Site Map
Quest. & Ans.
For Teachers
Review (1)
Review (2)
Central link


A Brief History of Magnetospheric
Physics During the Space Age

Reviews of Geophysics, 34, 1-31, 1996
David P. Stern, Laboratory for Extraterrestrial Physics
Goddard Space Flight Center, Greenbelt, MD 20771

Brief History: Substorms

Table of Contents

Clicking on any marked section on the list below brings up a file containing it and all unmarked sections immediately following it on the list. This list is repeated at the beginning of each file.

  1. Introduction
  2. Discovery of the Radiation Belts
  3. Artificial Belts and Early Studies
  4. The Large Scale Structure
  5. Convection
  6. Reconnection
  7. The Open Magnetosphere
  8. Observational Tests
  9. The Polar Aurora
  10. Field Aligned Voltage Drops
  11. Birkeland Currents
  12. Substorms: Early Observations
  13. Substorms: The Satellite Era
  14. Substorms: Theory
  15. Convection in the Geotail
  16. Planetary Magnetospheres
  17. Other Areas
  18. Assessment
References: A-G
References: H-P
References: Q-Z
Back to "Exploration"


12. Substorms: early observations

Following a great auroral display in Connecticut, on July 1, 1837, E.C. Herrick [1838, cited by Siscoe, 1980] wrote (italics in the original):

"It is worthy of notice that on this occasion there were two well marked and distinct seasons of greatest brilliance or fits of maximum intensity, at intervals of about four hours. It will be found on examination of former accounts, that this is a common feature of Auroral exhibitions of unusual brilliance."
Birkeland [1908, 1913; Bostrom, 1968;] observed conspicuous magnetic signatures of such "fits of maximum intensity" and proposed that here was a new type of magnetic activity, an "polar elementary magnetic storm" with a typical time scale of half an hour. More about Birkeland's pioneering work is given in BH-1 and its references, and also in the work of Stern [1991], from which parts of this section are taken.

Birkeland's polar storms are now known as (magnetic or magnetospheric) substorms. These violent twitches of the Earth's magnetic tail energize ions and electrons, inject some of them into the ring current, and greatly increase the rate at which energy is released in the magnetosphere. Indeed, many parallels exist between substorms and impulsive particle acceleration events on the Sun, and both are believed to be powered by the conversion of magnetic field energy. For all these reasons the substorm may be the most interesting problem in magnetospheric physics and a great challenge to both observer and theorist.

As noted in BH-1, Birkeland's ideas were opposed by Sidney Chapman [Chapman and Bartels, 1940; Fukushima, 1994, sect. 6]. Chapman contrasted the short time scale of polar disturbances with the much longer one of global magnetic storms and proposed that "polar storms" were merely phases of the global storm. Around the middle of the century such events were called "magnetic bays" because on a magnetogram (the plotted output of a magnetic observatory) they resembled bays on a coastline (according to Chapman and Bartels [1940], this term is due to Chree [1911,1912]). A study of bays was conducted by Silsbee and Vestine [1942; also Fig. 2 , Stern, 1977] who deduced a 2-cell pattern with strong electrojets near the boundary of the polar cap.

While Birkeland believed that auroral currents originated in distant space, the consensus in mid-century was that they resembled the well-known diurnal magnetic variation whose currents were attributed to tidal dynamo effects in the ionosphere, and that they were completely contained inside the ionosphere [Vestine and Chapman, 1938]. It was thought that the large polar magnetic variations arose by a similar process, but were much more intense because ionospheric conductivity was enhanced in regions bombarded by the aurora. Thus Harang [1951; p. 94] wrote, after explaining the tidal dynamo:

"The intrusion of electrically charged particles which produce the aurorae, strongly increases the ionisation and thus the conductivity of the ionised layers. Besides this, one must also assume secondary effects, such as expansion or heating of the upper atmosphere, which may increase the movements of the layers along the auroral zone. The polar storms are therefore, according to these views, due to an increase in the conductivity and velocity of movements of the upper layers."
Chapman became more involved with the aurora after 1951, when he accepted a visiting professorship at the University of Alaska at Fairbanks, and after he retired from Oxford in 1953 he used to stay in Fairbanks several months each year [Akasofu, 1970]. In 1958 he was joined there by Syun-Ichi Akasofu, a young Japanese who became his main associate.

