75

1 Introduction

Passing the Parker's singular point, mostof the internal energy of the solar wind is con-verted into kinetic energy, and from there on,kinetic energy becomes dominant in the solarwind. The primary process in the formation ofthe magnetosphere is the confinement of theEarth's magnetic field by this single-sideddynamic pressure from the solar wind. Inideal magnetohydrodynamic (MHD) process-es, the solar wind plasma and magnetosphericplasma (if present) would not mix, resulting ina quiet magnetosphere. In such a case, themagnetopause would represent a tangentialdiscontinuity. However, in reality, the solarwind plasma penetrates into the magnetos-phere through non-ideal MHD processes suchas reconnection, filling the magnetospherewith plasma. Furthermore, the penetration ofmomentum and energy induces large-scaleconvection in the magnetospheric plasma,involving in the ionospheric plasma as well.

Auroras and radiation belts are generated as apart of this convection. The convection alsogenerates electric currents in accordance withthe laws of MHD, creating a large-scale cur-rent system in the magnetosphere ionospheresystem. The various changes in the magneticfield observed on ground or in the magnetos-phere are manifestations of this current system(See Chapter 4).

The penetrated plasma is not uniformlydistributed in the magnetosphere, but tends toaccumulate in specific regions such as cusps,the low-latitude boundary layer (LLBL), themantle, plasma sheets, and in ring currents.This magnetospheric plasma structure can beunderstood to some extent through the equilib-rium configuration based on single-particledescriptions and local MHD. Furthermore,since these magnetospheric plasma structurescorrespond to the current in each region, themagnetic field structure of the magnetospherecan be explained approximately in terms ofcurrents derived from single-particle descrip-

TANAKA Takashi

3-2 Generation of Convection in the Magne-tosphere-Ionosphere Coupling System

TANAKA Takashi

Based on the magnetosphere-ionosphere (M-I) coupling scheme, convection as a com-plex (compound) system is considered including the generation of plasma populationregimes in the magnetosphere. To guarantee the self-consistency, the MHD simulation isadopted to analyze the problem. In these considerations, primary elements that must be setto a self-consistent configuration are convection flows in the magnetosphere and the iono-sphere, filed aligned current (FAC) systems, ionospheric currents, energy conversionprocesses, and plasma pressure. Then, global current systems coupled with plasma popu-lation regimes are derived from the magnetohydrodynamic (MHD) force balance controllingthe convection. The magnetospheric model derived from this consideration is the closedmagnetosphere with open cusp. Based on the convection model proposed in this paper, asuggestion is given for the substorm models in the next decade that they must develop froma modular model to a globally self-consistent model.

Keywords Convection, Field-aligned current, Dynamo, Substorm, State transition

76

tions, such as magnetization current and driftcurrent (See Chapter 1).

It is apparent that the plasma structure andcurrent system are not independent of eachother; on the contrary, they are inextricablylinked. Moreover, they are closely connectedto convection. However, to further relate plas-ma structure and current system with convec-tion, the 3D self-consistent structure of themagnetosphere and ionosphere must be inves-tigated. Although the self-consistent structuremust not violate any laws of basic physics, thelaws of basic physics alone cannot completelyexplain the interactions between the threecomponents, since the interactions are depend-ent on topologies specific to the magnetos-pheric and ionospheric regions. Thus, simula-tion studies become essential for such exami-nations to make the difference clear betweenphysics and the earth and planetary sciences.

It is well known that the concept of con-vection starts with Dungey (1961) and Axfordand Hines (1961). Of the two, Dungey (1961)had wide appeal and is still adopted by manyresearchers as the basic concepts for manystudies. The discovery of field-aligned cur-rents (FAC) by Iijima and Potemra (1976)ought to have given new meaning to the con-cept of convection. However, it took an unex-pectedly long time before this discovery wasunderstood as the other manifestation of con-vection (Iijima, 2000). This was a result of alimited understanding of the magnetosphericclosure process of FAC and uncertainties in itstopology; this was in its contrast to the fairlyadvanced studies of ionospheric closure ofFAC resulting in geomagnetic variations.Recent MHD simulations have reproduced themagnetospheric closure process of FAC(Tanaka, 1995) and revealed the relationshipbetween convection and FAC, which hasadvanced our understanding of the interactionsbetween plasma structure, current system, andconvection. Understanding the structure ofthe convection system is now considered to bean essential element in the comprehension notonly of disturbance phenomena such as sub-storms (Akasofu, 1964) and magnetic storms,

but also of apparently unique phenomena suchas sudden commencement (SC) and thetaauroras.

2 Magnetosphere-IonosphereCoupling in a Convection Sys-tem

In a coupled system of the magnetosphereand ionosphere, magnetospheric convectionproceeds involving the ionosphere; this con-vection becomes the most fundamentalprocess in transferring the free energy of dis-turbances in the magnetosphere-ionospheresystem. When examining this system, it isnecessary to recognize which is more essen-tial, BV (magnetic field & velocity) or EJ(electric field & current), as basic parametersto describe the plasma dynamics. In studyingmagnetospheric physics focusing on convec-tion, which is a fluid-dynamic phenomenon,the BV paradigm should be selected, sincefluid dynamic descriptions using BV as basicparameters automatically ensure dynamicalself-consistency. In this case, distortions arecreated in the magnetic field as it is carriedalong with the fluid, which in turn inducescurrents to affect the fluid motion. Therefore,in this paradigm, J is the result of changes in B(Parker, 2000).

In contrast to the BV paradigm of thefluid-dynamic model, in particle modeling,motions of particles generate electric currentsand fields, which then induce changes in themagnetic field that in turn controls particlemotion. Therefore, EJ are more basic param-eters (Lui, 2000), and J becomes the cause ofchanges in the magnetic field. If current dis-ruption and reconnection are assumed to bethe basic processes in magnetospheric distur-bances, the EJ paradigm should be selected toexamine the physics of the magnetosphere,since they cannot be incorporated into the BVparadigm. However, at present, some difficul-ties remain in satisfying the requirements ofglobal self-consistency in the EJ paradigm.

Journal of the Communications Research Laboratory Vol.49 No.3 2002

77

2.1 The Slab Model and the Genera-tion of a Global Current System

As the basis for understanding how mag-netosphere-ionosphere coupling constructs aconvection system, let us consider the slabmodel (Fig.1). In Fig.1, M, F, and E representthe magnetosphere, ionospheric F layer, andionospheric E layer, respectively. The entiresystem is coupled by a magnetic field. Thefigure represents a moment at which shearmotion is generated in the M region and thefront portion is just beginning to move. Theslab model is a simple analogy, but it is well-suited to explain why the generation of a glob-al current system is unavoidable when thevelocity (electric field) is projected from themagnetosphere onto the ionosphere. Themagnetospheric convection transfers the mag-netic field according to the frozen-in principle.However, if no ionospheric convection exists,growing distortions perturbations are generat-ed in the magnetic field, generating magnetictension that accelerates the ionosphere. Inreaction, the magnetosphere is decelerated,and the convective motion is controlled. The

actual solution is the point at which this inter-action is balanced. The characteristic of theslab model is that it considers convectionwithin the BV paradigm. In this model, theelectric current is a result of the distortions inthe magnetic field, and is not the cause ofchanges in the magnetic field. Furthermore,the electric field is included in the schemeonly when it is used for supplementary expla-nations. As a result, FAC, ionospheric current,and magnetospheric current form a closed cir-cuit in this model.

In the BV paradigm, current is not a mainparameter for the MHD process. However,the existence of the ionospheric E layer makesit difficult to exclude currents when discussingconvection. As evident from the slab model,currents are continuous, and the FAC accom-panies ionospheric closure on one hand andmagnetospheric closure on the other. There-fore, it is essential to clarify these closureprocesses in discussing the mechanics of con-vection. For a stationary convection, theremust be a force that counters the J×B force inthe magnetosphere. The work performed bythis force against the J×B force keeps J •Enegative. This action is referred to as thedynamo. In this process as well, considerationof the current aids in understanding themechanics of the magnetospheric convection.In the dynamo region, V and J×B must beanti-parallel. The force countering J×B(dynamo driver) determines the power of thedynamo, but the required force is dependentnot only on the convection velocity V but alsoon the ionospheric conductivityΣ.

2.2 Current System and the Projectionof Velocity and Electric Field

The explanation that magnetosphere-iono-sphere coupling in a convection system is real-ized through the projection of the magnetos-pheric electric field onto the ionosphere mayseem convincing. However, if the coupling isexplained as a projection of velocity from themagnetosphere onto the ionosphere accordingto the frozen-in concept.

TANAKA Takashi

M, F, and E represent the magnetosphericdomain, the ionospheric F layer, and theionospheric E layer, respectively.

