Interplanetary Conditions. L. R. Lyons. Department of Atmospheric Sciences. University of California, Los Angeles. Los Angeles, CA

Transcription

1 Geomagnetic Disturbances: Characteristics of, Distinction Between Types, and Relations to Interplanetary Conditions by L. R. Lyons Department of Atmospheric Sciences University of California, Los Angeles Los Angeles, CA Invited review based on review presented at the CEDAR-SHINE-GEM special session at 1999 CEDAR meeting Submitted to: Journal of Atmospheric and Solar-Terrestrial Physics November,

2 Abstract This review emphasis disturbances in auroral emissions and ionospheric currents and their relation to interplanetary conditions and the overall level of geomagnetic activity. Auroral zone disturbances are divided into three fundamentally different types: poleward boundary intensifications (PBIs), substorms, and effects of solar wind dynamic pressure enhancements. The most common type of auroral-zone disturbance is the PBI, which occurs during all levels of geomagnetic activity. PBIs have an auroral signature that often can be seen to move equatorward from the magnetic separatrix. They are typically associated with ground magnetic perturbations of few tens of nt, but perturbations can be as high as ~500 nt. Individual PBIs are longitudinally localized, associated with the longitudinally localized flow bursts in the tail plasma sheet, and occasionally traverse essentially the entire latitudinal extent of the plasma sheet. PBIs appear to generally be the dominant type of auroral zone disturbance during periods of enhanced magnetospheric convection, including the growth phase of substorms, convection bays, and the main phase of magnetic storms. Substorms are a far more dramatic and large-scale, but far less common, disturbance than PBIs. They occur after a ~ > 30 min growth-phase period of enhanced convection. It is now known that at least ~50% of substorms are associated with IMF changes that lead to a reduction in the strength of convection. However, it has not yet been shown whether or not all are most substorm onsets are caused by these types of IMF changes. Auroral activity during substorms typically initiates within a ~1-2 hr MLT sector near the equatorward boundary of the auroral oval and then expands both poleward and azimuthally. Substorms are associated with ground magnetic disturbances that range from ~50 nt to ~2000 nt, a reduction in strength of the cross-tail current, a poleward displacement of the inner edge of the plasma sheet, and a large release of plasma and magnetic field energy from the region earthward of the new inner edge of the plasma sheet. The reduction of cross-tail current is also believed to often be associated with a severance, and loss from the magnetotail, of the outer portion of the plasma sheet (r ~ > 25 R E ). Recent 2

3 studies have shown that solar wind dynamic pressure increases caused large auroral zone disturbances during a stormtime period of strongly enhanced convection, affecting the poleward boundary, latitudinal width, and intensity of the auroral oval. Dynamic pressure increases also appear to also enhance the entire magnetospheric current system, including the magnetopause, cross-tail, region 1 field-aligned, and global ionospheric currents. Thus, in addition to PBIs, significant variations in solar wind dynamic pressure should be considered as a possibly important source of geomagnetic disturbances during periods of enhanced magnetospheric convection. Introduction Transient enhancements of auroral emissions and ionospheric currents often occur within the Earth s auroral zones, which typically are a few to about ten degrees in latitude wide and lie at geomagnetic latitudes Λ between ~ 63 and ~75. There are several different types of transient disturbances, having time scales ranging from a few minutes to a few hours and ground magnetic field perturbations ranging from a few tens of nt to ~2000 nt. These auroral-zone disturbances occur within the ionospheric extension of the plasma sheet. The plasma sheet lies on closed magnetic field lines that cross the equatorial plane tailward of an inner edge, which lies at an equatorial radial distance r ~5-15 R E, depending on the strength of the convection electric field. Magnetic storms are a fundamentally different phenomenon from auroral zone disturbances. They occur when fluxes of energetic particles (~50 kev - 1 MeV) greatly increase in the inner portion of the radiation belts (2 < ~ L < ~ 5). The azimuthal drift of these particles significantly enhances the azimuthal current around the Earth known as the ring current. This ring current causes magnetic field depressions at low and mid latitudes that can be as large as nt. Ring current enhancements giving an average magnetic depression that exceeds ~50 nt are referred to as magnetic storms. Magnetic storms typically occur ~10-20 time per year, as compared to auroral zone disturbances which are observable essentially every day at some level of intensity. Also, storm time scales are 3

4 significantly longer than the time scales of auroral zone disturbances, ring current formation occurring during a several hour period known as the storm main phase and ring current decay during an ~2-3 day period known as the storm recovery phase. The relative locations of the plasma sheet and the stormtime ring current are illustrated in Figure 1. Note that the plasma sheet is always present whereas a significant stormtime ring current is only occasionally present. Auroral-zone disturbances are emphasized in this review. Their occurrence during magnetic storms is included; however, the physics of the formation and decay of the stormtime ring current is not included. The review starts with a general discussion of the driving of magnetospheric convection by the solar wind and the solar wind control of the plasma sheet. With this background in mind, different types of auroral zone disturbances that have so far been identified are described, distinguished from each other, and related to the strength of convection and to plasma sheet dynamics. Three different types of disturbances are considered: substorms, poleward boundary intensifications (PBIs), and effects of solar wind dynamic pressure enhancements. Each of these types has unique characteristics and reflects distinctly different physical processes occurring within the magnetosphere. These disturbances are also described in relation to the overall level of geomagnetic activity. Four overall levels of activity are considered: quiet-times, substorm periods, convection bays, and magnetic storms. Convection Driven by the Solar Wind Transfer of energy to the magnetosphere is controlled by the electric field which extends into the magnetosphere from the solar wind along open polar-cap field lines (those which cross the magnetopause and connect with the interplanetary magnetic field). As illustrated in Figure 1, this electric field, which is known as the convection electric field, extends into the closed field line region of the magnetosphere causing inward motion and energization of the plasma sheet on the nightside. Auroral-zone disturbances are related to 4