The word "substorm" first appeared in Chapman's writings in 1961 [Akasofu and Chapman, 1961, p. 1339], referring to a bay-like disturbance assumed to be a phase of the global magnetic storm. Two years later Akasofu and Chapman [1963] compared the signatures of magnetic storms near the equator and in the auroral zone (Figure 14). Near the equator the magnetic field variation was simple and familiar, a gradual weakening of the field on a typical time scale of 6 hours, followed by a slow recovery. In the auroral zone, on the other hand, the magnetic record was punctuated by many short but intense magnetic bays, which the authors again named substorms.

Figure 14

13. Substorms: the Satellite Era

The International Geophysical Year (IGY) 1957-8 (actually extended to a year and a half) brought not only the first scientific satellites but also a great expansion in auroral observations and widespread use of "all-sky cameras" [see Eather, 1980] which photographed the entire sky as reflected in a convex mirror. Using such records, Akasofu noted in 1964 that magnetic bays, which also occurred widely outside magnetic storms, were associated with a distinct pattern of auroral intensification and expansion, and proposed to name the phenomenon "auroral activation" [Akasofu, 1970]. Chapman however insisted on "auroral substorm" and that was the name used in the article's title [Akasofu, 1964]. Later Akasofu favored "magnetospheric substorm" [Akasofu, 1977; Siscoe, 1980], while Rostoker [1972a has used "polar magnetic substorm"; the commonly used term nowadays is "magnetic substorm" or simply "substorm." Today's view is that the "substorms" which Chapman identified inside magnetic storms are of a similar nature, except perhaps bigger and more frequent, capable of injecting appreciable numbers of ions and electrons into long-lived orbits of the ring current region [Lui et al., 1987].

During 1964-6 Akasofu and his group studied the morphology of substorms in great detail [Akasofu et al., 1964, 1965a,b,c; 1966a,b,c,d; Akasofu, 1966]. Their "classical" substorm phases (individual storms may vary) are still accepted: an initial brightening of a quiet arc, the expansion of the aurora polewards (either by the motion of existing arcs or by formation of new ones), a westward surge along the auroral oval, gradual breaking-up of arcs and final recovery.

A deeper understanding was gained after about 1965, when satellites began observing the great changes accompanying substorms in the Earth's magnetic tail. They observed magnetic field lines becoming stretched prior to substorm onset (Figure 15, from Fairfield and Ness [1970]) and then rebounding to more dipole-like shapes ("dipolarizing"). This was also noted by Heppner [1967] who wrote (p. 184):

"The view that is favored is that the tail field is partially collapsing back towards a less stressed condition during a negative bay."

Figure 15

The preliminary stretching may be observed as close as synchronous orbit, at 6.6 RE [e.g., Baker et al., 1981] and may begin as long as one hour before substorm onset: the prevalent idea (originally widely debated) is that this is the "growth phase" during which magnetic energy is stored [McPherron, 1970, 1972], and often an analogy is drawn between the stretching of field lines in this phase and the stretching of a slingshot. When the field lines rebound, the stored energy goes to accelerate particles and to drive currents. A large increase in the flux of ions and electrons is then observed at the nightside near the inner edge of the cross-tail current, e.g. in the synchronous orbit at 6.6 RE [De Forest and McIlwain, 1971], but what exactly goes on is still not understood [Mauk and Meng, 1987]. Typical injected energies are 1-10 keV, but particles up to 100 keV and more have also been observed [Nagai et al., 1983].

Satellites in the plasma sheet may observe disappearances ("drop-outs") of the plasma at the time of onset, and after onset a satellite which had been outside the plasma sheet may be suddenly engulfed by it. Early observations of such changes [Hones et al., 1967, 1970; Hones, 1979] were complicated by the fact that they were performed by "piggyback" instruments aboard the Vela satellites which carried no magnetometers, since their primary mission was to detect violations of the ban on nuclear tests in space. The ultimate explanation was that the plasma disappeared when the plasma sheet was severely stretched and became extremly thin, while the sudden appearances of plasma were associated with dipolarization, when plasma energized by the substorm rebounded earthward, swelled field lines of the near-earth tail and extended the plasma sheet past the satellite.