Slab model of convectionFig.1

78

then a simple question arises as to what forcesacted in the system to produce such results.This reflects the question of whether to choosethe BV or the EJ as the basic parameters ofplasma. When E is used in a fluid-dynamicdescription such as convection, the equation ofmotion becomes hidden, turning attentionaway from mechanics as a balance of force.Such confusing perceptions frequently ariseunder the EJ paradigm, and so the BV para-digm is considered more appropriate in thediscussion of magnetosphere-ionosphere cou-pling, as in the slab model. However, currentscannot be totally ignored, due to the presenceof the ionospheric E layer. This is becauseMHD does not apply to the E layer; instead,an analogy to an electric circuit (EJ paradigm)is applied. Therefore, it is essential to includethe effects of current in the discussion.

Sonnerup (1980) examined a convectionmodel of velocity projection onto the iono-sphere by generating FAC using viscous forcefor dynamo action. The topology of thismodel more closely resembles the actual mag-netosphere than does the slab model, and isinstructive in that it explains self-consistencybetween the magnetosphere and the iono-sphere. However, the convection is not recon-nection driven, and thus does not fully repro-duce the actual state of the magnetosphericconvection. The Rice Convection Model,which treats only the inner magnetosphere,generates FAC based on drift motion of parti-cles caused by the electric field, and projectsthe convection onto the ionosphere (Harel etal., 1981). Therefore, the Rice ConvectionModel falls under the EJ paradigm, althoughit assumes a boundary condition for E.

MHD simulations are required to generateFAC in a magnetospheric convection modelthat is closer to the actual state and to projectthe convection onto the ionosphere. However,there is a large gap in scale between the mag-netosphere and the ionosphere. To overcomethis gap and to project the convection onto the

ionosphere numerically, MHD simulationsusing unstructured grids are being studied(Tanaka, 1995; Siscoe et al., 2000; Gombosi etal., 2000). The section below will mainlyfocus on the results of unstructured-grid MHDsimulations.

3 Ionospheric Closure of FAC andIonospheric Convection

The process of ionospheric closure of FACis a classic problem in geomagnetism, and hasbeen a subject of intense study since Birkeland(Kamide et al., 1996; Cowley, 2000). Whenionospheric convection is described usingpotential electrical fieldφ, 2-dimensional elec-trical conductivity tensorΣ, and FAC J‖, thenionospheric closure can be expressed as

which essentially represents an energy loss

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Relationship between velocity andthe electric field inside plasma

Fig.2

The top and bottom panels are for rotation-al motion and divergent/convergentmotions, respectively.

79

process. Using potential to express the elec-tric field is equivalent to assuming that themagnetic field is stationary (∂B/∂t = 0) andthat the motions in the ionosphere are restrict-ed only to eddy motion (divV = 0). The toppanel in Fig.2 illustrates the relationship forthe above case. The bottom panel is providedfor comparison with a case of inductive elec-tric field. In the latter case, divergent and con-vergent motions are generated, and a station-ary magnetic field cannot be maintained.

It is widely known that whenΣis constant,only the Pedersen current must be accountedfor in current closure, since the Hall current isclosed within the ionosphere. SinceΣis notconstant due to EUV ionization and ionizationby particle precipitation, the Hall current mustalso be considered in ionospheric closure.Furthermore, it is known that non-uniformΣinthe auroral oval results in Cowling conductivi-ty, and this forms one of the central conceptsgeomagnetism: the electrojet.

The relationship between convection andFAC in ionospheric closure process can beapproximated to a condition that FAC is pres-ent in the center of shear in the convection.This situation can be understood from the rela-tionship between convection potential and

FAC presented in Fig.3. The left panel inFig.3 shows the case of uniformΣand theright panel illustrates the case of non-uniformΣcaused by EUV ionization and ionization byparticle precipitation. In these panels contoursrepresentφ, and the color scale represents J‖.WhenΣ is uniform, the region 1 currentapproximately corresponds to the shear creat-ed by the tailward flow in the polar caps andthe sunward flow in the auroral oval, andregion 2 current approximately corresponds tothe shear created between sunward flow in theoval and the stationary region at lower lati-tudes (Cowley, 2000; Tanaka, 2001). Howev-er, whenΣis non-uniform, the relationshipbetween convection shear and current isslightly displaced, because charge accumula-tion and resulting electrostatic field occur inthe ionosphere in order to maintain currentcontinuity. Thus, it can be seen that the projec-tion of velocity (electric field) is not a one-wayprocess from the magnetosphere to the iono-sphere; it also works in the reverse direction.

4 Magnetospheric Closure of FAC

The magnetosphere has a complex topolo-gy, and the magnetospheric closure of FAC in

TANAKA Takashi

Ionospheric potential and FACFAC is represented by the color scale, and potential is represented by contours. The plots on the left andright are for uniform and non-uniform electric conductivity of the ionosphere, respectively.

Fig.3

80

the slab model is not realistic. However, theprinciple can still be similarly applied to thereal magnetosphere. In the actual system, themagnetosphere is not necessarily incompress-ible, but near the ionosphere, there is a low-beta region that is considered to be incom-pressible. However, as in the slab model, themagnetospheric convection carries the mag-netic field together with it according to thefrozen-in concept, and then a kink structurewill be produced at the boundary between thelow-beta region, which is propagated throughthe low-beta region (Fig.4). This process rep-resents a case of magnetosphere-ionospherecoupling by FAC similar to that seen in theslab model. In this process, charge accumula-tion by FAC occurs, decelerating convection,because it is positive at low potential regionsand negative at high potential regions. There-fore, coupling with the dynamo is essential todrive convection against charge accumulation.These principles are similar to those of theslab model, and the driven convection mustaccompany the dynamo and FAC. Thus, theFAC path that connects the ionosphere withthe dynamo must be examined first (Stern,1983).

4.1 Magnetospheric Convection andthe Dynamo

It is clear from the above discussions thatthe dynamo must be included for magnetos-

pheric closure of FAC. However, energy mustbe supplied to form the dynamo. In this caseconvection is steadily driven by the energy-supplying process incorporated into magnetos-pheric convection; this is the mechanism thatsupplies electromagnetic energy to the iono-sphere. For this process, mechanical energy(kinetic or internal energy) must be convertedinto electromagnetic energy. However, toinvestigate this dynamo action in the actualmagnetosphere, the closure path of FAC in themagnetosphere must first be identified.

In recent years, MHD simulations havebeen undertaken to calculate how FAC isclosed in the magnetosphere (Tanaka, 1995;Siscoe et al., 2000). Fig.5 shows an exampleof the calculation for region 1 current closurein the magnetosphere. The region 1 current,which plays a central role in convection, clos-es at higher latitudes of the cusp. The FACfollows the magnetic field line in the low-betaregion, but does not necessary do so at higheraltitudes outside of the low-beta region (Tana-ka, 1995). Studying the J•E distribution iseffective in examining the closure process ofthe region 1 current at latitudes higher than thecusp. Fig.6 shows the J•E distribution at the12-0 o'clock meridian and the equatorialplanes. In this figure, the interplanetary mag-netic field (IMF) has a southward orientation,corresponding to the substorm growth phase.At the bow shock, J•E is negative, indicatingthat the magnetic field is compressed by solarwind that has lost its kinetic energy accordingto deceleration from supersonic to subsonicvelocity. At the magnetopause, J•E is posi-tive, which corresponds to a condition inwhich energy is supplied to the plasma bymagnetic tension resulting from the sharpbend in field lines at the reconnection.

In the region around the mantle and lati-tudes higher than the cusp, J•E is negative,and it can be seen from Fig.5 that this is thesource region of the region 1 current. It canbe observed from Fig.6 that the magnetic fieldlines in regions where J•E is intensely coloredconstitute the cusp (1) in both the magnetos-phere and the ionosphere. Energy conversion

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Schematic diagram of magnetos-phere-ionosphere coupling

Fig.4

The dashed and solid arrows represent thedirections of kink propagation and FAC,respectively.

81

is not observed along the magnetic field lineextending from the polar cap to the lobe (2).The primary factor in the formation of themagnetosphere is solar wind dynamic pres-sure. At this stage, the magnetosphere isclosed. In Fig.5, the magnetosphere is open atthe cusp due to reconnection, indicating thatforce is exerted at the cusp and that energy

conversion takes place there. The entire mag-netosphere does not open, as described inDungey. Therefore, there is a clear distinctionbetween the cusp and the polar cap.

On the other hand, region 2 current, repro-duced by simulation, is closed at the inneredge of the plasma sheet (Tanaka, 1995). Thisis the topology of a partial ring current, and isconsistent with the topology predicted in pre-vious studies (Cowley, 2000). At the inneredge of the plasma sheet, current is perpendi-cular to the magnetic field, and it is convertedinto J‖in the region where∇P is present in thelongitudinal direction of the inner magnetos-phere; it then flows into (or out of) the iono-sphere. The dynamo of the region 2 currentexists in this region, where∇P is present inthe longitudinal direction. In contrast, J•E ispositive in the plasma sheet. FAC having thisstructure corresponds to that calculated in theRice Model.