5 the energy and dynamics of the plasma sheet. Thus these disturbances depend strongly on the strength of the convection electric field. The potential drop across the polar cap φ pc is a good measure of the strength of the convection electric field. It depends strongly on the interplanetary electric field, which is a function of the y- and z-components, B y and B z, of the interplanetary magnetic field (IMF) and the solar wind speed V sw. The dependence of φ pc on B z is well known [e.g., Cowley, 1984, and references therein], φ pc increasing as B z becomes increasingly negative. This dependence is shown in Figure 2. φ pc also depends on B y. However, this dependence has not been evaluated in detail. A significant increase in φ pc with B y can be seen from Table 1. Table 1 gives φ pc obtained by G. Lu (private communication, 1998) using the Assimilative Mapping of Ionospheric Electrodynamics (AMIE) procedure [Richmond, 1992] for time intervals of stable IMF that were studied as a part of the Geospace Environment Modeling (GEM) Program. In the table, φ pc is shown for two pairs of examples. For each pair, the values of B z are nearly the same but the magnitudes of B y are small in one example and very large in the other. Following Dungey [1965] and Stern [1973], we can estimate the dimensions of the interplanetary region of field lines that connect with the geomagnetic field and along which the interplanetary electric field maps into the magnetosphere. The boundary between open and closed magnetic field lines (the magnetic separatrix) intersects the ionosphere in an approximately circular region at Λ ~ 73, so that the total amount of open magnetic flux emanating from each polar cap is ~7 x 10 8 T-m 2. This flux must be conserved everywhere along the mapping of this region of open field lines. Thus, for an interplanetary field strength of 5 nt, the magnetic flux through each polar cap must pass through a region 2 normal to the interplanetary magnetic field of area ~3400 R E. Approximating this region as a rectangle of width W in the direction normal to the solar wind and length L, as illustrated in Figure 3, we have that φ pc = B yz V sw W. Here B yz is the magnitude of B is the plane normal to the solar wind, which is assumed to be in the x direction. Taking B yz ~ 5

6 4 nt, V sw ~ 400 km/s and φ pc ~ 100 kv, gives W ~ 10 R E, which is considerably smaller than the diameter of the magnetotail (~50 R E ), and L ~ 340 R E. We thus find that the interplanetary regions magnetically connected to the magnetosphere are narrow and highly elongated as shown in Figure 3. The narrow and elongated shape of the interplanetary regions of interconnected field lines has two important consequences. First, it takes a 400 km/s solar wind ~90 min to traverse this region, so that IMF changes will also take ~90 min to traverse this region. We would thus expect changes in convection imparted to the magnetosphere to slowly propagate across the magnetosphere. However, observations in the ionosphere show that this is not the case [Nishida, 1968; Ridley et al., 1998; Ruohoniemi and Greenwald, 1998]. Instead, the strength of convection increases or decreases in response to IMF changes nearly simultaneously over the entire polar cap. Convection changes are observed to initiate ~10 min after an IMF change contacts the dayside magnetopause, and it takes another ~10 min for convection to fully adjust to a new IMF configuration. Thus convection changes at the low altitude end of nightside field lines occur long before electric field changes propagate to the interplanetary extension of these field lines, implying that convection changes at low altitudes cannot be understood solely in terms of a simple mapping of interplanetary electric fields along field lines. Understanding the propagation of the convection changes through the magnetosphere is an important problem, and one that is currently a topic of investigation. The rapid response of convection is seen throughout the ionosphere and is most likely appropriate to the inner portions of the magnetosphere; however, it may not apply to the more distant magnetotail (r ~ > 30 R E ). Furthermore, the rapid changes refer only to the potential portion of the electric field. In addition to this, magnetic field changes are associated with induced electric fields, which can at times be important at altitudes well above the ionosphere. The narrowness of the interplanetary region of interconnected field lines must also be considered when relating the measured IMF to geomagnetic activity. Figure 4 shows a 6

7 cross section of the magnetosphere in the y,z-plane and illustrates the extension of the region of interconnected field lines into the solar wind. Most IMF measurements are made somewhere near or within the plane shown in Figure 4. However, the region of interconnected field lines occupies only a small fraction of this plane, so that most IMF measurements are not made within the region of interconnected field lines and many measurements are made 10 s of R E away from this region. Thus, structure of the IMF in the y,z plane can cause significant differences between the IMF measured on a spacecraft in the vicinity of the Earth and that which actually affects the magnetosphere. This structure is illustrated by the representative (4 of 11 cases with multiple satellite observations from Lyons et al. [1997]) examples of simultaneous IMF measurements with 2-3 satellites shown in Figure 5. These examples show that at times there can be good correlation between IMF measurements at locations separated by 10 s of R E in the y,z-plane. However, often there are very significant differences in both the IMF structures and in the timing of specific structures. It is generally not possible to accurately determine how representative a particular IMF measurement is of that which actually affected the magnetosphere, so that it is often necessary to resort to statistical studies to determine the magnetospheric response to different interplanetary inputs. Solar Wind Control of the Plasma Sheet The solar wind not only drives the magnetospheric electric field, but it is also an important source of magnetospheric plasma. Early models of the magnetosphere considered the effect of a steady solar wind on the Earth s magnetic field, but did not include the effects of currents associated with plasma in the interior of the magnetosphere. These models gave a magnetosphere that was realistically compressed on the dayside, but was also compressed on the nightside [e.g., Mead, 1964]. Following discovery of the geomagnetic tail and the plasma sheet [Ness, 1965; Bame et al., 1966; Vasyliunas, 1968], it became recognized that the plasma sheet and its imbedded current sheet were fundamental features of the magnetosphere. Since auroral-zone disturbances occur within the plasma 7

