A Brief History of Magnetospheric
Physics During the Space Age

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

5. Convection

The processes producing the complex features of the magnetosphere must meet two requirements: sufficient energy must be available, and particles must somehow be accelerated to observed energies. As for the supply of energy, it was estimated [Axford, 1964; Stern, 1984] that about 1-2% of the solar wind energy impinging on the magnetopause cross-section is tapped by internal processes of the magnetosphere.

In the neutral atmosphere of the Earth, energy is usually transmitted by two mechanisms: by large-scale circulating flows which convect heat from the ground upwards, and by radiation which takes a more direct path. The magnetosphere, too, may transmit energy both by convective flows and by a more direct route, involving field-aligned currents.

In an ideal magnetized plasma, a steady bulk flow with velocity v requires the existence of an electric field E, satisfying the "ideal magnetohydrodynamic (MHD) condition" [e.g. Walen, 1946; Alfven, 1950]

Conversely, an electric field E impressed on a magnetospheric plasma produces a bulk flow satisfying (1). It is a property of (1) that "particles move with field lines", i.e. any group of ions or electrons sharing a field line at one time continues doing so ever after, and a "moving field line" in what follows will mean a moving string of plasma particles, threaded by a common field line. If dB/dt = 0, the magnetic configuration is fixed and on any "moving" line, the plasma population along its entire length migrates to an adjoining line: thus field lines can (for instance) transmit bulk motions from distant regions to their ionospheric ends. In inductive electric fields with dB/dt not zero, field line sharing also holds [Newcomb, 1958; Stern, 1966] but bulk motion is not necessarily transmitted along field lines [Stern, 1990, Figure 7].

The existence of the Chapman-Ferraro cavity (see BH-1) and hence of the magnetopause may be viewed as another consequence of field line sharing: as long as such sharing is rigorously enforced, there exists no way for interplanetary plasma, threaded (presumably) by fields of solar origin, to mix with plasmas of the Earth's field. For related reasons, as long as all terrestrial field lines are confined to the cavity's interior ("closed magnetosphere"), it is also difficult for energy, momentum and electric currents to enter the cavity from the outside. In the early days many scientists in fact believed that magnetospheric field lines were in this way completely confined inside the cavity. The alternative view of an "open" magnetosphere developed gradually and is discussed in sections 7 and 8.

Gold [1959, p. 1220] noted that the large scale flow of magnetospheric plasma (a type of which he was studying) "is quite analogous to thermal convection" and that led to the term "convection" used by Axford and Hines [1961] to describe large-scale circulation inside the magnetosphere, caused by the solar wind. The theory of Alfven [1939] (see BH-1, also Cowling [1942] and Stern, [1977]), although not consistently formulated, may also be viewed as a theory of magnetospheric convection. Contemporary theories began with Axford and Hines [1961, also Hines, 1974, p. 3, 933; Axford, 1962, 1964, 1994; Hines, 1964, 1986] and with the work of Dungey [1961] described further below. Axford and Hines proposed a convective circulation to explain an observed pattern of auroral motions [Davis, 1962, 1971] in which plasma seemed to circulate in the polar cap.

Axford and Hines visualized a magnetosphere whose field filled a cavity in the solar wind, elongated on its night side into a tail, as previously suggested by Johnson [1960], so that all field lines emanating near the magnetic pole extended into the tail. Their proposed convective flow pattern (Figure 5a, from Hill [1983])) carried plasma tailward along the flanks and returned it by means of a sunward flow near the x-axis, skirting around the region closest to Earth. They furthermore suggested that such a flow could be caused by a viscous-like momentum transfer from the solar wind to adjacent regions of the tail, although they admitted that other processes could produce similar flows in the polar cap, including Dungey's reconnection scenario (further below).

When the flow pattern of Figure 5a is mapped along field lines to the polar ionosphere, it produces a two-cell flow pattern, with plasma streaming nightwards across the pole and returning to the day side at lower latitudes, with flow lines similar to the contours in Figure 5b. By equation (1), if E =-grad V, it follows that v/gradV = 0 and therefore the plasma flow lines in Figure 5b are also lines of constant electric potential F, suggesting a dawn-to-dusk electric field across the polar caps. Satellites in a low-altitude polar orbit can observe such a field directly, by measuring the small voltage difference between the tips of a long antenna [Aggson, 1968; Cauffman and Gurnett, 1972]. The first satellites to successfully conduct such observations were Iowa's Injun 5 [Cauffman and Gurnett, 1971] and the OGO-6 observatory [Heppner, 1972a, 1977]; some later missions, e.g. Atmosphere Explorer 1 and Dynamics Explorer 2, measured E indirectly using "driftmeters" which observed v through the anisotropy of particle fluxes caused by the plasma's bulk motion [Hanson and Heelis, 1975; Heelis et al., 1981].

The observations confirmed the two-cell pattern and obtained typical voltage drops of 40-70 kV; this agreed with a prediction of the reconnection model by Levy et al. [1964; sect. VI]. Much depends upon the state of the interplanetary magnetic field (IMF). The two-cell pattern is most stable when the IMF has a southward slant (Bz < 0, see below), and it contains asymmetries correlated with the dawn-dusk By component of the IMF [Heppner, 1972c; Heppner and Maynard, 1987]. The cross-polar voltage drop DF on the average grows with southward Bz, though individual observations fluctuate greatly. With northward Bz the average DF sometimes decreases to less than 20 kV, and it has been suggested that at such times it may "bottom out" at a low level contributed by a viscous-like interaction at the flanks [Reiff et al., 1981; Wygant et al., 1983].

When the IMF has a northward slant, the 2-cell pattern becomes distorted and E is often irregular. At times non-standard patterns may develop, such as the 4-cell pattern deduced by Burke [1979]. More complex patterns have also been claimed [Reiff and Burch, 1985] but they are hard to confirm without simultaneous passes by a fairly large number of satellites.

In the innermost magnetosphere the plasma density n is dominated by the thermal ionospheric plasma which tends to co-rotate with Earth; this is another consequence of field line sharing [Ferraro, 1937] and is enforced by a corotation electric field ECR, which near Earth is much larger than the convection field E. It was found from whistler wave observations [Carpenter, 1963; Carpenter and Park, 1973] and later by in-situ observations that n often dropped precipitously from about 20-100 ions/cc to about 5 ions/cc at a "plasmapause" boundary on field lines that extended to 4-5 RE. Brice [1967; Kennel, 1985] and Nishida [1966] proposed that this was essentially the boundary of the region where low energy plasma shared the rotation of the Earth; beyond it the convection electric field E overpowered ECR. This view is now widely accepted, although the suggestion was also made [Lemaire, 1975] that the plasmapause was the limit beyond which low-energy plasma was easily lost through the interchange instability, an approach first explored by Brice [1973].