4.2 Energy Conversion by ConvectionLet us re-investigate the force balance in

the convection crossing the cusp and the polarcap (Fig.7). Let us assume here that the veloc-ity inside the magnetosphere is small (the rea-son for this will be given later), and that

On the lower-latitude side of the cusp,convection flows from the low-pressure to thehigh-pressure side. Magnetic tension forcesmakes it possible for the convection to flow inthe direction opposite the pressure gradient(electromagnetic pumping). The work here isdone by the electromagnetic force on the plas-ma, and therefore, this can indeed be consid-ered electromagnetic pumping. On the higher-latitude side of the cusp, this relationship isreversed and the plasma does the work anddrives the dynamo. Thus, work dW done bythe magnetic field is

therefore, when convection V is parallel to

TANAKA Takashi

Magnetic field (top) and region 1electric current (bottom) of the mag-netosphere shown for the 1/4 region of positive y and z values

Fig.5

The conditions correspond to the growthphase.

Magnetic field and J•E distributionshown for the 12-0 o'clock meridianand equatorial planes

Fig.6

The ionospheric electric potential is repre-sented by the color scale, and the circlesrepresent 60˚, 70˚, and 80˚ latitudes.

82

J×B, then work is done by the magneticfield, and pumping occurs with positive J•E ispositive. In contrast, when convection V isanti-parallel to J×B, then work is done on themagnetic field, and the dynamo is realized atnegative J•E (Tanaka, 2000a).

The dynamo may also be generated by theacceleration and deceleration of convection(Fig.8, top). In this case, J×B balances withdynamic pressure. However, the magnetos-pheric convection in this process contains acompressional component, making it difficultto realize self-consistent connection with low-beta shear motion generated by ionosphericFAC and ionospheric convection. As can beseen from the bottom panel in Fig.8, constant-velocity convection can more easily satisfyself-consistency throughout the entire system,because it contains no compressional compo-nent and connects smoothly with ionosphericconvection. In this case, the presence of the

cusp is essential for a dynamo driven by inter-nal energy. To further confirm this theory, acomparison of internal energy and kineticenergy is presented in Fig.9. Generally, kinet-ic motion inside the magnetosphere is small,and the conversion of energy by accelerationand deceleration does not play a major role.This situation becomes a latent factor in thesystem, which leads to a change due to statetransition; this point will be explained later infurther detail.

By applying similar logic to the plasmasheet, it becomes clear that convection repre-sents an energy conversion system. The rela-tionship in Eq. (3) also holds in the plasmasheet, and the magnetic tension is in balancewith the pressure gradient. Convection u isparallel to J×B and anti-parallel to －∇P,and electromagnetic energy is constantlybeing converted into mechanical energy. As aresult, the entire plasma sheet acts like apump; this structure causes internal energy to

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Balance of forces at various positionsin the magnetosphere

Fig.7

Schematic diagrams showing theenergy conversion process in thecusp

Fig.8

The top panel represents a case in whichkinetic energy predominates and the bot-tom panel represents one in which internalenergy predominates.

83

accumulate constantly at the inner edge.Using this internal energy as the source, con-vection extending from the inner edge of theplasma sheet to the dayside drives the region 2current. This convection V is anti-parallel toJ×B and parallel to －∇P.

Where is the energy to drive the plasmasheet pump transported from? Clearly, it isfrom theθcurrent system in the magnetos-

pheric tail. The energy driving theθcurrentsystem is exactly the same as that of theregion 1 current, and is generated in the man-tle. It can be said that theθcurrent systemtransports the electromagnetic energy generat-ed in the mantle to the plasma sheet.

As evident from the above discussions,magnetospheric convection is an energy con-version system that converts solar wind ener-

TANAKA Takashi

Distributions of kinetic and internal energies in the magnetosphere for (from top to bottom)the quiet state, the substorm growth phase, and the substorm expansion phase

Fig.9

84

gy into electromagnetic energy and succes-sively converts it to internal energy, and againconverts it back to electromagnetic energy,maintaining self-consistency on the whole.

4.3 Coupling Between Convectionand Plasma Structure

In the magnetosphere physics discussed atpresent, the generation of a plasma region inthe magnetosphere (including cusps and plas-ma sheets) is the result of localized mechani-cal balance and has no direct ties with globaldynamics such as convection and the dynamo.The generation of plasma regions is simply atopological problem, and is not a dynamicone. However, in a system in which internalenergy is dominant relative to kinetic energyand the dynamo is driven by internal energy,convection and the magnetospheric plasmaregion become coupled, and structures such ascusps and plasma sheets become dynamic.Without the cusps and plasma sheets, the con-vection system cannot achieve self-consisten-cy. The cusps are essential to the generationof region 1 current and plasma sheets forregion 2 currents.

Convection, current systems, and magne-tospheric plasma regions are all coupled, andcannot be discussed independently. Fig.10shows a composite system in which all threeare self-consistent. Convection of the sub-storm growth phase driven by dayside recon-nection, SC convection driven by compres-sion, and convection at the onset of substormsaccompanying dipolarization are all generatedin this system, and although they have distincttopologies, they share a common principle inthe realization of global self-consistency.

Convection is a phenomenon in a topolog-ical system. Earth and planetary sciencescharacteristically deal with topological sys-tems, rather than with universal laws ofphysics in a flat space. A discovery inphysics must be an explanation of a phenome-non based on universal principles, involvingthe reduction of the phenomenon to its basicelements. In contrast, given the significanceof topology in earth and planetary sciences,

phenomena cannot be reduced to their compo-nent parts without the loss of their essentialcharacteristics. Earth and planetary sciencesrepresent the physics of composite systems,which treat the systems as is, recognizing thattopology of these systems is tied to self-con-sistency. Another way to define such a com-posite system is to refer to it as a system thatallows state transitions.

5 Convection Structures under More General Conditions

Up to this point, the magnetosphere-iono-sphere convection system has been discussedmainly assuming a due southward IMF. Whatare the characteristics of convection in a moregeneral case? When dealing with magnetos-phere-ionosphere disturbance phenomena suchas substorms, an IMF oriented due south ornorth is generally assumed. However, suchorientations rarely exist in nature. To allowfor more generalized discussion, it is impor-tant to consider cases of an IMF with oblique-

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Schematic diagram of the compositesystem

Fig.10

85

ly southward or northward orientations.

5.1 Obliquely Southward IMFRotation in the Y-Z plane of IMF has a

significant effect on dayside reconnection thatis the first phenomenon induced after contactwith the magnetosphere. There are presentlytwo types of structures proposed for daysidereconnection: antiparallel merging and com-ponent merging (Crooker, 1990). The resultsof MHD simulations support the former(Ogino et al., 1986; Tanaka, 1999). Fig.11shows an MHD simulation result of daysidereconnection during obliquely southward IMF.The broken lines represent reconnection lines.The reconnection lines are orthogonal to theIMF at a proper distance from the center,which is true in the case of component merg-ing too. However, the lines converge at thenorth and south cusps near the Earth, and thisstructure is characteristic of antiparallel merg-ing. The first significant feature in Fig.11 isthe lack of bilateral symmetry of the closedmagnetic field. This implies that not onlyIMF is distorted by draping immediatelybefore reconnection, but also geomagneticfields involved with the reconnection are alsodistorted (Cowley, 1973). This is normallyreferred to as the presence of a diffusionregion.

Another peculiar structure can be seen inFig.11. Relatively far from the center, IMFstrucutures (such as 1, 2, 6, and 7) can recon-nect on a one-to-one basis with the geomag-netic field. However, near the center, all ofthe IMF structures 3, 4, and 5 must reconnectwith a single magnetic field line at the cusp.In other words, in the central region sand-wiched between the north and south cusps, adead zone of reconnection is produced, and allIMFs impinging on this dead zone are put tothe cusp. To date there have been no studiesinvestigating this structure in detail. This maybe related to the FTE or to the unsteadydynamics of the cusp ionosphere.