8 sheet, understanding the source of plasma sheet particles and the energy of the plasma sheet is a critical aspect of understanding auroral zone disturbances. We know that the energy of the plasma sheet is highly variable and controlled by the IMF, the solar wind density n sw, and the solar wind dynamic pressure; however this control has not yet been the subject of intensive study so that current knowledge of this subject is quite limited. It is now generally accepted that the solar wind is the primary source of particles for the tail plasma sheet [Walker et al., 1999]. An important means by which solar wind particles access the plasma sheet is via the plasma mantle as illustrated in Figure 6 [Pilipp and Morfill, 1978]. This figure shows solar wind particles crossing the dayside magnetopause along open polar-cap field lines in the region of the dayside cusp, then mirroring and forming the region interior to the nightside magnetopause that is know as the plasma mantle, and eventually reaching the plasma sheet as a result of electric field drift within the tail lobes. Strong observational confirmation of this type of plasma access to the plasma sheet has been recently obtained from the GEOTAIL spacecraft [Maezawa and Hori, 1998]. This plasma access should give a plasma sheet density that correlates well with n sw after a delay appropriate for the transport of solar wind particles to the plasma sheet, and such a correlation have been found to exist [Borovsky 1997]. Thus n sw is an important contributor to the energy content of the plasma sheet. The interplanetary media should also control plasma sheet energy through the strength of the convection electric field. It is well accepted that particles are significantly energized by the cross-tail electric field when they enter the tail current sheet [Speiser, 1965, Lyons and Speiser, 1982], and this energization is directly proportional to the strength of the cross-tail electric field. In addition, the strength of the cross-tail electric field should be important for controlling how many particles from the mantle have access to the closed field line portion of the plasma sheet. For a weak cross tail electric field, only a small fraction of mantle particles can reach the current sheet earthward of the tail magnetic x-line. (For example, for a cross tail potential drop of 50 kv, a tail width of 50 R E, a lobe 8

9 magnetic field strength of 15 nt, and a distance of 15 R E between the mantle and the center of the current sheet, we find that it takes mantle particles ~2.7 hr to drift across the lobes to the current sheet so that only particles with a parallel velocity v < ~ 70 km/s can reach the closed field line portion of the current sheet. This velocity is considerably below typical mantle flow speeds of ~150 km/s [Rosenbaur et al., 1975]). However, for a large crosstail potential drop, the number of particles reaching the current sheet earthward of the x-line should be significantly increased. (A potential drop of 120 kv would give a drift time to the current sheet of ~1.1 hr., so that particles with v up to ~160 km/s can reach the closed field line portion of the current sheet.). In addition, the solar wind dynamic pressure (which depends upon n sw and V sw ) should affect the energy of the tail plasma sheet because of the compression of the magnetosphere that occurs when the dynamic pressure increases. This effect has been seen in case studies [Russell et al., 1994; Collier et al., 1998]; however this potentially important effect has not been comprehensively evaluated. We thus expect there to be very large variations in the energy of the tail plasma sheet, and these variations should be strongly dependent on the IMF, n sw (with time delay), and solar wind dynamic pressure. A good measure of the energy density of plasma at the center of the tail current sheet is the energy density of the magnetic field in the tail lobes. While a study correlating lobe energy densities to the IMF and solar wind parameters has not been performed, Yamamoto et al. [1994] looked at a large number of lobe magnetic field energy densities and found very large variability, the total variation being nearly a factor of 100. The largest values were found to occur exclusively during magnetic storms, which are associated with periods of strongly southward IMF, and thus strongly enhanced convection, and with greatly enhanced solar wind dynamic pressure. An example of dramatic enhancements of lobe energy densities during a very weak storm (minimum Dst -51 nt) on November 27, 1995 is shown in Figure 7. The bottom panels show the total pressure P tot (sum of the magnetic and plasma pressure) and ion 9

