Interplanetary magnetic field By and auroral conductance effects on high-latitude ionospheric convection patterns

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106, NO. All, PAGES 24,505-24,516, NOVEMBER 1, 2001 Interplanetary magnetic field By and auroral conductance effects on high-latitude ionospheric convection patterns T. Tanaka Communications Research Laboratory, Tokyo, Japan Abstract. The dependence of the ionospheric electric potential (convection) on the interplanetary magnetic field (IMF) and the ionospheri conductivity is investigated to understand the generation of convection patterns in the framework of the solar windmagnetosphere-ionosphere (S-M-I) coupling scheme and the merging concept. A numerical magnetohydrodynamic (MHD) simulation is adopted for the study of the present problem. To achieve a high resolution in the ionosphere, the MHD calculation employs the finite volume (FV) total-variation diminishing (TVD) scheme with an unstructured grid system. The two-cell convection patterns reproduced from simulation are shown for several cases under the southward IMF condition during the growth-phase interval. In the investigation of these results, special attention is paid to the analysis of mirror symmetry in the convection patterns with respect to the IMF By. On the dayside in the Northern Hemisphere, IMF By_ (By+) generates flow deflection on newly opened field lines toward the dusk (dawn) without a severe violation of the mirror symmetry. While the mirror symmetry of the convection pattern is maintained even on the nightside when the ionospheric conductivity is uniform, it is not maintained on the nightside when the ionospheric conductivity is nonuniform. A realistic ionospheric conductivity modifies the convection pattern in the Northern (Southern) Hemisphere so as to emphasize distinctive featureseen for IMF By+ (By_) under a uniform conductivity, and the reproduced convection patterns coincide with the observation quite well including fine signatures on the nightside, both for IMF By_ and By+. Because of the nonuniform conductivity, cell centers of convection are shifted to the earlier magnetic local times, and the antisunward flow in the northern polar cap is nearly aligned with noon-midnight meridian for IMF By_, while the flow in the northern polar cap has a significant inclination from prenoon to premidnight for IMF B +. These convection patterns can be understood by considering the effect due to the Hal current closure of the region-1 field- aligned current. The analysis for the dependence of nightside convection IMF By and ionospheric conductivity shows that the Harang discontinuity is attributed partially to the structure of magnetospheric driver but mainly to the effect of nonuniform auroral conductivity. As a consequence, it is more adequate to say that convection patterns are more or less caused by the synthesized effect of more than one process rather than a single elementary process. Reproduced convection patterns in this paper show a particular coincidence with satellite observationsummarized by adopting the pattern-recognitionbased approach. 1. Introduction The high-latitude ionospheric convection extending from the polar cap to the auroral zone is a low-altitude manifestation of the magnetospheri convection. The ionospheric convection, however, is not a mere projection of the magnetospheric driver. The magnetospheric and ionospheric convections construct a mutual coupling system through the projection of convection electric field and the exchange of field-aligned current (FAC) [Vasyliunas, 1970; Sonnerup, 1980; Stem, 1983; Caudal and Blanc, 1988]. Since the projection of electric field is equivalento the projection of motion, they are achieved through the transmission of stress by the FAC [Iijima and Potemra, 1976; Iijima, 2000]. In the magnetosphereionosphere (M-I) coupling system, therefore, the electric field Copyright 2001 by the American Geophysical Union. Paper number 2001JA /01/2001JA $09.00 aspect and the FAC aspect of the convection must be the different points of view for the same physical phenomenon occurring there [Tanaka, 2000a]. When the interplanetary magnetic field (IMF) is southward, tailward transport of open field lines following the bow reconnection is mapped down as antisunward flow at the cusp and in the polar cap, and sunward return of closed field lines following the reclosure at downstream is mapped down as sunward flow in the auroral oval. As a whole, these structures are recognized as the two-cell convection pattern in the polar ionosphere. In this pattern the polar cap potential drop is the mapping of the potential distributing along the reconnection line on the dayside magnetopause. This potential distributing along the dayside reconnection line is, in turn, the mapping of the solar wind potential between the outmost reconnecting IMFs [Crooker, 1988]. These structures construct the macroscopic convection-driving system in the solar wind-magnetosphere-ionosphere (S-M-I) system. 24,505

2 24,506 TANAKA: IMF By AND AURORAL CONDUCTANCE EFFECTS Recently, detailed morphologies of high-latitude convection have been derived from the statistical analyses of data obtained by polar-orbiting satellites [Heppner and Maynard, 1987; Weimer, 1995], ground-based incoherent scatter (IS) and highfrequency (HF) radars [de la Beaujardieret al., 1986; Foster et al., 1986; Holt et al., 1987; Leonard eta!., 1995; Ruohoniemi and GreenwaM, 1996], and magnetometer-observation-based techniques [Richmond and Kamide, 1988; Lu et al., 1994; Lyons et IMF By effects are seen even on the nightside [Heppner and Maynard, 1987; Weimer, 1995; Ruohoniemi and GreenwaM, 1996]. In the Northern Hemisphere the antisunward flow in the nightside polar cap is nearly aligned with the noon-midnight meridian for IMF By_, while the flow in the nightside polar cap has a significant inclination from prenoon to premidnight for IMF By+. From the radar observations, the convection reversal, a distinct feature near 70 ø latitude on the nightside, al., 1996]. These analyses have revealed that the IMF By mod- can be discerned only in the sector of the crescent-shaped cell, ulates the basic two-cell morphology of the