The effect of oblique IMF, or By, is mostevident in the structure of the ionospheric con-vection cell. This problem has been investi-

gated using various methods, such as satellite,radar, and geomagnetic observations (Heppnerand Maynard, 1987; Weimer, 1995; Ruohonie-mi and Greenwald, 1996). Fig.12 shows theasymmetry of convection cells due to the Byeffect (Tanaka, 2001). The cell structure dis-plays antisymmetric patterns with respect tothe morning and evening sectors for By+ andBy－, although the two are not perfectly mir-ror-symmetrical. First, convection near thecusp heads duskwards and dawnwards forBy－and By+, respectively. This can be attrib-uted to tension due to the By effect. From theright panel in Fig.3, it can be seen that corre-sponding to the condition in which convectionat the cusp is duskward; the morning sectorregion 1 current stretches along the higher-lat-itude side of the evening sector region 1 cur-rent and extends to the evening sector. Thedynamo is thought to lose its symmetry and toacquire a 3D structure. Many problemsremain to be resolved, such as the 3D structureof the dynamo and the relationship betweenthe morning sector region 1 current structureextending to the duskside and the cusp FAC.

Next, in the central part of the polar cap,convection heads straight in the 12-0 o'clockdirection for By－, whereas it is oblique forBy+ in the 9-21 o'clock direction. The posi-tions of the evening and morning sector cellsare shifted to the dayside and nightside,

TANAKA Takashi

Dayside reconnection for southwardIMF

Fig.11

86

respectively, for both By－ and By+, display-ing a clockwise rotation. On the whole, theevening sector cell is dominant. This resultsfrom charge accumulation along the termina-tor line which prevents discontinuity in theHall current arising from the difference inconductivity between the day and night sec-tors, to modulate the potential distribution(Atkinson and Hutchison, 1978). However,this is insufficient to generate a clockwiserotation, which is considered to be a manifes-tation of a more global balance (Tanaka,2001).

In the night sector, convection is stronglyaffected by the auroral oval. Near midnight,the equipotential lines in the oval are deflecteddawnward. This corresponds to the generationof Cowling conductivity in response to thewestward electrojet (WEJ). Near midnight,the electrojet is always westward, and so theequipotential lines cannot be deflected in theopposite direction (duskwards). Therefore,the convection does not display mirror symme-try for By. For By+, the equipotential linesconcentrate in the morning sector at the cusp,and unless these lines are returned to theevening sector in the polar cap, dawnwarddeflection of the equipotential lines in the

auroral oval and the WEJ cannot be produced.This results in oblique equipotential lines with-in the polar cap. These structures are involvedin the generation of the Harang discontinuity(Harang, 1946). Within the WEJ, Harang dis-continuity is generated by shear at the equato-rial edge of the auroral oval, and the eastwardelectrojet (EEJ) is generated by shear at thepoleward edge (Amm et al., 2000). It has beengenerally accepted that the Harang discontinu-ity is generated by magnetospheric convections(Erickson et al., 1991), but ionospheric con-vections are equally likely to be the cause.

The By component during southward IMFis involved in the generation of transient phe-nomena in the cusp ionosphere (such as mag-netopause FTE and the related auroral patch,the dayside auroral transient) (Newell andSibeck, 1993). However, some schools ofthought prefer a straightforward interpretationof these phenomena in terms of pulsed recon-nection, without giving much consideration tothe possibility that the dayside reconnectioncontaining By may produce a complex magne-tospheric structure (Lockwood et al., 1995).Since the ionospheric convection is incom-pressible, the fast flow at the cusp presents aconcentrated region of equipotential lines.

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Electric conductivity and potential are represented by color scale and contours, respectively. The leftand right plots are for negative and positive By, respectively.

By effect on ionospheric potential distributionFig.12

87

This can be realized either through the genera-tion of a small vortex or, as in Fig.12, bybending the equipotential lines at the cusp inthe east-west direction. With pulsed reconnec-tion phenomena, the former is more likely.However, problems relating to field-alignedirregularities (FAI) and the mechanism ofdynamo generation remain unresolved in thiscase. Fluctuation phenomena in the cusp ion-osphere should be viewed differently－i.e.,ought not to be seen as a strictly localizedproblem－when considering the cusp's role asan element of magnetospheric convection (asshown in Fig.8) and when considering recon-nection involving By.

The effect of By is significant in daysidereconnection and ionospheric convection. Onthe nightside in contrast, oblique IMF (largeBy) does not seem to affect the tail structuregreatly.

5.2 Obliquely Northward IMFA due-northward IMF produces a short

magnetotail and 4 cell convection in the iono-sphere (Gombosi et al., 2000). The daysidereconnection occurs simultaneously in bothhemispheres on the higher-latitude side of thecusp between the IMF and the closed magnet-ic field of the short tail. As a result, a closedmagnetic field filled with solar wind plasmaappears near noon, and another field (isolatedfrom the geomagnetic field) appears in the tail.The closed magnetic field is assimilated by themagnetospheric low-latitude boundary layers(LLBL; Song and Russell, 1992), drifts alongthe flanks to the tail (convection), and thenreturns to the reconnection point (Troshichev,1990). Due to this structure, a due-northwardIMF generally creates a thick LLBL.

In the case of obliquely northward IMF,the convection structure is more complex,with a magnetospheric convection composedof a lobe cell and a merging cell and an ionos-pheric convection composed of a round celland a crescent cell (Crooker et al., 1998;Tanaka, 1999). Here, the lobe and the merg-ing cells are named after their physical struc-tures. The former convection is confined to

the interior of the lobe and does not reach theplasma sheet (Fig.13) and remains as perpetu-ally open field lines (Type 1 open magneticfield line). The latter convection reaches theplasma sheet, and through a reconnectionprocess in the tail, experiences an interval ofclosed magnetic field lines (Type 2 & Type 3open magnetic field lines) (Fig.14 & 15). Incontrast, the round and the crescent cells ofthe ionosphere are only classified according toexternal shape. Physically speaking, theround cell consists of both the lobe cell (Type1 open magnetic file line) and the merging cell

TANAKA Takashi

Lobe cell for obliquely northward IMF(Type 1 magnetic field)

Fig.13

Merging cell for obliquely northwardIMF (Type 2 magnetic field)

Fig.14

88

(Type 2 open magnetic line), while the cres-cent cell consists only of the merging cell(Type 3 open field line).

In Fig.13, the lobe cell is initiated byreconnection of the IMF and open magneticfield line 4. This results in the generation ofline 1, which is a Type 1 magnetic field line.The reconnection produces an isolated mag-netic field line "d" along with line 1. Howev-er, line d is positioned away from the Earthand is not a part of the convection. Line 1remains as an open magnetic field line, con-vects to positions 2 and 3, and finally returnsto the original position at 4. The Type 1 mag-netic field line always drifts near the magne-topause and does not reach the plasma sheet.As seen from Fig.13, the source of Type 1magnetic field lines shifts position in the cen-tral part of the round cell inside the ionosphere(Crooker et al., 1998).

The merging cell is initiated by the recon-

nection between the IMF and a closed magnet-ic field line. Here, the magnetic field lines ofthe merging cell can be classified as Type 2 onthe round cell and Type 3 on the crescent cellin the ionosphere (Tanaka, 1999). The appar-ent differences between these types (viewed interms of magnetospheric convection) require asomewhat complicated explanation. Type 2,shown in Fig.14, has the same basic convec-tion structure as Type 1 at positions 1, 2, 3,and 4. However, at position 4, the field linebecomes a closed magnetic field line (C4)through tail reconnection. It then returns toC1 as a closed magnetic field line. It can beseen from Fig.14 that the source of Type 1magnetic field lines drifts along the peripheryof the round cell in the ionosphere (Crooker etal., 1998).

Type 3 magnetic field lines display aneven more complex structure, and in additionto the open-closed structure of Type 2 fieldlines, it features a distinct cell switchoverprocess (Fig.15). When the magnetic fieldline at 1 created by the dayside reconnectionarrives at position 2, it reconnects again withC2' and becomes line 2'. Along with thisswitchover, the footpoint in the ionospherealso switches from C2 to C2'. Thus, the con-vection from position 1 to 2 can be consideredto relay the open magnetic field line createdby reconnection to the crescent cell, producinga third ionospheric cell type (in addition to theround and crescent cell types), which may bereferred to as the "exchange" cell (Tanaka,1999). This structure, present at around 9o'clock, is a 4-layer structure when seen fromthe FAC. Whereas Ohtani and Higuchi (2000)explained a 4-layer structure based on viscouscells, the exchange cell may be considered analternative explanation of the 4-layer struc-ture. The Type 3 magnetic field line moves toa position nearest to the plasma sheet, andreconnects with the Type 2 line in the oppositehemisphere inside the plasma sheet. The Type3 field line displays strange motion inside themagnetosphere. From position 2' to 3 inFig.15, the Type 3 line progresses from theevening sector to the morning sector, but

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Merging cell for obliquely northwardIMF (Type 3 magnetic field)

Fig.15

89

returns to the evening sector before tail recon-nection (Tanaka, 1999).

