurred near the main-phase minimum of the March 5 storm. However, fluctuations in Dst associated with auroral zone activity at this time make such a determination difficult. Finally, since the ring current exists at quiet times [Frank, 1967], we expect that there is threshold for storm development at quiet times.

The fact that storms are caused by the occurrence of large southward interplanetary fields leads quite naturally to the association of storms with shock waves in the following way. First, the southward component of the interplanetary field associated with even a moderate-sized storm is quite large. The small- to moderate-sized storms on March 3 and 5 and May 1 and 2 were associated with southward fields of about 7 γ. The moderately large storm of April 5 and 6 was associated with a 20-γ southward field at the peak of the storm. On the other hand, the average interplanetary magnetic field strength is only about 6 γ, the most probable field strength is about 5 γ, and the magnitude of the field exceeds 7 γ less than 20% of the time [Schatten, 1972]. An occurrence of a southward field of such a magnitude would of course be much rarer, since the interplanetary field is usually near the solar equatorial plane [Coleman et al., 1969]. Thus to obtain a large southward component of the interplanetary field, it is necessary to both compress and twist the interplanetary field. Colliding streams of solar plasma should be effective in accomplishing both effects, although we would not expect every interaction to produce a southward component.


A study of the solar wind velocity and density and the interplanetary field strength and its solar magnetospheric Z component for four geomagnetic storms in the spring of 1968 shows that the occurrence of large southward interplanetary fields is associated with each main phase. A threshold effect is apparent in these data, so that if the interplanetary field is southward but is less than the threshold, the strength of the ring current does not increase even though this southward component persists. The data also show evidence of a saturation Dst for a particular sustained southward component. The fact that the threshold appears to vary with ring current strength indicates that the apparent threshold and the saturation Dst are manifestations of the same phenomenon: the dissipation of ring current energy proportional to ring current strength. The principal role of the sudden increases in solar wind dynamic pressure in the development of main-phase minimums is apparently to generate large sustained southward components. Since the solar wind velocity did not change significantly during our events or from event to event, we have presented our data in terms of the southward magnetic field. However, we expect that the interplanetary westward electric field is instead the critical parameter, i.e., the rate of transport of southward field to the magnetopause. We note that whereas the data presented here are not sufficient to quantify the relationship between interplanetary parameters and the strength of the ring current, they do indicate the form of this relationship. Namely, there is a source term that is a function of the interplanetary magnetic (or electric) field and a loss term dependent on the ring current strength, and if the interplanetary conditions remain constant, these two terms will reach an equilibrium. The data suggest that a 1.5-mV/m westward electric field maintains Dst at roughly -25 γ and a 2.5-mV/m westward electric field maintains Dst at roughly -60 γ.

Acknowledgments. We would first like to thank C. P. Sonett and D. S. Colburn for providing the Ames Research Center magnetic field data and J. H. Binsack for providing the MIT plasma probe data from both the Explorer 33 and the Explorer 35 satellites on magnetic tape so that such a correlative study would be possible. We would also like to thank C. R. Clauer for the creation of the 2.5-min values of Dst and G. Paulishak of the World Data Center A for providing the digital data from which Dst was created and for providing the AE index tapes. We have also had many useful discussions of these data with F. V. Coroniti, C. F. Kennel, and M. G. Kivelson. We also wish to thank J. Hirshberg, A. J. Hundhausen, and G. Rostoker for many useful comments on an earlier version of this manuscript. This work was supported by the National Science Foundation under NSF grant GA 34148-X. Creation of the Dst index was supported by the Office of Naval Research under contract ONR N00014-69-4016.

* * *

The Editor thanks R. L. Arnoldy, A. Nishida, and two other referees for their assistance in evaluating this report.


Akasofu, S.-I., A source of the energy for geomagnetic storms and auroras, Planet. Space Sci., 12, 801, 1964.

Burlaga, L. F., and K. W. Ogilvie, Causes of sudden commencements and sudden impulses, J. Geophys. Res., 74, 2815, 1969.

Coleman, P. J., Jr., E. J. Smith, L. Davis, Jr., and D. E. Jones, The radial dependence of the interplanetary magnetic field: 1.0-1.5 AU, J. Geophys. Res., 74, 2826, 1969.

Davis, L., Jr., The interplanetary magnetic field, Solar Wind, NASA Spec. Publ. 308, p. 93, 1972.

Davis, T. N., Temporal behavior of energy injection into the geomagnetic ring current, J. Geophys. Res., 74, 6266, 1969.

Davis, T. N., and R. Parthasarathy, The relationship between polar magnetic activity DP and growth of the geomagnetic ring current, J. Geophys. Res., 72, 5825, 1967.

Frank, L. A., On the extraterrestrial ring current during geomagnetic storms, J. Geophys. Res., 72, 3753, 1967.

Hirshberg, J., Note on ring currents in the absence of sudden commencement storms, J. Geophys. Res., 68, 6201, 1963.

Montgomery, M. D., S. J. Bame, and A. J. Hundhausen, Energy and mass content of high-speed solar wind streams, J. Geophys. Res., 77, 5432, 1972.

Piddington, J. H., Theories of the geomagnetic storm main phase, Planet. Space Sci., 11, 1277, 1963.

Pudovkin, M. I., 0. 1. Shumilov, and S. A. Zaitzeva, Polar storms and development of the DR-currents, Planet. Space Sci., 16, 891, 1968.

Russell, C. T., and R. L. McPherron, Semiannual variation of geomagnetic activity, J. Geophys. Res., 78, 92, 1973.

Schatten, K. H., Large-scale properties of the interplanetary magnetic field, Solar Wind, NASA Spec. Publ. 308, p. 65, 1972.

Sugiura, M., and S. Chapman, The average morphology of geomagnetic storms with sudden commencement, Abh. Akad. Wiss. Göttingen Math. Phys. Kl., 4, 1, 1960.

Back to CT Russell's page More On-line Resources
Back to the SSC Home Page