ionospheric con- that is, premidnight for IMF By_ and postmidnight for IMF vection under the southward IMF condition. Through the By+. According to radar observations, consequently, breakbreakdown of dawn-dusk symmetry, the IMF By effects reveal down of mirror symmetry is not so severe even in the nightside. more clearly the electrodynamic structures in the convection In this case the dusk (dawn) cell makes the larger contribution system, which are outwardly hidden in a due southward IMF to the convection on the midnight meridian for IMF By_ case under the symmetry of the system. (By+). This feature, however, must be considered in relation In the Northern Hemisphere, for instance, the well-known with the long-lasting problem about the Harang discontinuity dayside IMF By_ (By + ) effect is to give duskward (dawnward) [Harang, 1946; Erickson et al., 1991], because its position overflow deflection on newly opened field lines. Southern Hemi- laps that of the Harang discontinuity. sphere dayside flows become mirror symmetric to those in the In general, the Harang discontinuity is defined as the locus Northern Hemisphere for the same sign of IMF By. Discus- of points at which the zonal velocity reverses from westward to sions in this paper are therefore mainly concentrated on the eastward with increasing latitude and is generally understood Northern Hemisphere. Thus initial flow deflection on its tranto slant equatorward with increasing MLT in the evening secsient from the cusp to the polar cap reveals operation of magtor [Erickson et al., 1991]. From the radar observations, as netic tensionormal to the reconnection line at By-influenced mentioned above, the Harang discontinuity is clear in the reconnection on the dayside [Crooker, 1979]. Through the pro- Northern (Southern) Hemisphere only for IMF By_ (By+). jection of merging line, the north-south asymmetry also ap- For the other sign of the!mf By, latitudinal velocity shear is pears in the position of merging gap in the cusp ionosphere, concentrated in the dawn sector. On the contrary, Heppner and reflecting the fact that the antiparallel merging line on the Maynard [1987] observed the Harang discontinuity in the magnetopause cannot be accumulated on the component evening for both IMF By + and By_ cases. In their observations merging line under the nonzero IMF By condition [Crooker, the evening Harang discontinuity for IMF By+ is at lower 1988; Moses et al., 1988]. latitude than it is for By_. Breaking of mirror symmetry is The IMF By effects on the convection also appear in the unavoidable in the results of Heppner and Maynard [1987], global shaping of cells and the orientation of the basic two-cell since the Harang discontinuity being stationed in the evening position in magnetic local time (MLT) [Heppner and Maynard, gives a symmetric structure rather than mirror symmetry. In 1987; Weimer, 1995; Ruohoniemi and GreenwaM, 1996]. For this problem, numerical treatment would be inevitable to clar- IMF By_ (By+), the northern dusk (dawn) cell is crescentify the relative role acted by the magnetospheric driver and shaped and the northern dawn (dusk) cell is round due to the initial flow deflection. In addition to this crescent/round shapauroral conductivity for the Harang discontinuity. ing, observed convection patterns include a component which It is the purpose of this paper to investigate the nonuniformconductivity effect in a rigorous manner to clarify its role in breaks mirror symmetry with respect to IMF By. For IMF By_ the flow in the northern polar cap is antisunward and crescent/ breaking mirror symmetry of ionospheric convection patterns round cells are of similar size, while for IMF By + the northern with respect to IMF By and in generating fine signatures on the dawn (crescent) cell is smaller than dusk (round) cell and is nightside. We investigate how the combined effect of IMF By more crescent-shaped, and the orientation of the overall pat- and nonuniform conductivity generates modulations on the tern is such that the cell centers are shifted to earlier MLTs. convection patterns and how it breaks the mirror symmetry of Atkinson and Hutchison [1978] proposed a mechanism which the system with respecto IMF By. In this paper, numerical breaks the mirror symmetry with respecto IMF By. Within magnetohydrodynamic (MHD) simulation of the S-M-I couthe polar cap the day-night conductivity gradient modifies cur- pling system is adopted to analyze this problem. The strategy of vature of the equipotentials so that they concentrate on the this paper is as follows: After a brief explanation of adopted dawnside [Atkinson and Hutchison, 1978]. This makes the dusk MHD simulation in section 2, we will show in section 3 simucell look round and the dawn cell look crescent-shaped, in the lation results obtained under several different conditions and sense to exaggerate northern convection pattern under IMF discuss generation mechanisms of various convection patterns, By +. For IMF By +, therefore, the round/crescent cell pattern not only global configuration but also fine structures, by comin the Northern Hemisphere can be enhanced while for IMF paring the IMF By effects for different conditions. The last By_ the flow lines across the polar cap can have a more uniform distribution. This process violating the mirror symmetry can be a cause of complex cell orientation, as pointed out by Ruohoniemi and Greenwa M [1996]. It may explain some of the section includes the conclusions obtained from the present investigation. Substorm processes may perturb the convection in the midnight sector in a manner independent of IMF-driven dayside discrepancies between the observations and the empirical merging. Avoiding this complexity, we treat in this paper the models. However, in order to rigorously incorporate nonuniform-conductivity effect into the conceptual models including auroral conductivity, numerical treatment considering a selfconvection pattern during the growth-phase interval that continues about an hour after a southward turning of the IMF. We look upon this convection during the growth phase as a typical consistent M-I convection is unavoidable. convection pattern under the southward IMF condition.