10 pressure P ion obtained from GEOTAIL. Assuming pressure balance in the direction perpendicular to the tail current sheet, P tot should approximately equal the lobe energy density. In Figure 7, P ion has been used for the plasma pressure. This is generally a good approximation because electron temperatures are well below ion temperatures in the plasma sheet. The relative contribution of P ion to P tot reflects the proximity of the spacecraft to the center of the plasma sheet, negligible values of P ion occurring when GEOTAIL was within the lobes. Also shown is the lobe pressure based on the average lobe magnetic field, B lobe = (124 nt) x -0.54, obtained by Slavin et al. [1983]. The upper panels in Figure 7 show the IMF and n sw (approximately proportional to the solar wind dynamic pressure, since the V sw varied far less than did the density) from the WIND spacecraft located slightly in front of, and well to the dusk side, of the magnetosphere. Times for the WIND measurements have been delayed by 16 min so as to give as good as possible agreement with IMF measurements from IMP located just outside the dawn magnetopause and at about the same x as GEOTAIL. The total variation in P tot was about a factor of 8 during the 12 hr interval shown in the figure, and the initiation of two sharp increases in P tot are identified by vertical dashed lines. Figure 7 shows significant IMF and dynamic pressure effects on the lobe energy density. The gradual increase in P tot from ~0715 UT to 0830 UT, which occurred as B y gradually increased, and the variations in P tot between 12 and 18 UT, when n sw was nearly constant, are clearly IMF related. On the other hand, the initiation of the strong enhancement in P tot at 0830 UT and the spike in P tot that began at 1100 UT are primarily responses to solar wind dynamic pressure. The large peak values of P tot of ~2 npa clearly resulted from the simultaneous strong enhancement of both IMF driven convection and the solar wind dynamic pressure. Figure 7 demonstrates that IMF and solar wind control of tail energy is an important aspect of magnetospheric dynamics. However, current knowledge of this connection is insufficient and significant further study is clearly needed. 10

11 A critically important aspect of the dynamics of the plasma sheet is its inward motion during periods of enhanced convection. Vasyliunas [1968] found that the inner edge of plasma sheet electrons varied dramatically with geomagnetic activity, being located at r ~ 11 R E during quiet intervals and at r ~ 6-8 R E during periods of geomagnetic activity. It has been seen as close in as r = 5 R E [Smith and Hoffman, 1974]. Vasyliunas [1968] concluded that the plasma sheet moves inward during geomagnetically active periods as a result of inward plasma motion driven by the cross-tail electric field. Vasyliunas s idea that the inner boundary of the plasma sheet moves inward in response to enhancements in convection is supported by calculations of the trajectories of plasma sheet particles [e.g., Kavanagh et al., 1968; Jaggi and Wolf, 1973; Wolf, 1995] and by observations of the MLT of the inner edge of the plasma sheet at synchronous orbit as a function of the level of geomagnetic activity [Elphic et al., 1999; Korth et al., 1999]. Figure 8 illustrates trajectories (based on the calculations of Chen et al. [1993]) of equatorially mirroring protons in the equatorial plane for conditions of weak and enhanced convection. The trajectories are sketched for a magnetic moment of 20 MeV/Gauss, which is representative of plasma sheet protons (~20 kev at r = 7 R E ). In general, plasma sheet particles convect earthward until they reach the radial distance where magnetic drift deflects them around azimuthally around the Earth, electrons toward the dawn side and positive ions toward the dusk side. Once particles reach the dayside, the convection electric field carries them sunward across the dayside magnetopause. When convection is weak, the deflection by magnetic drift prevents plasma sheet particles from reaching the <10 R E portion of the nightside magnetosphere. However, enhanced convection brings particles well into this region. The strong effect of convection strength on the radial distance to which plasma sheet particles have access ought to have very significant effects on particle distributions in the ~6-10 R E region of the nightside magnetosphere. To measure the response of the near- Earth plasma sheet to variations in convection and the relation of this response to 11

12 geomagnetic activity requires measurements of plasma sheet particles as a function of equatorial radial distance. Unfortunately, such measurements have been very limited in the past. However the data of Schield and Frank [1970] show that the inner edge of the plasma sheet is displaced inward during the main phase of moderate storms, and Kivelson et al. [1973] inferred that the boundary moves inward during the growth phase of substorms. Both the main phase of storms and the growth phase of substorms are conditions of enhanced electric fields, for which inward motion of the inner edge of the plasma sheet is expected. The expected response of the plasma sheet to changes in magnetospheric convection can be seen in the particle data from the POLAR satellite in Figure 9. Figure 9 shows spectrograms of ion and electron counts from ~0.01 to ~15 kev as obtained from the HYDRA instrument [Scudder et al., 1995] on POLAR on May 9 (top) and May 12, 1996 (bottom). For both passes, the satellite was near 2245 MLT and moved equatorward and earthward from the northern hemisphere lobe, entering the plasma sheet at an invariant latitude Λ ~ 69 on May 9 and ~70 on May 12. The poleward boundary of the plasma sheet is clearly identified by an abrupt increase in ion and electron count rates at energies from a few hundred ev to the highest energy of detectable counts. As the satellite continued equatorward and earthward, the inner boundary of the plasma sheet was encountered. This boundary is particularly abrupt for electrons, and is seen at significantly different locations for the two orbits. The May 12 orbit was during an extended period of positive interplanetary magnetic field (IMF) B z (~5 nt) so that convection is expected to have been weak. The inner boundary of plasma sheet electrons on this orbit was located at a model invariant latitude of Λ ~ During the period of the May 9 orbit, the IMF B z was negative ( 2 to 5 nt). As expected from enhanced convection during a period of negative IMF B z, the inner edge of the plasma sheet was encountered at significantly lower Λ (~63 ) on this orbit than on the May 12 orbit. While the invariant latitudes in Figure 9 were obtained from a magnetic field 12