As a result, a magnetotail with a twistedmagnetic structure is produced during oblique-ly northward IMF. The nightside reconnectioninside the merging cell takes place under thistwisted magnetic field structure. Since twotwisted magnetic fields cannot satisfy theantiparallel condition with respect to eachother, they undergo an untwisting process at adistance and then reconnect. This is the causeof the strange motion observed with the Type3 magnetic field line. In this way, the obliqueIMF model can explain the formation of thedistant-tail neutral line. This is an importantinitial condition for the substorm growth

phase.The convection flow during a northward

IMF, as explained above, is slower than thatfor a southward IMF. This is because, gener-ally, IMF connecting to the geomagnetic fieldthrough reconnection structures is decreased,and results in an extremely suppressed pene-tration of electric field from the solar wind.From Figs.13, 14, and 15, it can be observedthat IMF with larger y-coordinate values con-vects antisunward in the ionosphere deeper inthe morning sector, and produces an electricfield in the dawn-to-dusk direction in the iono-sphere. However, this interpretation is basedon the EJ paradigm, and does not apply toFig.10. It is necessary to explain this convec-

TANAKA Takashi

Formation of a theta auroraFig.16The plots proceed in the following chronological order (0 to 25 min): upper-right, upper-left, lower-right,and lower-left.

90

tion in terms of the dynamo and FAC, even fora northward IMF. A week convection undernorthward IMF should be interpreted as aresult of insufficient release of tension, as seenin Fig.8, which leads to less increased internalenergy in the cusps.

5.3 The Effects of IMF By and ThetaAurora

One of the most interesting phenomena ofthe magnetosphere during northward IMF isthe theta aurora. To study the theta aurora, itis necessary to have a firm understanding ofthe magnetospheric structure during north-ward IMF. A theta aurora appears after a largeshift in the By component involving a sign

change, during relatively long durations ofnorthward IMF having absolute value exceed-ing 10 nT (Cumnock et al., 1997). Fig.16, 17,and 18 show the results of simulation underthe above conditions, with a negative-to-posi-tive shift in By. Fig.16 shows the develop-ment of a pressure distribution in the iono-sphere projected from the magnetosphere. Inthe beginning, an auroral oval expands in themorning sector, and eventually, a gap appearsbetween the primary oval and this extendedauroral oval. The detached part of the auroraloval drifts further duskwards and becomes atheta aurora. The behavior of the auroral ovalduring this process agrees well with actualobservations (Cumnock et al., 1997).

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Changes in the plasma sheet structure during theta aurora formationFig.17The plots are for the same time points as in Fig. 16.

91

How does the magnetospheric structurerespond to these changes in the ionosphere?Fig.17 shows the temperature distribution of adeveloping plasma sheet at various stages onthe y-z plane at x = 15Re. This figure corre-sponds to the sight line from the tail to theEarth. In the beginning, the plasma sheet is

twisted in the counterclockwise direction, cor-responding to negative By. Some time afterBy turns positive, the twisted structure weak-ens and the plasma sheet thickens. As timeprogresses, the plasma sheet begins to twist inthe opposite direction, and a kink appearsbetween the central part and the edge. This

TANAKA Takashi

Magnetic field topology during theta aurora formationFig.18Black, red, and blue dots in the ionosphere show the sources of By-, By+, and closed magnetic fields,respectively. Circles represent 60˚, 70˚, and 80˚ latitudes. The lower-right-hand plot in Fig.16 is shownfor comparison.

92

kink, projected onto the ionosphere, is thetheta aurora. At this stage, the northern lobeseparates into the morning and evening parts.In the northern hemisphere, the former con-sists of a lobe derived from a new IMF withpositive By, and the latter consists of a lobederived from the old IMF with negative By. Itcan be clearly seen from Fig.18 that there arecounterparts to this structure in the iono-sphere. Fig.18 presents the magnetic fieldstructure observed from the tail, showing withthis footpoint in the ionosphere. In the morn-ing sector, there is a new polar cap with amagnetic field connected to positive By, andin the evening sector, there is an old polar capwith a magnetic field connected to negativeBy. Note here that the closed magnetic fieldlines are accumulated above the theta aurora,extending to the polar cap.

The various morphologies obtained in sim-ulations can be explain from the convectionstructure during northward IMF. Let us exam-ine the convection in the northern hemisphere,assuming that the By has shifted from negativeto positive. In the beginning, there is a cres-cent cell in the evening sector and a round cellin the sector from the center of the polar capto the morning (Tanaka, 1999). The Type 2magnetic field lines of the merging cell, whichhave opened at the cusp, drift in the counter-clockwise direction over the round cell. Uponreaching nightside, the Type 2 magnetic fieldlines reconnect with the Type 3 magnetic fieldlines that have drifted over the crescent cell inthe morning sector in the southern hemisphere(same as the structure shown in Fig.15, butreversed in the N-S hemisphere). The result-ing closed magnetic field line returns to thereconnection region near the cusp, passingthrough the magnetosphere in the morningsector. However, when the By switches topositive, a new round cell drifting in a clock-wise direction appears at the central part of thepolar cap. Thus, the circuit to return to themorning sector is blocked for the closed mag-netic field lines created in the old convection.In the new convection, the crescent cell isformed in the morning sector and drifts in the

counterclockwise direction, and this alsoblocks the return path of the closed magneticfield lines created in the old convection.Therefore, the closed magnetic field lines cre-ated from the lobe magnetic field with nega-tive By accumulate in the nightside, forming atheta aurora.

6 Behavior of a Transient Convec-tion

The study of the structure of the magnetos-phere-ionosphere convection system forms thebasis of all studies of magnetosphere-iono-sphere phenomena. All disturbance phenome-na are derived from this fundamental struc-ture, and one can achieve a significant under-standing of these phenomena by regardingthem as manifestations of variations in con-vection. Here, we will examine how the con-vection structure consists of the incompress-ible ionosphere and compressible magnetos-phere and how the magnetosphere respondsmost efficiently to changes in solar wind con-ditions while simultaneously maintaining self-consistency.

6.1 Reconstruction of the IonosphericConvection

Convection is often explained as beinginduced by the tailward motion of the magne-tospheric plasma by the tension from IMFanchored to the solar wind, and by transport-ing plasma sunwards by the shortening ofmagnetic field lines after nightside reconnec-tion events. This is because, for a long time,these ingenious ideas on the magnetospherepresented by Dungey led many to accept themunconditionally. The well-known problemwith this interpretation, leading to inconsisten-cies, is the problem of different response timesof the magnetospheric and ionospheric con-vections (Ridley et al., 1997; Khan and Cow-ley, 1999; Murr and Hughes, 2001).

When the IMF switches northward tosouthward, ionospheric convection becomesstronger near the cusp, which is consistentwith the conventional view that the cusp exists

Journal of the Communications Research Laboratory Vol.49 No.3 2002

93

below the dayside reconnection. According toDungey's image of convection, it takesapproximately 30 minutes for fast magnetos-pheric convection to reach the tail, and thusmany believe that the intensity of the nightsideionospheric convection begins to increaseafter 30 minutes. This perception is based onthe erroneous premise that the downstream(nightside) ionospheric convection cannot beinduced by the magnetotail unless the changeon the dayside has been relayed to the magne-totail. However, in reality, nightside convec-tion responds immediately to changes in IMF,as does dayside convection (Ridley et al,1997; Murr and Hughes, 2001). This hasbeen interpreted as the propagation of cuspinformation to the nightside through the iono-sphere at a large Alven velocity (See Chapter3 for details).

This explanation is equivalent to statingthat the ionosphere is incompressible. In otherwords, it proves that the ionospheric electricfield can be expressed by potential, that themotion in the ionosphere is restricted to aneddy (vortex) motion (divV = 0), and that themagnetic field is stationary (∂B/∂t = 0). Forionospheric closure of FAC, Eq. (2) repre-sents necessary and sufficient conditions.Therefore, convection can be generated in thenightside ionosphere if there is a supply of J‖.Here, also, the problem lies in the discussionof the actual position and structure of thedynamo. The FAC remains an FAC only inthe low-beta region. In the high-beta region, itbecomes immediately nonparallel to the mag-netic field. This causes the dynamo of theregion 1 current to be situated relatively nearthe dayside (Fig.6). This results in a quickionospheric response.

This naturally raises the question ofwhether ionospheric convection can controlmagnetospheric convection. Such dominationof the ionosphere may lead to contradictingstates in magnetospheric convection. Howev-er, this problem is easily solved, since not onlyshear motion but also compressible motion arepossible in the magnetosphere to accommo-date any contradicting states. A typical exam-

ple of this problem may be found in convec-tion in the substorm growth phase, asexplained below.