3 TANAKA: IMF B. AND AURORAL CONDUCTANCE EFFECTS 24,507 Table 1. IMF and Ionospheric Conductivity Parameters Bz at Bz at By at By at t < 0 t > 0 t < 0 t > 0 Conductivity Case _ uniform Case _ uniform Case _ +_ realistic 2. Methods of Numerical Simulation A numerical MHD simulation is adopted for the study of the present problem [Tanaka, 1994, 1995, 1999, 2000a, 2000b]. The plasma convection and PAC play a central role in the M-I coupling, whereas the state of the energy source for this convection system depends on the S-M interaction. Thus a selfconsistent treatment of the S-M-I coupling process is required for the investigation of the convection system. The MHD calculation employs the finite volume (PV) total-variation diminishing (TVD) scheme with an unstructured grid system [Tanaka, 1994], to achieve a high resolution and an excellent shock capturing. In this paper the x axis is pointing toward the Sun, the y axis is pointing in the direction opposite the Earth's orbital motion, and the z axis is pointing north. The outer and inner boundaries for the simulation are at 200 and 3 R r. In the calculation of the M-I coupling process, dependent variables are projected along the field line from the inner boundary to the ionosphere. Under these circumstances the number of effective grid points in the ionosphere becomes equal to the number of grid points on the inner boundary. Consequently, a sufficient number of grid points must be allocated on the inner boundary to resolve the structures generated in the ionosphere. Such a requirement is achieved through the use of an unstructured grid sys- tem. In the ionosphere, Ohm's law is solved to match the divergence of the Pedersen and Hall currents with the PAC. In two cases of simulation (cases 1 and 2 in Table 1), the ionospheric conductivity is set to a constant value (5 mhos). In the other case (case 3 in Table 1) the ionosphericonductivity is set to be more realistic and is calculated from the solar EUV flux, diffuse precipitation modeled by the pressure and temperature, and discreet precipitation modeled by the upward PAC. In this paper the Hall conductivity is set to be 2 times as large as the Pedersen conductivity. Details for the calculation of conductivity follow Tanaka [2000b]. To obtain the ionospheric potential, a two-dimensional partial differential equation is solved on a sphere. Since it is an elliptic equation, it can be solved quite easily by the biconjugate residual method. The outer boundary conditions give a solar wind flow on the upstream side at x = 40 R r and give zero gradient conditions on the downstream side at x = -200 R r. A uniform solar wind with a speed of 350 km s - and a density of 10 cm -3 is assumed at the upstream boundary. The results shown in this paper as convection patterns for the southward IMF correspond to the solutions during the growth-phase interval. In general, a stationary-state solution cannot be obtained in the solar S-M-I coupling system under a southward IMF condition [Tanaka, 2000b]. In order to clarify the situation with which the discussions are concerned, there- fore, simulation is started from a stationary solution under a northward IMP condition [Tanaka, 1999, 2000b] in which the IMF magnitude is 5 nt and the IMP direction is inclined 30 ø from due northwardirection (Bx = 0.0 nt, By = nt, and Bz = 4.33 nt). In the successive calculation the growth phase is simulated by changing the IMF direction at the upstream boundary by 120 ø or 180 ø to southward (Bx = 0.0 nt, By = +_2.50 nt, and Bz nt). The time axis is selected so that the contact of the southward IMF on the dayside magnetopause is at t - 0. After a southward turning of the IMF, the solution changes its configuration gradually until a sudden breakdown of monotonous tail structure [Tanaka, 2000b]. The growth phase is the interval before the sudden breakdown. Characteristic features in the magnetosphere occurring during this interval are erosion of the dayside magnetosphere, thinning of the plasma sheet, and an increase in the flaring angle of the magnetopause [Tanaka, 2000b]. In summary, six simulation results are obtained for three different cases as shown in Table 1. In case 1 the solutions are obtained for IMF By_ and By+ under a constant ionospheric conductivity of 5 mhos, while supposing that the sign of IMF By is unchanged when IMF B changes from northward to southward. In case 2 the solutions are obtained for IMF By_ and By + under a constant ionospheric conductivity of 5 mhos, while supposing that the sign of IMF By changes when IMF B changes from northward to southward. Finally, in case 3, the solutions are obtained for IMF By_ and By+ under a realistic ionosphericonductivity, while supposing that the sign of IMF By does not change when IMF B z changes from northward to southward. 3. Results and Discussion This section shows the simulation results obtained under respective conditions defined in the previous section and compares them with each other. Since overall signatures in the Southern Hemisphere for an IMF By_ (By+) are the same as for those for an IMP By + (By_) in the Northern Hemisphere, all discussion in the remainder of this paper is concentrated only on the Northern Hemisphere. In the M-I system generating the ionospheric convection shown in this section, plasma and magnetic field motions emerging from the magnetosphere the ionosphere must be organized in a circulation configuration without a considerable accumulation at all heights. In this process the magnetospheric convection following the bow reconnection is mapped down to the polar cap ionosphere as a dawn-to-dusk electric field. If the ionospheri current is not connected to the FAC, div J = 0 is violated on the polar cap boundary owing to the dawn-to-dusk ionospheric Pedersen current associated with the antisunward flow [Caudal and Blanc, 1988]. The FAC automatically satisfies div J = 0 to enable a steady ionospheri convection following the magnetosphericonvection. Seeing from a different angle, the J x B force of the ionospheric Pedersen current is such as to maintain the ionospheric plasma convection caused by the mapped-down electric field against the atmospheric friction [Stem, 1983; Tanaka, 2000a]. In opposition, the dissipation effect in the ionosphere decelerates and modifies the convection velocity to generate so-called ionospheric secondary electric field, which is mapped back to the magnetosphere. In the results shown in this section, these processes are taken into consideration in a self-consistent manner from the M-I cou- pling simulation. Plate 1 shows calculated distributions of the ionospheric potential (contours) and FACs (colors) in case 1 for (left) IMP By_ and (right) IMP By+. Plate 1 is seen from above the

4 24,508 TANAKA: IMF By AND AURORAL CONDUCTANCE EFFECTS IMF By- 11 ' otential and field-aligned current Case 1 IMF By I 16 X, 19 la ' -O Plate 1. Calculated distributions of the ionospheric potential (contours) and field-aligned currents (FACs) (colors) in the Northern Hemisphere under the constant conductivity condition (case 1). The interplanetary magnetic field (IMF) angle changes 120 ø at the southward turning of the IMF. Noon is at top, and three circles show the 60 ø, 70 ø, and 80 ø north latitudes. The difference between contours is 6 kv, and the unit for color level is/za m-2. IMF By- 21 Potential and field-aligned current Case 2 MF By Plate 2. Calculated distributions of the ionospheric potential (contours) and FACs (colors) in the Northern Hemisphere under the constant-conductivity condition (case 2). The IMF angle changes 180 ø at the southward turning of the IMF. Noon is at top, and three circles show the 60 ø, 70 ø, and 80 ø north latitudes. The difference between contours is 6 kv, and the unit for color level is/za m -2.