13 model which does not consider magnetic field changes due to plasma sheet dynamics, Figure 9 shows a clear difference in the location of the inner edge of the plasma sheet and in the width of the plasma sheet between the two orbits that relates to the expected strength of convection. For both orbits, the inner edge of plasma sheet ions is located very close to the inner edge of plasma sheet electrons. (The ions extend further earthward over a narrow range of energies near 8 kev. At these energies, the westward magnetic drift speeds and eastward corotation speeds for ions are nearly equal, so that the total azimuthal drift speed is very low. Thus azimuthal drift, which limits the earthward penetration of most particles, is less effective over this energy range and allows enhanced earthward penetration via convection [Smith and Hoffman, 1974]). Earthward penetration of the plasma sheet during periods of enhanced convection has very significant effects on plasma pressures and magnetic fields in the inner portion of the plasma sheet. Figure 10 shows equatorial plasma pressures that have been obtained during quiet times [Spence et al., 1989], when convection is expected to be weak, and during periods just preceding the onset of substorms [Kistler et al., 1992], when convection is expected to be strong. It can be seen that plasma pressures are about an order of magnitude higher at r < ~ 10 R E when convection is strong than when it is weak. The equatorial pressure of the Earth s dipole field is also shown in Figure 10. Even during quiet times, the plasma sheet pressure is significantly greater than the dipole field at r ~ > 12 R E, giving the large magnetic distortion of the magnetotail which exists for all levels of magnetic activity. However, during quiet times plasma pressures are below the dipole field pressure at r < ~ 10 R E so that the magnetic field in this region is nearly dipolar. Enhanced convection causes plasma pressures to exceed the dipolar field pressure down to r 6-7 R E, leading to significant tail-like distortion of the magnetic field in this inner region. The large stretching of the magnetic field that occurs during periods of enhanced convection can be seen in Figure 11. Here field-lines emanating from the Earth at five fixed latitudes are shown as obtained from the Tsyganenko-96 magnetic field model for 13

14 quiet and for disturbed conditions. The stretching of the field in the r 6-10 R E region is illustrated by the field line that cross the equatorial plane at x 5.8 R E in the quiet time case and extends to x 9.3 R E in the disturbed case and by the field line that crosses the equatorial plane at x 7.3 R E in the quiet case and extends to x 19.8 R E in the disturbed case. Auroral activity Ion motion along tail-like magnetic field lines violates the guiding center approximation near the magnetic equator, resulting in strong pitch angle scattering of ions throughout the plasma sheet [Lyons, 1997]. This leads to precipitation, which causes atmospheric emissions that can be measured with ground-based photometers. Measurements of these emissions can be used to monitor the intensity, motion, and location of the plasma sheet as a function of time. Combined with observations of auroral disturbances, such measurements can be used to determine the location and motion of auroral disturbances relative to plasma sheet boundaries. Figure 12 shows an example of auroral-zone emissions obtained with the CANOPUS meridian-scanning photometers during a 12 hr period of low geomagnetic activity from UT on 18 December The middle panels show emission intensities as a function Λ and UT as obtained from a merging of data obtained along the magnetic meridian from CANOPUS [Rostoker et al., 1995] stations Gillam (GILL, Λ = 67 ) and Rankin Inlet (RANK, Λ = 74 ), which are at approximately the same longitude. The photometers measure emission intensities as a function of elevation angle along the magnetic meridian; latitude profiles have been obtained from these measurements by assigning a fixed altitude to each of the emissions. The lower panels show ground magnetic field data from the CANOPUS meridian chain of magnetometers extending from Pinawa (PINA) to Rankin Inlet, and the upper panels show IMF data from IMP 8. Photometer data for three different wavelengths are shown in Figure 12. The 6300 Å emission reflects precipitation of plasma-sheet electrons at energies < ~ 1 kev. This 14

15 emission shows a clear poleward boundary which tracks the poleward boundary of the plasma sheet and provides a good identification of the separatrix that separates the region of open polar-cap field lines from the lower latitude region of closed field lines [Samson et al., 1992; Blanchard et al. 1995]. The location of this boundary as determined by an automated procedure developed by Blanchard et al. [1997] is shown in the figure as a black and white line. The 4861 Å emission is due to proton precipitation. The most intense 4861 Å emissions correspond to the innermost portion of the plasma sheet proton distribution [Samson et al., 1992], so that the band of most intense emissions tracks the temporal evolution of the inner plasma sheet as mapped to the ionosphere. A black and white line is drawn in Figure 12 at the latitude of the peak in the 4861 Å emission profile. The 5577 Å emission responds primarily to > ~ 1 kev electron precipitation, but also responds weakly to protons. This emission is particularly useful for identifying and tracking auroral disturbances. 1. Quiet times-auroral poleward boundary intensifications The photometer and magnetometer data in Figure 12 are typical of periods of weak convection when the IMF B z remains mostly positive. Notice that the peak of the 4861 Å emission remained poleward of invariant latitude Λ = 68 and relatively weak (as compared to what is seen during more disturbed periods), and that the poleward boundary of the plasma sheet remained relatively stable at Λ ~ 73 until over an hour past magnetic midnight. Consistent with the lack of significant auroral activity, the magnetometer data for this period are extremely quiet. Note, however, that the 5577 Å emissions in Figure 12 show some weak auroral enhancements near the polar-cap boundary. These enhancement, referred to here as poleward boundary intensifications, or PBIs, are the most common type of auroral-zone disturbance. PBIs occur repetitively, so that many individual disturbances can occur during time intervals of ~1 hr, and they generally appear to be the most intense auroral disturbance at times other than the expansion phase of substorms. These disturbances 15