6.2 Distant-Tail Neutral Line and theGrowth Phase

A northward-to-southward turning of theIMF enhances dayside reconnection due to itsantiparallel-merging characteristic. Observa-tions have revealed that the progress of mag-netospheric convection accompanies a thin-ning of the plasma sheet, and that in the iono-sphere, two-cell convection is induced accom-panied by a gradual increase in current intensi-ty. These represent well-known features of thegrowth phase (Baker et al ., 1996). Asexplained in the previous section, it is predict-ed that at least 30 minutes is required for themagnetospheric convection to reach the night-side depending on the solar wind velocity.However, it takes the ionosphere only approx-imately 5 minutes to turn into a two-cell con-vection involving the nightside. This is clear-ly the result of selective generation of shearmotion by the incompressible ionosphere.

The flaring angle theory is often proposedto explain plasma sheet thinning (Baker et al.,1996). Under this theory, the open magneticfield lines accumulate on the nightside due tothe transport after reconnection on the day-side, producing a larger flaring angle. There-fore, the lobes are more strongly sailed bysolar wind pressure, which leads to increasedlobe magnetic pressure. This increased lobepressure compresses the plasma sheet andthins it. It is obvious that this theory reliesupon an extension of the magnetosphericimage based on localized MHD balance orsingle-particle description.

In contrast, from a global magnetosphericimage based on convection, plasma sheet thin-ning is generated due to the "gap" between themagnetospheric and ionospheric convectionsduring the growth of convection (Tanaka,2000b). Fig.19 presents the magnetic topolo-gy during the growth phase, 30 minutes afterthe turning of the IMF orientation fromobliquely northward to obliquely southward.

TANAKA Takashi

94

It is evident that the topology still retains fea-tures of northward IMF. The southward IMFis present only near the surface of the magne-tosphere, and the core part still retains thenorthward IMF. The mechanism of thegrowth phase under such a structure is shownin Fig.20. With the development of the day-side reconnection, the magnetospheric convec-tion from the dayside toward the nightside isstrengthened. However, the distant-tail neu-tral line does not immediately disappear, andso the plasma sheet backflow does not imme-diately increase along with the convection.

This fast forward and slow backward convec-tion accompanies a compressional componentand cannot be directly projected onto the iono-sphere. However, the ionosphere does notneed to incorporate the entire state of the mag-netosphere. In this case magnetospheric con-vection need only be seen from the ionosphereas a shear flow. The magnetospheric convec-tion can generate a shear flow against the ion-osphere by creating a divergent flow thatsweeps out the plasma accumulated in theplasma sheet to dayside. This causes the out-going dayward flux from the plasma sheet toexceed the supply flux to the plasma sheetfrom the distant-tail, resulting in plasma sheetthinning. In this way, the thinning of the plas-ma sheet during the growth phase is a naturalconsequence of convection development. Inthe diagrams presented in Fig.2, the divergentflow corresponds to the condition∂B/∂t≠0,and thinning is an expected outcome.

6.3 Substorm OnsetNo matter how large in magnitude the

magnetosphere-ionosphere disturbance maybe, it would not be regarded as a substorm if itwere to start gradually. A major object of sub-storm study is to find an explanation for theappearance of discontinuity at onset. A dipo-larization event that definitely corresponds tothe condition∂B/∂t≠0 characterizes onset,and reveals the convergent property of magne-tospheric convection (Fig.2). Naturally, thisconvergent motion is not projected onto theionosphere. Conversely, the magnetosphereneeds to generate such motion as that confinedto the magnetosphere.

The near-earth neutral line (NENL) modelin the lower panel of Fig.21 (Baker et al.,1996) explains flow convergence (dipolariza-tion) as a pileup of fast flow from the NENL.Therefore, the motion is in a fast wave mode,in which both the magnetic field and fluid aresimilarly compressed. This produces tailwardpressures that are balanced by the earthwarddynamic pressure. This structure is the sameas that for the bow shock in Fig.6, and thenegative J•E generates a dawnward current.

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Magnetic field topology during thegrowth phase 30 minutes afterchange in IMF

Fig.19

Schematic diagram of growth phaseconvection

Fig.20

The red lines show the regions ofenhanced convection.

95

This current is considered to correspond to thecurrent wedge. The substorm discontinuity inthis model results from the rapid developmentof NENL. The reconnection must be an insta-bility in this case, and is regarded as a result ofkinetic processes. However, this model isinconsistent with the results shown in Fig.9.In this model, the onset proceeds from the tailto the inner magnetosphere (steps 1, 2, and 3in Fig.21, bottom). The other weakness of thismodel lies in the discordance in the brighten-ing order of quiet arcs (which proceed fromthe equator to polar direction) that are alreadypresent before the onset. Furthermore, it failsto explain the fact that substorm onset is trig-gered by a northward turning of IMF (Lyons,1997).

Performing a rotation operation on depo-larization corresponds to the development of acurrent disruption (CD), and under the BVparadigm, these two are equivalent processes.

In contrast to the NENL model, the CD modelattributes the substorm discontinuity to the CD(depolarization). In the CD model, distur-bance produced as a result of CD propagatesto the tail and triggers the NENL (Fig.21, top;Lui, 1996). In this model, the onset proceedsfrom the inner magnetosphere to the tail (steps1, 2, and 3 in Fig.21, top). The CD is a kineticprocess and is also unstable. Supporters ofthis model associate the severe magnetic oscil-lations accompanying depolarization withnon-MHD processes, and seem to believe thatthere may even be a separation between themagnetic field and plasma. Therefore, thismodel is interpreted under the EJ paradigm.The CD model is consistent with the brighten-ing order of the quiet arcs.

The two models above both attribute thecause of onset to local instabilities. In con-trast, the state transition model (Tanaka,2000b) regards the onset as extension of thedevelopment process of convection. Duringthe growth phase, the balance between plasmapressure and magnetic pressure in the z-direc-tion is the dominating element in the forcebalance, and depolarization is considered to bea restoration of tension resulting from theshrinkage of the elongated magnetic field.The restored tension balances with the tail-ward －∇P from the high-pressure region cre-ated by injection. In other words, tension con-fines the plasma and enables convergence ofthe convection, allowing convection motion tobe driven even when the magnetosphere andionosphere are partially out of sync. Thisdriven convection motion corresponds to theonset. Plasma accumulates along the magnet-ic field lines without compression of the mag-netic field, and forms slow wave variations.Fig.22 shows the development of pressure dis-tribution within the plasma sheet. The varia-tions in pressure distribution are similar tothose observed by Kistler et al. (1992), andgenerate a maximum within 10 Re after onset.These variations are rapid, and within 1minute, a transition in distribution occurs fromthe growth phase to the expansion phase.

The state-transition model can also explain

TANAKA Takashi

Schematic diagrams of two substormmodels

Fig.21

Top and bottom figures represent CD andNENL models, respectively. Both modelsassume a principal cause for the onset,and hence the resulting effects occur insequence.

96

the triggering effect caused by the northward-turning of IMF. The deceleration of the ionos-pheric convection reduces the flux exiting theplasma sheet to the dayside, creating a regionprone to flow convergence. The triggeringeffect by northward-turning of IMF is alsoexplained by Lyons (1995). However, Lyons'theory attempts to explain the triggering effectas originating from the deceleration of con-vection caused by the penetration of the weak-ened solar wind electric field into the magne-tosphere; this leads to the mistake of the EJparadigm pointed out by Parker (2000).

The NENL model and the CD model bothassume that there must be a central player ini-tiating the onset. In contrast, there is no cen-tral player in the state transition model. Thestate transition model resembles the economicmodel of major depressions, the Ising modelfor magnetization, or the avalanche model forsubstorms (Chapman et al., 1998). Thesemodels are all based on cooperative phenome-na, and no central players exist. Instead, inthese models, many similar elements coexistand interact with one another. The state tran-sition of the interacting systems correspondsto the conditions of a major depression and ofmagnetization. The onset of a substorm is the

occurrence of state transition in the interactingsystem, as shown in Fig.10. A substorm doesnot involve as many elements as the economicmodel or the model for magnetic bodies.However, all elements in substorm have char-acteristic topologies. Since the conditions dif-fer significantly from those of a typical com-plex system (found in economic models andmodels for magnetic bodies), we will refer tothe substorm as a composite system. The con-cept of this type of state transition has alsobeen suggested by Atkinson (1991), followedby Tanaka (2000b).

6.4 Pseudo Breakup and SMCPseudo breakup and steady magnetospher-

ic convention (SMC, or convection bay) arephenomena resembling a substorm but whichare grouped in a different category. The pseu-do breakup displays a common morphologywith substorms until the onset. However, itdoes not accompany poleward expansion or awestward traveling surge (WTS), and termi-nates before changes are propagated to thewhole magnetosphere (Pulkkinen et al., 1998).The magnitude of disturbance in an SMC iscomparable to that of the substorm, but anSMC progresses to the expansion phase with-out displaying a clear discontinuity, or onset(Yahnin et al., 1994; Sergeev et al., 1996). Inan SMC, a strong, stationary two-cell ionos-pheric convection is realized.