5 TANAKA: IMF B v AND AURORAL CONDUCTANCE EFFECTS 24,509 IM,y Potential and field-aligned curren Case 3 IMF By+ // 3b :37 Plate 3. Calculated distributions of the ionospheric potential (contours) and FACs (colors) in the Northern Hemisphere under the realistic conductivity distribution (case 3). Noon is at top, and three circles show the 60 ø, 70 ø, and 80 ø north latitudes. The difference between contours is 6 kv, and the unit for color level is/ A --2 m. IM By- Potential and conductivity Case 3 MF By Plate 4. Calculated distributions of the ionospheric potential (contours) and conductivity (colors) in the Northern Hemisphere under the realistic conductivity distribution (case 3). Noon is at top, and three circles show the 60 ø, 70 ø, and 80 ø north latitudes. The difference between contours is 6 kv, and the unit for color level is mhos.

6 ..,.. 24,510 TANAKA: IMF Br AND AURORAL CONDUCTANCE EFFECTS IMF By- Conductivity and field-aligned cun'ent Case 3 IMF By Plate 5. Calculatedistributions of the ionospheric conductivity (contours) and FACs (colors) in the Northern Hemisphere under the realisticonductivity distribution (case 3). Noon is at top, and three circles show the 60 ø, 70 ø, and 80 ø north latitudes. The difference between contours is 1.6 mhos, and the unit for color level is/ A m -2. Hall Current divergence and FAC case 3 By- Pedersen Plate 6. Calculatedistributions of the (left) Hall and (right) Pedersen divergences (colors) and their summation (contours) in the Northern Hemisphere for IMF By_ under the realistic conductivity distribution (case 3). In this plate, the size of the polaregion is enlarged compared with the other plates. Noon is at top, and three circleshow the 60 ø, 70 ø, and 80 ø north latitudes. The difference between contours is 0.1/ A m -2, and the unit for color level is/ A m -2.

7 TANAKA: IMF B, AND AURORAL CONDUCTANCE EFFECTS 24,511 North Pole, and noon is at the top. The three circles show the north latitudes of 60 ø, 70 ø, and 80 ø. The basic convection pattern seen in Plate 1 is two-cell for both IMF By_ and By +. By the color codes, the FAC flowing out from (into) the ionosphere is shown as positive (negative) current. The result reproduces the region-1 and region-2 FACs observed by Iijima the IMF and the geomagnetic field for both the southward IMF condition and the northward IMF condition [Nishida et a!., 1998; Tanaka, 1999]. This merging hypothesis has been the base of conceptual models for the IMF-dependent highlatitude ionosphericonvection [Crooker, 1979, 1988]. According to the above concept, there are good understandand Potemra [1976] and Iijima et al. [1984]. The magneto- ings from phenomenological models for processes that lead to spheric processes responsible for the generation of these FACs must include the energy conservation process as discussed by Harel et al. [1981], Stem [1983], Tanaka [1995], and Tanaka [2000a]. dayside convection patterns. The model of Crooker [1979] predicts the morphology of the convection on the dayside as a function of!mf angle in the y-z plane. In this model, dayside merging occurs along the line where the IMF and the geomag- Plate 2 shows calculated distributions of the ionospheric netic field are antiparallel [Ogino et al., 1986], and the associpotential (contours) and FACs (colors) in case 2 for (left) IMF By_ and (right) IMF By +. The format of Plate 2 is the same as that in Plate 1. As seen in Table 1, the difference between ated plasma acquires antisunward motion normal to this line. Concerning these discussions, we cannot conclude which mechanism, antiparallel merging or component merging, is Plates 1 and 2 is the direction of IMF By before the southward actually operating from a mere comparison of phenomenologturning of the IMF (t < 0). In Plate 2, IMF By as well as IMF B z changes its sign associated with the IMF southward turning (t = 0). ical models. However, Ogino et al. [1986] showed from the simulation study that antiparallel merging is more plausible. Under the southward IMF condition, the dominant dayside Plate 3 shows calculated distributions of ionospheric poten- IMF By_ (By+) effect is to give duskward (dawnward) flow on tial (contours) and FACs (colors) for (left) IMF By_ and newly opened field lines. From this process, the model predicts (right) IMF By+ under a realisti conductivity distribution two merging cells for IMF Bz_, one crescent-shaped which lies (case 3). The format of Plate 3 is the same as that of Plate 1. In this plate the mutual-coupling effects between the magnetosphere and ionosphere in the M-I coupling system are taken into account more correctly by considering the modulation of on the dusk (13)(dawn, 16) side for IMF By_ (By+) and the other round which lies on the dawn (15) (dusk, 14) side for IMF By_ (By+). With this cell arrangement, flow deflected zonally on the dayside due to By-influenced reconnection rethe ionosphericonductivity due to the particle precipitation. tains the imprint of its initial deflection on its transient through In the situation considered in this paper the main source of auroral ionization is diffusive particle precipitation due to thermal energy. In the model, therefore, oval position is determined as a projection of the plasma sheet. An important consequence of the simulation is that the relative position of the FAC and the auroral oval is determined retaining a selfconsistency required in the M-I coupling system. the polar cap. These predictions were previously confirmed from the simulation by Fedder et al. [1998], and they are confirmed again in the present simulation. The distribution of FAC shows that the magnetospheric driving term is mainly responsible for the IMF By effects (flow deflection) in the dayside. These IMF By effects on the convection pattern can be evidence for the ionospheri convection Arrows and numbers in Plates 1-4 identify corresponding driven through the merging process in the S-M-I system. These parts of discussions