The deep magnetic bays accompanying the substorm arise from greatly enhanced electrojets, and the AE index [Sugiura and Davis, 1966] is often taken as a gauge of the level of substorm activity. Observations have suggested [McPherron et al., 1973, Fig. 7-8] that during substorms Birkeland currents are reinforced on the nightside by a system with region 1 polarity, arising from a diversion of part of the cross-tail current through the ionosphere. Because the shape of the diversion circuit tapers towards the Earth, this is known as the "substorm wedge current". In the ionosphere the wedge current follows the auroral oval, whose conductivity is greatly increased by the aurora, and thus reinforces the westward electroject around midnight. The wedge current may be intense enough for its magnetic effects to be observed at middle latitudes [Clauer and McPherron, 1974a, b] as well as in synchronous orbit [McPherron and Barfield, 1980].

By 1972 most observational features of the substorm had been identified, and they were reviewed by Rostoker [1972a] who also summed up the history of substorms (p. 163, 200), and by Aubry [1972]. A conference on substorms was held in October 1972 [Vasyliunas and Wolf, 1973] and an initial coordinated study of the substorms of 15 August 1973 was undertaken. The results of that study appeared in 1973 in 9 consecutive articles (J. Geophys. Res., 78, p. 3044-3149) the last of which [McPherron et al., 1973] presented an interpretation which included the "wedge circuit."

14. Substorms: Theory

As the features of substorms became known, attempts were made to explain them. One important feature was the strong correlation between substorms and "southward IMF". If the magnetosphere was quiet during a spell of northward IMF Bz, and suddenly Bz turned southward and stayed that way, it was found that a high probability existed for a substorm to erupt within an hour or so.

A further link was provided by Aubry et al. [1970; Aubry and McPherron, 1971] who found evidence that, other things being equal, the "nose" of the magnetosphere was pushed in closer to Earth at times of southward IMF Bz, a phenomenon qualitatively evaluated by Holzer and Slavin [1978] and by Sibeck et al. [1991; Roelof and Sibeck, 1993]. Aubry termed this phenomenon "erosion" of the magnetopause and claimed it occurred because closed field lines were being reconnected to interplanetary ones near N1 (Figure 7) faster than closed lines were arriving from the tail to take their place.

Coroniti and Kennel [1972] explored later developments of this scenario. If magnetic flux is removed from N1 faster than the sunward flow initiated near N2 brings it back, additional flux will pile up in the tail lobes. The lobes then swell and present a larger obstacle to the solar wind, which therefore compresses them more, increasing the lobe field BL: they identified this process with the growth phase of substorms, noting that the magnetic energy in the tail, whose density is proportional to BL squared, would increase.

Figure 16

Ultimately, in this scenario, the increased southward IMF reaches N2 and increases the rate of reconnection there, and after a while the supply of returning magnetic flux reaching N1 again matches the demand. But meanwhile other processes may intervene. The increased pressure on the lobe may squeeze the plasma sheet, causing reconnection at an internal neutral line N3 (Figure 16), so that the flux returning sunward is now supplied by reconnection at N3. Tailward of N3 an isolated magnetic bubble will be created, named "plasmoid" by Hones [1976; p. 567], a term previously applied by Bostick [1956, 1957, 1986] to a type of transient plasma bubble observed in the laboratory. Hones [1979, p. 393] described it as "...a blob of magnetospheric plasma ... detached from the magnetotail plasma...". The reconnection process at N3 was assumed to provide the substorm's energy and to accelerate particles; observations of impulsively accelerated particles in the tail [e.g. Keath et al., 1976; Roelof et al., 1976] were believed to indicate proximity to to N3.

The preceding scenario, with various modifications, has dominated the community's view of substorms in the last 20 years. In particular, Hones [1979] has collected evidence in its favor, and has argued that once a new neutral line N3 was formed, the tailward motion of the plasmoid would stretch it into a neutral sheet, explaining the formation of a very thin section of the plasma sheet ("thinning") inferred from the observed disappearance of plasma. This was disputed by Frank et al. [1976] who claimed substorms originated in "fireballs", possibly associated with boundary layers. The controversy persisted for a while [Hones, 1977, 1978a, b, c; Frank and Ackerson, 1977; Frank et al., 1978a, b], with Hones claiming that Frank's examples were in fact ordinary reconnection events, but it ultimately died down when no alternative scenario based on fireballs emerged.

Since that time several alternative theories of the substorm have been proposed [Kan, 1990]. Most interpretations place N3 fairly close to Earth (about 15 RE), but some views [Rostoker and Eastman, 1987] argued that substorms may reflect enhanced reconnection (and other processes?) at N2. Alternative theories have invoked a "thermal catastrophe" [Goertz and Smith, 1989], Alfven waves bouncing between the ionosphere and the tail [Kan et al., 1988] and disruption of the cross-tail current [advocated by Lui, 1991]. The proliferation of models has led to some skepticism [Stern, 1989b], and a meeting was held in Victoria, B.C. to seek some general agreement [Rostoker et al., 1980], but it did not clarify much.