A pseudo breakup can be explained interms of the NENL model as a mid-course ter-mination of an initiated reconnection. In con-trast, in the state transition model, it can beinterpreted as follows. Normally, state transi-tion is initiated in the inner magnetospherewith the extinction of the distant-tail neutralline shown in Fig.19, and the substorm pro-ceeds to the expansion phase. However, ifstate transition occurs before the extinction ofthe distant-tail neutral line shown in Fig.19,the discontinuity at onset is realized, but theNENL cannot be formed at this stage due tothe presence of the core structure shown inFig.19. This results in a failure to proceed tothe expansion phase, and instead, the event

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Development of plasma sheet pres-sure distribution during substorms

Fig.22

97

ends in a pseudo breakup. If convection ispromoted without state transition or after thestate transition has been initiated, and the stateof no distant-tail neutral line is maintained ,then the event is an SMC. In this way, thestate transition model provides simple expla-nations for pseudo breakups and SMC.

6.5 SC and ConvectionSudden commencement (SC) is a nonsta-

tionary response of the magnetosphere whensubject to sudden and strong increase indynamic pressure of the solar wind. The mag-netosphere must achieve a new state of equi-librium to incorporate this strong dynamicpressure. This is a classic problem, and hasbeen the subject of extended study; in first-order approximation, this response is inter-preted as an increase in Chapman-Ferraro cur-rent. This has often led to the misapplicationof Bio-Savart's Law in an attempt to solve theproblem. As we have repeatedly stressed inprevious sections, variations in the magnetos-phere-ionosphere system must be regarded asa response in convection. The magnetosphereis not a vacuum, and therefore, SC must beinterpreted not in terms of Bio-Savart's Law,but in terms of convection physics or in termsof composite system. The pattern of geomag-netic field variation during a typical SCinvolves the initial short-period reduction ofhorizontal components, followed by a rapidincrease in horizontal components, after whichthis increased level is maintained for sometime. The initial reduction is referred to as thepreliminary reverse impulse (PRI) and the fol-lowing increase is called the main impulse(MI).

Results of recent SC studies using MHDsimulations have revealed the nature of the SCin terms of composite system. Figs.23 and 24show the simulation results of SC developingout of the conditions shown in Figs.13, 14,and 15. Fig.23 shows the dayside magnetos-phere structure, the magnetic field, and currentlines at the PRI stage. At this stage, a com-pressional motion is propagated within themagnetosphere, and a polarization current is

formed at the propagation front. The non-uni-formity of the medium couples the polariza-tion current with the FAC. The middle panelshows how a PRI current loop is formed fromthe increased CF current, polarization current,and FAC. From the top panel, it can be seenthat the magnetic field extending from theregion of FAC inflow is located far within the

TANAKA Takashi

Dayside magnetic field topology,electric current system, and energyconversion (J•E) during preliminaryreverse impulse (PRI)

Fig.23

The color scale represents the pressuredistribution on the 12-0 meridian andequatorial planes, and the red line repre-sents the magnetic field extending fromthe region where PRI is most observed.The contours represent the intensity ofelectric current. The middle panel is thesame as the top panel, except for the redline representing electric current. In thebottom panel, J•E is superimposed on theelectric current.

98

magnetopause. From J•E in Fig.23, it can beseen that the CF current is the dynamo, andthat work is done by the solar wind dynamicpressure. Since momentum is transferredahead at the polarization current, the com-pressed magnetospheric magnetic field pushesthe magnetospheric plasma, and J•E is posi-tive. Even though the topology is different,the structure shown in Fig.10 is generallyestablished for the PRI.

Fig.24 illustrates MI convection. Thepanel on the lower right shows the develop-ment of an ionospheric convection afterarrival of high dynamic pressure. The secondplot still shows residual of PRI. MI developsin the third panel, and then becomes graduallydamped in the ensuing panels. The upper-leftand central panels show the ionospheric con-vection and magnetic field structure at the tim-ing between plots 3 and 4 in the lower-rightpanel. The small circles in the ionosphericconvection represent the footpoints of themagnetic field lines presented in the central

panel. At high latitudes, a lobe cell is seenaccompanied by northward IMF. In contrast,the ionospheric convection cells at low lati-tudes in the dusk sector accompany region 1currents; this is MI convection. The twogroups of small circles in the ionospheric con-vection show the flow paths of footpoint ofmagnetic field lines constructing the lobe celland the MI cell. The lobe cell structure in thecenter panel corresponds to that in Fig.13, andconsists only of open magnetic field lines. Onthe other hand, the MI convection has onlyclosed magnetic field lines, and clearly differsfrom convections induced by reconnection. Atthe end of PRI, the dayside magnetic field isexcessively compressed compared to thenightside, and is in a state of pressure imbal-ance. The MI convection is generated torestore balance, using this dayside over-com-pression as an energy source. As expected,convection is a process that transfers the mag-netosphere into a new state of equilibrium, asshown in Fig.10.

The magnetic variations during SC arestrong in the initial stage, or PRI, when energyis supplied to the magnetosphere. During MI,when this energy is consumed, the magnetos-pheric variations are relatively mild. Never-theless, PRI produces only slight variations inthe ionosphere, while the MI is observed as alarge disturbance. This is a direct conse-quence of the fundamental dynamics of themagnetosphere-ionosphere system shown inFig.2. In other words, PRI is a compressionmotion and is basically invisible in the iono-sphere. Due to the non-uniformity of themedium, there occurs only a small disturbanceas PRI due to mode conversion in the magne-tosphere. In contrast, the MI is a shear motioncoupled between the magnetosphere and iono-sphere, and can therefore be observed as alarge disturbance.

7 Conclusions

Convection is the fundamental process ofmagnetosphere physics. The main problem inunderstanding this process lies in determining

Journal of the Communications Research Laboratory Vol.49 No.3 2002

Convection of the main pulse (MI)Fig.24The upper-left figure shows the ionos-pheric potential and the footpoints of themagnetic field, the panel in the centershows the magnetic field topology, andthe lower-right panel represents the devel-opment of the ionospheric potential.

99

how the magnetosphere and the ionospheremaintain self-consistency while responding togiven solar wind conditions. The physicsunderlying any space weather phenomena canbe identified by examining how the convec-tion structure shown in Fig.10 can be realizedin a self-consistent manner, whether the phe-nomena consist of substorms, SMC (steadymagnetospheric convection), pseudo breakup,SC (sudden commencement), or geomagnetic

storms.The physics of magnetosphere-ionosphere

coupling has been described in terms of thephysics of complex systems, taking into con-sideration the importance of self-consistencyin the magnetosphere-ionosphere coupling, theimportance in the earth and planetary sciencesof treating systems with topology, and the pos-siblity of state transitions as cooperative phe-nomena within entire systems.

TANAKA Takashi

References1 Akasofu, S. -I., "The development of the auroral substorm", Planet. Space Sci., 12, 273, 1964.

2 Amm, O., P. Janhunen, H. J. Opgennoorth, T. I. Pulkkinen, and A. Viljanen, "Ionospheric shear flow situation

observed by the MIRACLE network, and the concept od Harang discontinuity, in Magnetospheric current

systems ", Geophys. Monogr. Ser., edited by S. Ohtani et al., p.227, AGU, Washington, D. C., 2000.

3 Atkinson, G., "A magnetosphere WAGS the tail model of substorms, in Magnetospheric substorms", Geo-

phys. Monogr. Ser., Vol. 64, edited by J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima, p. 191, AGU,

Washington D.C., 1991.

4 Atkinson, G., and D. Hutchison, "Effect of the day night ionospheric conductivity gradient on polar cap con-

vection flow ", J. Geophys. Res., 83, 725, 1978.

5 Axford, W. I., and C. O. Hines, "A unifying theory of high latitude phenomena and the geomagnetic storm ",

Can. J. Phys., 39, 1433, 1961.

6 Baker, D. N., T. I. Pulkkinen, V. Angelopoulos, W. Baumjohann, and R. L. McPherron, "Neutral line model of

substorms: Past results and present view ", J. Geophys. Res., 101, 12,975, 1996.

7 Chapman, S. C., N. W. Watkins, R. O. Dendy, P. Helander, and G. Rowlands, "A simple avalanshe model as

an analogue for magnetospheric activity", Geophys. Res. Lett., 25, 2397, 1998.

8 Cowley, S. W. H., "A qualitative study of the reconnection between the Earth's magnetic field and an inter-

planetary field of arbitrary orientation ", Radio Sci., 8, 903, 1973.

9 Cowley, S. W. H., "Magnetosphere-ionosphere interactions: A tutorial review, in Magnetospheric current

systems ", Geophys. Monogr. Ser., edited by S. Ohtani et al., p.91, AGU, Washington, D. C., 2000.