in the text. In the remainder of this paper, each structure is referred to by these numbers. features are reproduced in the present simulation (Plates 1-3) under a proper self-consistency in the S-M-I coupling process, and the results obtained here coincidence well with observa Dayside Convection tions. In some results, however, there remains an inexplicable In Plates 1 and 2 the two-cell convection pattern and the feature in that the region-1 FAC shows a pair of distinct FAC distribution are shown under a constant-ionospheric- patches around the cusp region. This result may indicate that conductivity condition. From Plates 1 and 2 we can see the the global feature of antiparallel merging between the oblique mutual relation between the convection and FAC. It is noted IMF and dayside magnetosphere can be more complicated that the convection reversal coincide strictly with the region-1 FAC under a constant-ionospheric-conductivity condition. In the dayside region from the cusp to the polar cap, flow deflecthan the present understanding. It may not guarantee a stationary-state solution. Reconnection line mapping to the ionosphere through the tion toward dusk (11 and 21) (dawn, 12 and 22) is observable cusp having a finite width becomes very complicated [Crooker, for IMF By_ (By +). Flow deflection in the convection pattern 1988]. The reconnection line standing on the dayside magnearound the cusp is apparently correlated with a well-developed topause is projected to the equatorward boundary of the dayregion-1 FAC in the postnoon (prenoon) region. Even in Plate 3, the overall signature of flow deflection in the dayside is almost similar to those in Plates 1 and 2. Dayside flow in Plate 3 for IMF By_ is especially more similar to those in Plates 1 and 2 than for IMF By+. Comparing dayside convection in Plate 3 with those in Plates 1 and 2, only little difference appears for IMF By+ that a wide zone of westward flow (31) appears in ø latitude. The reason for this small difference will be shown in the next section. side cusp as the open/closed boundary. It is believed that flow in the ionosphere enters from the auroral oval to the polar cap through this footprint of the reconnection line. If the exchange of plasma and magnetic flux through the boundary between the auroral zone and polar cap takes place through a narrow opening or gap around the noon, this merging gap becomes throatlike and the rema. ining boundary is a shear reversal. Alternatively, the auroral zone-polar cap boundary is a rotational reversal, if the exchange of plasma and magnetic flux occurs in The convection of high-latitude ionospheric plasma must be understood in the framework of the solar wind interaction with the M-I system. In recent years, it has become more and more apparent that the principal driving mechanism of the conve c- tion in the M-I system is the magnetic reconnection between a broad local time region [Moses et al., 1988]. The present simulation results coincide with the model that the auroral zone-polar cap boundary in the dayside is a rotational reversal rather than a shear reversal, over a wide local time range. Since IMF By causes the boundary around noon to be aligned along

8 24,512 TANAKA: IMF By AND AURORAL CONDUCTANCE EFFECTS the meridian rather than a latitudinal circle, it produces an east-west flow componenthrough the boundary, duskward for IMF By_ or dawnward for IMF By + [Moses et al., 1988; Lockwood et al., 1995]. These configurations of the dayside convection do not strongly depend on the conductivity model Cell Position The rotation of cell centers to the earlier MLTs, however, cannot directly be derived from the squeezing mechanism by Atkinson and Hutchison [1978]. It is a result of more synthesized effect. The fact that violation of the mirror symmetry occurs more severely on the nightside suggests that the model for the cell rotation must consider the oval structure. In Plate 1 the overall convection pattern is mirror symmetric 3.3. Nightside Convection with respect to the sign of IMF By, and the line connecting cell centers is aligned along the dawn-dusk direction. In Plate 1 the distribution of FAC is also mirror symmetric with respect to The IMF By dependence on ionospheric convection extends well into the nightside. In the results for constant conductivity cases shown in Plates 1 and 2 (cases 1 and 2), the nightside the sign of IMF By. These feature are maintained even in Plate convection does not exhibit a remarkable breakdown in mirror 2; that is, the overall patterns are mirror symmetric with respecto the sign of IMF By and the line connecting centers of convection cells is aligned along the dawn-dusk direction. Comparing Plate 3 with Plates 1 and 2, many changes are seen in the convection patterns both for IMF By_ and By +. A drastic change is seen in the overall pattern of convections in Plate 3 in that the mirror symmetry with respecto the sign of IMF By is no longer maintained and cell centers are shifted to the earlier MLTs (32, 33, 34, and 35). Violation of the mirror symmetry occurs more severely on the nightside. From a mutual comparison among Plates 1, 2, and 3, it is apparent that a symmetry. It is apparent that the results under constant conductivity assumption do not reproduce observed nightside convection patterns completely. In Plate 1, well-developed convection shear is observable around 70 ø latitude in the premidnight (17) (postmidnight, 18) for IMF By_ (By+) correlating with the region-1 FAC extending from the dusk (dawn). For IMF By_, consequently, the dusk cell is more crescent-shaped (13) and the flow in the nightside polar cap is slightly inclined (19) from postnoon to postmidnight, whereas for IMF By+ the dawn cell is more crescent-shaped (16) and the flow in the nightside polar cap is nonuniform ionosphericonductivity can modify the simulated slightly inclined (la) from prenoon to premidnight. global convection pattern so as to match both the satellite