16 occur during all levels of geomagnetic activity [Lyons et al., 1998], and they have an auroral signature that often can be seen to move equatorward from the magnetic separatrix. A quiet-time example (from 18 January 1997) with stronger PBIs is shown in Figure 13. Magnetic and auroral activity during the period shown (03-08 UT) was quiet except for the poleward boundary intensifications. The color panel in the middle of the figure shows 5577 Å emissions obtained from a merging of data from the meridian scanning photometers at Gillam and Rankin Inlet; the white-dashed line in that panel gives the poleward boundary of the plasma sheet as obtained from the photometer observations of 6300 Å emissions (not shown). The PBIs can be seen to have occurred repetitively and were particularly bright at Λ ~ from UT. They may have been just as bright from UT; however during this earlier period the optical emissions were partially obscured and significantly scattered by clouds at Rankin Inlet. Some of the PBIs can be faintly seen to propagate equatorward to ~68 latitude (e.g., those at 0555 and 0605 UT), indicating that these disturbances can propagate large distances through the plasma sheet. Figure 13 shows that PBIs can be associated with disturbances in the ground magnetic field at the location of the auroral enhancement. The PBIs from ~ ~0615 UT were associated with a series of ~50 nt ground magnetic perturbations and a continuous series of Pi-2 pulsations at Rankin Inlet, which is at approximately the same invariant latitude as the brightest auroral activity. At UT, PBIs initiated near Λ = 72, and ground magnetic perturbations associated with these PBIs were seen at the station (ESKI) at that latitude. A good conjunction with GEOTAIL at x~ -30 R E in the tail occurred during the time of the CANOPUS measurements shown in Figure 13, and line plots of the x-component of magnetic field B x and the x- and y-components of the plasma velocity V x and V y measured by GEOTAIL are shown in the upper panels of the figure. The heavier lines in this panel give V perp,x and V perp,y, which are the x and y components, respectively, of the velocity component perpendicular to the measured magnetic field. GEOTAIL was located at 16

17 longitudes just to the dawn side of 24 MLT during the time period shown in Figure 13. Thus the best conjunction between GEOTAIL and CANOPUS (magnetic midnight at ~0640 UT) occurred from ~ UT. Throughout the UT period, GEOTAIL was within the plasma sheet and saw several periods of bursty flows (several min periods of structured flows with peak values of V x ~ > 200 km/s), which are now believed to be an important means of plasma transport in the magnetotail [Baumjohann et al., 1990; Angelopoulos et al., 1992]. Figure 13 shows that bursty flows were observed primarily during the time periods when PBIs were observed (identified by shading in the figure), and that the plasma sheet was very stable with little flow during the time periods when poleward boundary intensifications were not seen. This suggests that the bursty flows are associated with the PBIs, a suggestion that has been supported by other case studies [de la Beaujardière et al., 1994; Kauristie et al., 1996; Yeoman and Lühr, 1997]. The connection between the bursty flows and the PBIs arises because the bursty flows, which are quite localized in longitude [Angelopolous et al., 1994, 1996], are associated with significant (~5 nt) structure in the magnetic field indicating that the bursty flows are associated with structured currents in the tail [Lyons et al, 2000a]. These currents are estimated to be sufficiently intense, that when mapped to the ionosphere, they require the formation of the magnetic field-aligned potential drop which accelerates electrons downwards towards the ionosphere [e.g., Lyons, 1992]. All-sky auroral images show that PBIs that extend equatorward from the poleward boundary of the auroral zone often are elongated in the north-south direction [Zesta et al., 1999]. North-south auroral structures have been occasionally reported in two-dimensional images of the aurora [Rostoker et al., 1987b; Nakamura et al., 1993], and these structures are apparently the same phenomena as the poleward boundary intensifications discussed here. Consistent with this, Henderson et al. [1998] suggested that the north-south structures are related to the bursty flows in the tail, and Sergeev [2000] found direct 17

18 correspondence between an individual north-south auroral structure and an earthward going flow burst observed in the tail. 2. Substorms Substorms are the most dramatic auroral zone disturbance. The concept of a global substorm was based on ground-based observations of aurora [Akasofu, 1964], which show a sudden brightening of a quiet auroral arc within the equatorward portion of the auroral zone near magnetic midnight at the onset of the substorm expansion phase. Onset is accompanied by suddenly enhanced Pi 2 micropulsations [Rostoker, 1968]. After the initial auroral brightening, the region of active aurora expands poleward and azimuthally at speeds on the order of 1 km/s. The region of westward-moving, active aurora is known as the "westward traveling surge". Expansion-phase active aurora is accompanied by an enhanced westward ionospheric current that causes a large negative change in the ground magnetic field [e.g., Akasofu and Meng, 1967; Akasofu, 1968; Nishida and Kokubun, 1971; Kisabeth and Rostoker, 1971]. This enhanced current, known as the "westward electrojet", moves poleward with the region of active aurora during the expansion phase [Kisabeth and Rostoker, 1971]. Figure 14 shows CANOPUS photometer and ground magnetometer data for 1-13 UT on 12 January 1991, which includes a substorm period with onsets (indicated by dotted lines) at ~06 and ~0630 UT, and for 1-13 UT on 12 February 1991, which had larger substorm onsets at ~0215 and ~0800 UT and a smaller onset at ~0710 UT. IMF data are available for the 12 January 1991 interval. The photometer data in Figure 14 show typical substorm characteristics. Prior to substorm onset, the peak of the 4861 Å emissions moved equatorward to a Λ < ~ 67 (64-66 for the cases shown). This is the expected manifestation of the earthward motion of the plasma sheet that results from enhanced convection during the substorm growth-phase. The IMF data for the 12 January example shows that this growth phase signature initiated at about the time that IMF turned southward. After each of the periods of expansion phase activity in Figure 14, the band of 18