10 Crooker, N. U., "Morphology of magnetic merging at the magnetopause ", J. Atmos. Terr. Phys., 52, 1123,

1990.

11 Crooker, N. U., J. G. Lyon, and J. A. Fedder, "MHD model merging with IMF By: Lobe cells, sunward polar

cap convection, and overdraped lobes ", J. Geophys. Res., 103, 9143, 1998.

12 Cumnock, J. A., J. R. Sharber, R. A. Heelis, M. R. Hairston, and J. D. Craven, "Evolution of the global aurora

during positive IMF Bz and varying IMF By conditions ", J. Geophys. Res., 102, 17,489, 1997.

13 Dungey, J. W., "Interplanetary magnetic field and the auroral zones ", Phys. Rev. Lett., 6, 47, 1961.

14 Erickson, G. M., R. W. Spiro, and R. A. Wolf, "The physics of the Harang discontinuity ", J. Geophys. Res.,

96, 1633, 1991.

15 Gombosi, T. I., K. G. Powell, and B. van Leer, "Comment on Modeling the magnetosphere for northward

interplanetary magnetic field: Effects of electrical resistivity by Joachim Raeder ", J. Geophys. Res., 105,

13,141, 2000.

100 Journal of the Communications Research Laboratory Vol.49 No.3 2002

16 Harang, L., "Polar geomagnetic storms", J. Geophys. Res., 51, 353,1946.

17 Harel, M., R. A. Wolf, P. H. Reiff, R. W. Spiro, W. J. Burke, F. J. Rich, and M. Smiddy, "Quantitative simula-

tion of a magnetospheric substorm, 1, Model logic and overview ", J. Geophys. Res., 86, 2217, 1981.

18 Heppner, J. P., and N. C. Maynard, "Empirical high-latitude electric field models ", J. Geophys. Res., 92,

4467, 1987.

19 Iijima, T., "Field-aligned currents in geospace: Substance and significance, in Magnetospheric current sys-

tems ", Geophys. Monogr. Ser., edited by S. Ohtani et al., p.107, AGU, Washington, D. C., 2000.

20 Iijima, T., and T. A. Potemra, "The amplitude distribution of field-aligned currents at northern high latitudes

observed by Triad ", J. Geophys. Res., 81, 2165, 1976.

21 Kamide, Y., W. Sun, and S. -I. Akasofu, "The average ionospheric electrodynamics for the different sub-

storm phases ", J. Geophys. Res., 101, 99, 1996.

22 Khan, H., and S. W. H. Cowley, "Observations of the response time of the high-latitude ionospheric convec-

tion to variations in the interplanetary magnetic field using EISCAT and IMP-8 data ", Annales Geophysicae,

17, 1306, 1999.

23 Kistler, L. M., E. Mobius, W. Baumjohann, and G. Paschmann, "Pressure changes in the plasma sheet dur-

ing substorm injection ", J. Geophys. Res., 97, 2973, 1992.

24 Lockwood, M., S. W. H. Cowley, P. E. Sandholt, and U. P. Lovhaug, "Causes of plasma flow bursts and

dayside auroral transients: An evaluation of two models invoking reconnection pulses and changes in the Y

component of the magnetosheath field ", J. Geophys. Res., 100, 7613, 1995.

25 Lui, A. T. Y., "Current disruption in the earth's magnetosphere: Observations and models ", J. Geophys.

Res., 101, 13,067, 1996.

26 Lui, A. T. Y., "Electric current approach to magnetospheric physics and the distinction between current dis-

ruption and magnetic reconnection, in Magnetospheric current systems ", Geophys. Monogr. Ser., edited by

S. Ohtani et al., p.31, AGU, Washington, D. C., 2000.

27 Lyons, L. R., "A new theory for magnetospheric substorms", J. Geophys. Res., 100, 19,069, 1995.

28 Lyons, L. R., G. T. Blanchard, J. C. Samson, R. P. Lepping, T. Yamamoto, and T. Moretto, "Coordinated

observation demonstrating external substorm triggering ", J. Geophys. Res., 102, 27,039, 1997.

29 Murr, D. L., and W. J. Hughes, "Reconfiguration timescales of ionospheric convection ", Geophys. Res.

Lett., 28, 2145, 2001.

30 Newell, P. T., and D. G. Sibeck, "By fluctuation in the magnetosheath and azimuthal flow velocity transients

in the dayside ionosphere ", Geophys. Res. Lett., 20, 1719, 1993.

31 Ogino, T., R. J. Walker, M. Ashour-Abdalla, and J. M. Dawson, "An MHD simulation of the effects of inter-

planetary magnetic field By component on the interaction of the solar wind with the earth's magnetosphere

during southward interplanetary magnetic field ", J. Geophys. Res., 91, 10,029, 1986.

32 Ohtani, S., and T. Higuchi, "Four-sheet structure of dayside field-aligned currents: Statistical study", J. Geo-

phys. Res., 105, 25,317, 2000.

33 Parker, E. N., Newton, Maxwell, and Magnetospheric Physics, in Magnetospheric current systems, Geo-

phys. Monogr. Ser., edited by S. Ohtani et al., p.1, AGU, Washington, D. C., 1999.

34 Pulkkinen, T. I., D. N. Baker, M. Wiltberger, C. Goodrich, R. E. Lopez, and J. G. Lyon, "Pseudobreakup and

substorm onset: Observations and MHD simulations compared ", J. Geophys. Res., 103, 14,847, 1998.

35 Ridley, A. J., G. Lu, C. R. Clauer, and V. O. Papitashvili, "Ionospheric convection during nonsteady interplan-

etary magnetic field conditions ", J. Geophys. Res., 102, 14563, 1997.

36 Ruohoniemi, J. M., and R. A. Greenwald, "Statistical patterns of high-latitude convection obtained from

Goose Bay HF radar observations ", J. Geophys. Res., 101, 21,743, 1996.

101TANAKA Takashi

37 Sergeev, V. A., R. J. Pellinen, and T. I. Pulkkinen, "Steady magnetospheric convection: A review of recent

results ", Space Sci. Rev., 75, 551, 1996.

38 Siscoe, G. L., N. U. Crooker, G. M. Erickson, B. U. O. Sonnerup, K. D. Siebert, D. R. Weimer, W. W. White,

and N. C. Maynard, "Global geometry of magnetospheric currents inferred from MHD simulations, in Magne-

tospheric current systems ", Geophys. Monogr. Ser., edited by S. Ohtani et al., p.41, AGU, Washington, D.

C., 2000.

39 Song, P., and C. T. Russell, "Model of the formation of the low-latitude boundary layer for strongly north-

ward interplanetary magnetic field ", J. Geophys. Res., 97, 1411, 1992.

40 Sonnerup, B. U. O., "Theory of low-latitude boundary layer ", J. Geophys. Res., 85, 2017, 1980.

41 Stern, D. P., "The origin of Birkeland current ", Rev. Geophys., 21, 125, 1983.

42 Tanaka, T., "Generation mechanisms for magnetosphere-ionosphere current systems deduced from a

three-dimensional MHD simulation of the solar wind-magnetosphere-ionosphere coupling processes ", J.

Geophys. Res., 100, 12,057, 1995.

43 Tanaka, T., "Configuration of the magnetosphere-ionosphere convection system under northward IMF con-

dition with non-zero IMF By ", J. Geophys. Res., 104, 14,683, 1999.

44 Tanaka, T., "Field-aligned current systems in the numerically simulated magnetosphere, in Magnetospheric

current systems ", Geophys. Monogr. Ser., edited by S. Ohtani et al., p.53, AGU, Washington, D. C., 2000a.

45 Tanaka, T., "The state transition model of the substorm onset ", J. Geophys. Res., 105, 21,081, 2000b.

46 Tanaka, T., "IMF By and auroral conductance effects on high-latitude ionospheric convection ", J. Geophys.

Res., 106, 24,505, 2001.

47 Troshichev, O. A., "Global dynamics of the magnetosphere for northward IMF conditions ", J. Atmos. Terr.

Phys., 52, 1135, 1990.

48 Weimer, D. R., "Models of high-latitude electric potentials derived with a least error fit of spherical harmonic

coefficients ", J. Geophys. Res., 100, 19,595, 1995.

49 Yahnin, A., M. V. Malkov, V. A. Sergeev, R. J. Pellinen, O. Aulamo, S. Vennerstrom, E. Friis-Christensen, K.

Lassen, C. Danielsen, J. D. Craven, C. Deehr, and L. A. Frank, "Features of steady magnetospheric convec-

tion ", J. Geophys. Res., 99, 4039, 1994.

TANAKA Takashi, Dr. Sci.

Professar of Graduate School of Sci-ence, Kyusyu University

Magnetosphere-Ionosphere Physics