observations [Heppner and Maynard, 1987; Weimer, 1995] and ground-based observations [Ruohoniemi and Greenwald, 1996; Contrary to Plate 1, the nightside convection reversal around 70 ø in Plate 2 is more apparent in postmidnight (23) (premidnight, 24) for IMF By_ (By+). Consequently, distinc- Lyons et al., 1996] quite well. From a more precise comparison, tive features characterizing the crescent/round cells become the ionospheric convection pattern simulated under the realistic ionospheric conductivity (Plate 3) shows a particular coless apparent in Plate 2 compared with Plate 1. This change in the convection pattern is apparently correlated with that in the incidence with that obtained from satellite observations distribution of nightside region-1 FAC. It is concluded from through the pattern-recognition-based approach [Heppner and Maynard, 1987], with respect to the features in breakdown of mirror symmetry, in the flow line connection between the polar these results that the nightside region-1 FAC is mainly controlled by the magnetospheric structure before the southward turning of the IMF (t < 0). It depends on the merging cell cap and the auroral oval (36 and 37), and in the Harang convection under a northward IMF condition with nonzero discontinuity (38 and 39). About the latter two points, details will be discussed in the following sections. Atkinson and Hutchison [1978] proposed a mechanism for the effect of day-night conductivity gradient on the breakdown of the mirror symmetry. Ionospheric conductance produced by solar illumination decreases from the dayside to the nightside across the polar cap. Assume a sharp day-nighterminator that crosses the polar cap from 0600 to 1800 LT separating two regions of uniform conductance on either side. The primary electric field is imposed across the polar cap parallel to the dawn-dusk axis. While the primary Pedersen currents flow everywhere along regions of constant conductivity, the primary Hall currents, which flow parallel to the midnight-to-noon direction, are interrupted by the conductivity change at the terminator. In order to satisfy div J = 0 in this circumstance, the IMF By [Tanaka, 1999]. The nightside convection shown in Plate 1 coincides with the radar observations in a sense that the convection reversal, a distinct feature near 70 ø latitude of the nightside, can be discerned only in the sector of the crescent-shaped cell, that is, premidnight (17) for IMF By_ and postmidnight (18) for IMF By+ [Leonard et al., 1995; Rouhoniemi and Greenwald, 1996]. To explain this feature seen in radar observations, Cowley et al. [1991] proposed a conceptual model for the IMF By dependence of the nightside ionospheric convection that the asymmetrical transport of tubes of open flux into the crescentshaped cell results in compressive force and compensatory drift across the pole. Comparing Plate 1 (case 1) with Plate 2 (case 2), however, this mechanism is shown not to be plausible. From Plates 1 and 2 we can conclude that nightside convection is electric field is modified so as to include a secondary compo- rather a remnant of the crescent cell convection under the nent which is oriented toward the terminator on both sides of northward IMF condition (t < 0) [Tanaka, 1999]. The posiit. Consequently, the antisunward flow in the polar cap is squeezed to the dawnside to break the mirror symmetry with respecto IMF By. This makes the dusk cell look round and the dawn cell look crescent-shaped, in the same sense for tion of nightside convection reversal depends on the previous IMF By at t < 0. Therefore present results do not support Cowley et al. [1991]. In statistical results adopting the binaveraging approach, IMF By effects on the nightside convecnorthern convection under IMF By +. For IMF By +, therefore, tion may depend on the nature of IMF time variation. The the round/crescent cell pattern in the Northern Hemisphere is enhanced, while for IMF By_ the flow lines across the polar cap have a more uniform distribution. This squeezing of the antisunward flow also generates a wide zone of westward flow (31) that appears for IMF By+ in the dayside around 70o-80 ø latitude (Plate 3). convection reversal appears in premidnight (postmidnight) for IMF By_ (By+) if the probability of constant IMF By case prevails rather than that of randomly changing IMF By case. As shown in the previous section, mirror symmetry in the convection patterns beaks down in Plate 3. Present simulation reproduces this feature together with IMF By-dependent fine

9 TANAKA: IMF By AND AURORAL CONDUCTANCE EFFECTS 24,513 signatures about the flow line connection between the polar cap and the auroral oval. In the present simulation a fine grid spacing is realized in the ionosphere to make the fine signatures clearer than previous simulations [Fedder et al., 1998]. In Plate 3 the antisunward flow in the nightside polar cap is nearly aligned with the noon-midnight meridian (3a) for IMF part of this section that the closing Hall current requires southward electric field along the nightside auroral oval regardless of the sign of IMF By, and causes the modification of convection so as to break mirror symmetry, as seen in the position of cell centers, the inclination of polar cap flow, and the configuration of nightside convection reversals (Plates 3). By _, while the flow in the polar cap has a significant inclination Plate 5 shows calculated distributions of the ionospheric from prenoon to premidnight (3b) for IMF By+. These struc- conductivity (contours) and FAC (colors) in case 3. The format tures emphasize the crescent/round cell configuration (35 and of Plate 5 is the same as that of Plate 1. It is seen from Plate 33). The flow line inclination for IMF By+ acts to give an 5 that the center of the nightside region-1 FAC is situated in additional rotation of the cell centers to the earlier MLTs. the region of steep conductivity gradient on the poleward edge Clarification of such synthesized