19 proton precipitation reappeared at higher latitudes (centered at Λ ~ > 67 ), consistent with an expansion-phase reduction in plasma pressure in the inner portion of the growth-phase plasma sheet. For each onset, expansion phase auroral activity begins several degrees equatorward of the poleward boundary of the plasma sheet. The larger substorm onsets are followed by a large poleward motion of the region of active aurora (The structure of the poleward-moving active aurora during the 12 January example is somewhat masked by higher than usual atmospheric scattering on this night). The two dimensional evolution of the aurora during a substorm expansion phase with an onset at ~1850 UT on 1 April 1986, as seen from the VIKING satellite located well above the auroral zone, is shown in Figure 15. The images in this figure show the localized onset of bright aurora (covering ~1 hr from MLT), and the subsequent azimuthal and polar expansion of the region of bright, expansion-phase aurora. The region of bright aurora that forms during the substorm expansion phase is known as the substorm bulge. Note that the 12 February photometer data in Figure 14 show a number of PBIs from ~0330 to ~0930 UT, which indicates that these disturbances can occur during the growth, expansion, and recovery phases of substorms as well as during quiet times. This example very nicely illustrates the much smaller scale size, but much more frequent occurrence, of PBIs as compared to substorms. The smaller scale size of PBIs can also be seen in the lower panel of Figure 15, which shows a VIKING image from the recovery phase of the substorm shown in the upper panels of the figure. This images shows relatively weak emissions throughout the decaying substorm bulge except for a series of small bright areas along the poleward boundary. Except perhaps for the bright areas before 22 MLT, which may be related to the westward traveling surge, these bright areas are PBIs that have formed along the poleward boundary of the auroral oval within the longitude range of the bulge. 19

20 The inward motion of the inner edge of the plasma-sheet during the substorm growth phase and the poleward displacement of this boundary following expansion-phase activity has a significant effect on magnetic fields observed at synchronous orbit. This is because synchronous orbit (r = 6.6 R E ) lies near the inner edge of the plasma sheet during times of enhanced convection, but earthward of the inner edge during quiet periods. Figure 16 (from McPherron and Manka, 1985) shows the typical relation between the magnetic field observed at geosynchronous orbit (by the GOES 3 satellite ) and the IMF (observed by IMP 8) and ground magnetometers during substorm periods. Times of the onsets of two large substorm expansion phases are indicated by vertical lines in the figure. The GOES data is shown for times surrounding the first onset, when the satellite was in the vicinity of magnetic midnight. Notice that the southward turning of the IMF at 1008 UT leads to an ~70 nt increase in the magnitude of the V (radial) component of the magnetic field observed on GOES. This increase occurred over the ~35 min growth-phase period from ~1015 to 1050 UT, and implies a large increase in cross-tail current near the equator on geosynchronous field lines. The H-component (parallel to the geomagnetic dipole and positive northward) of the magnetic field decreased as the V-component decreased and the magnetic field became increasingly tail-like. The decrease in magnitude of the V- component, and increase in the H-component, following the onset was more rapid than was the increase during the substorm growth phase. These changes reflect the return of the magnetic field to a more dipolar configuration. This dipolarization is associated with a decrease in the cross-tail current, and implies a decrease in equatorial plasma pressure. The expansion-phase decrease in cross-tail current and increase in westward electrojet current are connected via field-aligned currents to the ionosphere forming what is known as the substorm current wedge [e.g., Atkinson, 1967; Akasofu and Meng, 1969; Cummings and Coleman, 1968; McPherron et al., 1973]. This current wedge initiates within a localized region of the inner plasma sheet, and then expands in longitude both eastward and westward [Arnoldy and Moore, 1983; Nagai, 1987] and tailward [Jacquey et 20

21 al., 1991, 1993; Ohtani et al., 1992]. The westward expansion of the current wedge is associated with the westward traveling surge observed in the ionosphere and the tailward expansion of the current wedge is associated with the poleward motion of the region of active aurora in the ionosphere. Figure 17 illustrates the formation and expansion of the current wedge within the equatorial plane. Tail observations show that the expansion phase reduction of cross-tail current is often associated with a severance, and loss from the magnetotail, of the outer portion of the plasma sheet (r ~ > 25 R E ) [e.g., Moldwin and Hughes, 1993; Nagai et al., 1994, 1998; Machida et al., 1998]. However, signatures of this severance are absent during many substorms, even when there are tail observations within the longitude and radial distance range where signatures are expected [Lui et al., 1998, Machida et al., 1998; Pulkkinen et al., 2000]. It has not yet been determined whether severance of the outer portion of the tail occurs during all (or most) substorms and should thus be viewed as a fundamental aspect of the substorm process. However, whether or not such a severance generally occurs during the expansion phase, it is clearly an important aspect of substorm energy transfer when it does occur. The substorm onsets in Figures 14 and 16 for which IMF data is available show that onset occurs in association with an IMF change that is expected to lead to a reduction in large-scale convection. It is now known that there is a definitive statistical association between such IMF changes and substorm onsets, onsets occurring ~10 min after the estimated magnetopause impact of corresponding IMF changes [Lyons et al., 1997]. There is currently an important debate in the community concerning the extent to which substorms are a response to such IMF changes that occur after a ~ > 30 min period of enhanced convection. The consensus view at the 1999 GEM summer workshop was that ~ > 50% of substorms occur in response to IMF changes that lead to a reduction in convection; however whether the actual number is closer to 50% or closer to 100% has not been determined. That changes in large-scale convection also occur ~10 min after the impact of 21