One approach to studying substorms is to try correlate their variations with interplanetary stimuli, and thus seek to identify their causes or "triggers." Part of the problem here is the gauging of a substorm's intensity, and many studies have used for this the auroral AE index, or the related AU and AL indices. These reflect the strength of the auroral electrojets and therefore of the Birkeland current system, and they are known to become very high during large substorms.

Coroniti and Kennel [1972], Hones [1979] and most other researchers argued that the substorm obtained its energy from magnetic energy stored in the tail lobes, accumulated during a "growth phase" preceding substorm onset, during which the tail's magnetic flux increased [Caan et al., 1975]. Perreault and Akasofu [1978], however, found good correlation of AE with an "epsilon-parameter," constructed from solar wind characteristics and very sensitive to IMF Bz. They therefore proposed that substorms represented periods of stronger coupling between the solar wind and the magnetosphere, enabling the latter to extract more energy, and were thus "driven" by interplanetary conditions rather than representing the "unloading" of stored energy [Akasofu, 1980]. Some physicists now claim that both "unloading" and "driven" processes are involved. Predictive linear filters have also been used to study the way the solar wind input relates to AE [Iyemori et al., 1979; Clauer et al., 1981].

Theorists who support the "near-earth neutral line" (NENL) scenario have sought the "trigger" mechanism which initiates the onset of substorms, possibly some plasma instability. Computer simulations based on idealized MHD equations and assuming (purely) southward IMF [e.g., Walker et al., 1993] have supported the NENL scenario, yielding (with southward IMF) even more pronounced reconnection than seems to be observed.

15. Convection in the geotail

An interesting idea about the origin of substorms was proposed by Erickson and Wolf [1980; Hau et al., 1989; Erickson, 1984, 1992]. In the ideal MHD approximation, in the absence of rapid accelerations, it is expected that the magnetosphere is always close to force balance, mainly between pressure gradients and the magnetic force:

grad p= j x B      (2)

Properly p is a tensor, but in the plasma sheet the observed distribution of ions is close to isotropic and hence a scalar p is often used there. Solutions of (2) appropriate to a realistic 3-dimensional magnetosphere are not known, but two-dimensional solutions for a scalar p, assuming a linear dipole that extends indefinitely in the y-direction, can be obtained (at least numerically) from the Grad-Shafranov equation [e.g. Voigt and Wolf, 1988].

Erickson and Wolf [1980] noted that the existence of a convective plasma flow in the tail imposes additional restrictions. If the tail's field lines move with the flowing plasma, either the magnetic pattern is static and satisfies (2), and then the convection is such that each field line of the pattern is carried into another one; or else the pattern evolves, in which case (2) must hold at each intermediate stage. Erickson and Wolf showed that observed patterns of B and expected patterns of E were not compatible with a static scenario.

Erickson [1992] later simulated the field's evolution on a computer, satisfying (2) at all times; his calculation was two-dimensional, but flux arriving near Earth was allowed to "escape sideways" rather than piling up. This process led to a very weak B near the earthward edge of the plasma sheet (x near -12 RE), suggesting in almost all cases the imminent formation of a near-earth neutral line (the simulation could not go far enough to confirm it). By this scenario, substorm-type events may be an inevitable outcome of convection in the tail.

Attempts to actually observe this convection raise new problems. The double-probe method for observing E in low Earth orbit fails in the rarefied plasma of the tail, but the plasma's bulk flow can be inferred by measuring ion flux anisotropies. The first attempt [Frank and Ackerson, 1979] suggested a great deal of back-and-forth sloshing of plasma, but no underlying persistent earthward motion. A later study by Huang and Frank [1986] filtered out plasma sheet boundary layer observations, which contained field-aligned flows, and obtained an average earthward flow in the plasma sheet (r < 22 RE) of about 20 km/s, much below the expected rate. Recent studies by Angelopoulos et al. [1992] suggest that high-speed earthward plasma flows do exist (at about 150 km/s) but only about 7% of the time. The problem is thus still unsolved.

On to Section 16: Planetary Magnetospheres

Last updated 25 November 2001