effect is an important result of the auroral oval. This tendency is more apparent in the dusk obtained from simulations. Plate 4 shows calculated distribu- tions of the ionospheric potential (contours) and conductivity (colors) in case 3. The format of Plate 4 is the same as that of Plate 1. A high-conductivity area in the dayside is due to the solar EUV ionization, while a high-conductivity area in the nightside, which is surrounding the North Pole, is due to the auroral precipitation. Plate 4 reveals the relative role played by the polar cap and the auroral oval for the generation of convection pattern in the nightside ionosphere. For IMF By_, it is region than in the dawn region. If the ionosphericonductivity is spatially uniform, the FAC can be connected only to the Pedersen current, because the Hall current cannot have a divergent component in such a situation. In the nonuniform- conductivity region around the nightside auroral oval (Plate 5), on the other hand, not only the Pedersen current but also the Hall current can have a divergent component. This possibility can be seen from Plate 5 in the position of region-1 FAC with respect to the auroral oval. obvious from Plate 4 that flow lines which extend from the Plate 6 shows the divergences of ionospheric Hall (left) and polar cap to the auroral oval along the noon-midnight meridian divert toward dawn (36 and 41) when they enter the auroral Pedersen (right) currents by color together with their summation by contour, for IMF By_ under a realistic ionospheric oval. For IMF By+, on the other hand, a protruding tongue conductivity condition (case 3). In this plate the polar region is from the morning convection cell is extending along the polar cap-auroral oval boundary (37 and 42). Similar to the case in the right panel of Plate 1, the effects of the prior interval of the northward IMF B z remain on the nightside to make the tongue stand out more. Plate 4 clearly shows that these fine features shown with an enlargement, and current divergence (convergence), which can be connected to the downward (upward) FAC, is shown by a minus (plus) value. It is a natural consequence expected from the definition to see in Plate 6 that the summation of two divergence components drawn by the concome from the correlation between the nightside convection tour reproduces the region 1 and 2 FACs that are shown in the pattern and the distribution of the ionospheric conductivity. In Plate 4 the difference is apparent in the conductivity distributions for IMF By_ and By+. This difference is looked upon as an interhemispheric asymmetry. In the growth-phase interval the high-conductivity area in the nightside is determined mainly by diffuse precipitation as the projection of the plasma sheet. Therefore the interhemispheric asymmetry indicates that plasma distribution is affected by the ionospheric convection. As shown hitherto, nightside fine signatures of convection reproduced in the present simulation are in excellent agreement with the satellite observations, particularly with that of left panels of Plates 3 and 5. On the dayside the distribution of downward (upward) region-1 FAC coincides with that of the Pedersen divergence (convergence) quite well, indicating that the dayside region-1 FAC is mainly connected to the Pedersen current. It is quite natural to observe such a situation, because ionospheric conductivity is nearly uniform in the dayside region. In the region extending from premidnight to dawn, on the other hand, the distribution of region-1 FAC shows no remarkable coincidence with either the Hall or Pedersen divergence. A unique feature seen in this region is that two-band structures having opposite signs are extending longitudinally on both Heppner and Maynard [1987]. We can explain IMF Bysides of the auroral oval. In the band structure the Hall and dependent modification of the nightside convection seen here by considering the Hall current closure process of region-1 FAC, as will be shown in the next section. Pedersen divergences almost perfectly cancel each other to cause small FAC. This is a well-known configuration of the Cowling channel, which connects the morning region-1 FAC to the evening region-1 FAC through the Hall current construct Hall Current Closure ing the westward electrojet [Richmond and Kamide, 1988]. Under a uniform-conductivity condition the ionospheric On the duskside the upward region-1 FAC must overlap Hall current is perfectly separated from other current systems with current convergence (red area), since total current must in the M-I system. Since the Hall current is, in this case, closed inside the ionosphere, it plays no role in the construction of current loops in the M-I coupling system. Under the nonuniform-conductivity condition, on the other hand, the Hall current can participate in the formation of the M-I current system. For instance, the Hall current through the auroral oval can connect the morning and evening region-1 FACs under the nonuniform-conductivity condition, whereas only the Pedersen satisfy the current continuity. It is apparent in Plate 6 that in the evening sector the poleward half of the region-1 FAC is owed to the Hall current convergence while the equatorward half is owed to the Pedersen current convergence. Even in the morning region, the downward region-1 FAC is connected to the Hall current at the poleward edge. In this region the region-1 FAC must overlap with current divergence (blue area). A small blue spot of the Hall current divergence is seen around current can connect the FACs under the uniform-conductivity 0800 MLT to occupy the poleward side of the dawnside recondition. This closure process including the Hall current is the key mechanism for the breakdown of mirror symmetry with gion-1 FAC. As shown in Plate 6, the Hall current can be directly connected to the FAC under the nonuniformrespect to the sign of IMF By. It will be shown in the following conductivity condition.