22 appropriate IMF implies that it is the reduction in strength large-scale convection that leads to the onset of substorms. Such a reduction quite naturally accounts for the reduction in pressure, and thus the reduction in cross tail current, that forms within the near-earth plasma sheet during the substorm expansion phase. During the substorm growth phase, the inner boundary of the plasma sheet moves slowly earthward with the drift of plasma sheet particles. Considerable energy is imparted to the particles as they move earthward and to the nightside magnetic field as the cross-tail current increases in the near-earth region. After a reduction in convection, the plasma sheet must adjust to a tailward displacement of the inner plasma sheet boundary. Thus energy stored in the plasma between the old and new locations of the inner edge of the plasma sheet and energy stored in the magnetic field in association with the enhanced cross-tail in this region must be removed from the nightside [Atkinson, 1991]. Before and after a reduction in the strength of convection, azimuthal magnetic drift carries plasma sheet particles away from the night side. While convection is enhanced, particles removed by magnetic drift are replaced by new earthward convecting nightside plasma sheet particles. However, after a reduction in convection, this replenishment stops earthward of the new inner boundary of the plasma sheet. This will result in a significant loss of plasma, which has been proposed to lead to the large reduction in cross-tail current and the associated current-wedge formation that occurs within the near- Earth plasma sheet during the expansion phase of substorms[lyons, 1995]. A reduction in convection also reduces the energization of plasma sheet particles tailward of the new inner edge of the plasma sheet, which should lead to a reduction in cross-tail current throughout the tail plasma sheet. While expansion-phase reductions in cross-tail current tailward of the inner plasma sheet are significantly smaller than those within the near-earth plasma sheet, magnetic field measurements show that such reductions are a common feature of the substorm expansion phase [Caan et al., 1975]. 22

23 3. Convection bays Substorms typically occur after an ~30-90 min period of enhanced convection. Occasionally, however, a stable IMF can lead to periods of enhanced convection that persists for ~ > 2-3 hr. Such extended periods of relatively steady, enhanced convection are known as convection bays or steady magnetospheric convection periods. There can be considerable auroral zone activity during convection bays, and this activity is occasionally interpreted to be a signature of substorms. However, studies that have looked for signatures of substorm current wedge formation have found that such signatures are absent during convection bays, implying that substorms are not the primary cause of auroral activity during periods of steady, enhanced convection [Kokubun et al., Pytte et al., 1978; Yahnin et al., 1994; Lyons, 1996]. Because of the enhanced convection, the plasma sheet is expected to penetrate inward during convection bays as it does during the substorm growth phase. However, during convection bays, the enhanced plasma pressure and enhanced cross-tail current are expected to be maintained for extended periods in the inner plasma sheet without the large reductions that occur during substorms. This implies that the magnetic field in the region of the near-earth plasma sheet should be highly stretched throughout convection bay periods, and such a persistent, highly stretched field has been observed [Sergeev et al., 1994]. To illustrate the type of auroral-zone activity that occurs during convection bays, CANOPUS photometer and ground magnetometer data for convection bays periods on 24 January 1991 and 13 December 1990 are shown in Figure 18 [from Lyons et al., 1998]. IMF data was available for the first period, and shows significantly negative B z and/or large B y throughout most of the period shown in the figure. While the observed detail of the measured IMF may not accurately reflect the detailed structure that affected the magnetosphere, the measurements indicate that convection should have remained enhanced throughout most of the interval. The inner plasma sheet dynamics seen during these two intervals is significantly different than that seen in association with substorms. Rather than 23

24 moving equatorward and then poleward, as it does before and after a substorm expansion, the inner plasma sheet is observed to remain well equatorward of its quiet-time location from 01 UT (the beginning UT of the observations shown in the figure) until ~09 UT. The poleward displacement of the plasma sheet at later UT may be a MLT effect, since the duskward drifting plasma sheet protons do not have ready access to the dawn side (see Figure 8) and magnetic midnight at CANOPUS is ~0640 UT. Considerable auroral activity can be seen during the convection bay periods shown in Figure 18. However, other than a possible weak substorm with an onset at 0800 UT on 24 January 1991 (identified by a vertical dotted line in the figure), the poleward propagating regions of active auroral that occur during substorms are absent. Instead the auroral activity is dominated by strong PBIs, many of which maintain high intensity as they extend equatorward from the poleward boundary of the auroral zone (typically at Λ ~ 73 ) through nearly the entire plasma sheet to Λ ~ The start of several of these PBIs is identified in the figure by vertical dashed lines. It can be seen that it takes ~20-30 min for each PBI to extend through the plasma sheet. Taking a length for the plasma sheet in the equatorial plane of 80 R E, min corresponds to an average propagation speed of ~ km/s through the equatorial plasma sheet. This corresponds quite well with the typical speed of flow bursts in the plasma sheet, which are the probable cause of the PBIs. Notice that, in addition to the enhanced auroral emissions, the PBIs identified in Figure 18 are associated with significant (~ nt) perturbations in the magnetic X-component on the ground. Such perturbations are of the same magnitude as those associated with weak-to-moderate substorm. The above discussion suggests that PBIs are by far the dominant type of auroral disturbance during periods of steady, enhanced convection, and that substorms are uncommon. The PBIs for the events shown here are quite intense, and can be seen to propagate through nearly the entire latitudinal extent of the plasma sheet. However auroral zone activity during convection bay periods has not yet received a lot of attention, and the 24