10 24,514 TANAKA: IMF By AND AURORAL CONDUCTANCE EFFECTS In order to establish the Hall current closure process of the region-1 FACs through the auroral oval, it is essential to generate southward electric field in the conduction channel at the midnight. To maintain this electric field, in turn, a sufficient number of flow lines must enter the premidnight auroral oval and divert toward dawn after entering the auroral oval. Thus tinuity is extending slant equatorward with increasing MLT while the region-1 FAC distributes along constant latitude. From the satellite observation [Heppner and Maynard, 1987; Weimer, 1995], the Harang discontinuity in the evening can be seen for both cases of IMF By_ and By+. For IMF By_ the Harang discontinuity is formed through the diversion of flow sufficient flow lines must intersect the polar cap-oval boundary toward dawn (36), which occurs on the high-latitude side at in the premidnight region. For IMF By_, this requirement is about 70 ø latitude, and successive diversion toward dusk (38), automatically satisfied even if the flow line in the polar cap is which occurs on the low-latitude side at about 65 ø latitude. For directed from noon to midnight, because in the dayside suffi- IMF By+, in addition, a protruding tongue structure (37) of cient flow lines are concentrated to the duskside due to the the convection is extending along 70 ø latitude from the mornflow deflection toward dusk. For IMF By+, on the other hand, ing convection cell and helping the generation of the Harang flow lines must be transferred in the polar cap from prenoon to premidnight in order to maintain sufficient flow lines to the premidnight oval. The protruding tongue also plays a role in concentrating flow lines to the premidnight region. These rediscontinuity. Although the position of the tongue structure roughly coincides with that of the morning region-1 FAC, there remains a little discrepancy between them. Consequently, eastward flow on the high-latitude side of the Harang discontinuity sults indicate that nightside ionospheri convection is strongly is bordered on the poleward side of 70 ø latitude by a region of controlled by the auroral conductivity. It is equivalent to say that the nightside convection in the M-I coupling system is westward flow. Because of this structure, the position of the Harang discontinuity between 2200 and 0030 MLT is pushed controlled and modified to a large extent by the secondary down to lower latitudes for IMF By+ than for By_. electric field in the ionosphere. In order to maintain perfect mirror symmetry with respect to IMF By, electric field in the midnight auroral oval must be At a fixed latitude near 65 ø it implies that the discontinuity will appear roughly 1 hour earlier in the Northern Hemisphere with IMF By+ than for IMF By_ [Heppner and Maynard, 1987]. northward for IMF By+ and, in this case, eastward electrojet Rodger et al. [1984] also found the Harang discontinuities in the must be generated in the auroral oval. This situation (mirror symmetry and eastward electrojet) is prohibited because it is against the formation of the Cowling channel. Thus we can conclude that the nightside convection is determined so as to Northern Hemisphere to be earlier for IMF By+ than IMF By_. These features coincide with the results in Plates 3 and 4. The traditional theories of the Harang discontinuity find its primary cause in the magnetosphere [Erickson et al., 1991]. In maintain the Cowling channel and the current closure process, these theories, magnetosphericonvection plays a primary role and consequently mirror symmetry cannot be held in the nightside. in the formation of the Harang discontinuity, as shown below. The dawnside plasma depletion caused by the curvature/ gradient drift in the tail causes cross-tail divergence of the 3.5. Itarang Discontinuity The ionospheric control of the nightside convection discussed in the previous section is apparently related to a longcross-tail drift current to require upward field-aligned current from the ionosphere. Current closure requires electric fields that are directed toward the center of the upward current. The lasting question about the cause of the Harang discontinuity present results, however, are against these theories. The dom- [Erickson et al., 1991]. On the basis of magnetometer observations, Harang [1946] first found that a discontinuity exists in the auroral current system in the premidnight region. These curinant role of the ionospheric secondary electric field in the formation of the Harang discontinuity is apparent in Plate 4. This result is deduced from two important features seen in the rents are predominantly Hall currents so that the discontinuity convection pattern obtained from the simulation. At first, disdelineates reversals of the ionospheric electric field and the crepancy between the region-1 FAC and the convection reverplasma flow. Thus poleward of the discontinuity one should sal is apparent in the premidnight region in Plate 3. This result observe a southward electric field and eastward plasma flow, and equatorward of the discontinuity one should observe a northward electric field and westward plasma flow. In Plate 1 the dusk (13) (dawn, 16) cell makes a larger contribution to the convection on the midnight meridian for shows that the Harang discontinuity is not directly generated by the magnetospheric driver. Second, the result shown in Plate 4 indicates that the nightside convection is strongly controlled by the ionospheric conductivity, showing that the secondary electric field generated in the ionosphere plays an im- IMF By_ (By+), since the convection reversal in the night portant role in the formation of the Harang discontinuity. sector, a distinct feature, can be discerned only in the sector of the crescent-shaped cell. The Harang discontinuity-like structure in the evening [Harang, 1946] therefore appears only for Heppner and Maynard [1987] observed the Harang discontinuity in the evening for both IMF By_ and By +. On the other hand, Rouhoniemi and GreenwaM [1996] observed it only for IMF By_ that produces a crescent-shaped cell in the dusk IMF By_. The reason for this discrepancy due perhaps, to sector. The origin of the Harang discontinuity is, in this case, the differences in methods of statistical treatment and in the estimated to be in the magnetosphere, because nightside convection reversal is strictly correlated with the region-1 FAC extending from dayside. However, its feature does not match observable area. Heppner and Maynard [1987] estimated the convection patterns from the pattern-recognition-based approach. In their statistics the condition for the IMF is that the with an observational feature that the Harang discontinuity IMF remains within a single sector throughout the period 15 extends slant equatorward with increasing MLT. In Plates 3 and 4 the configurations of the Harang discontinuity are reproduced both for IMF By_ (38 and 43) and IMF and 90 min in advance. On the other hand, Rouhoniemi and Greenwald [1996] used a bin-averaging approach. In this method, all IMF data were averaged within 12-min intervals. By+ (39 and 44). Contrary to Plates 1 and 2, a discrepancy On the nightside, convection data were tagged with the IMF between the flow reversal and region-1 FAC is apparent along the Harang discontinuity in the evening. The Harang disconaveraged over three previous intervals. It is naturally expected that the pattern-recognition-based approach tends to prevent

11 TANAKA: IMF By AND AURORAL CONDUCTANCE EFFECTS 24,515 the smoothing out of fine signatures. In addition, the HF radars pointing toward the polar direction are generally insensimidnight meridian for IMF By_, while the flow in the nightside polar cap has a significant inclination from prenoon to tive to the observation of lower-latitude area. premidnight for IMF By+. This feature and other fine signa- As mentioned before, the nonuniform-conductivity effect tures about the flow connection between the polar cap and the modifies the convection pattern so as to emphasize distinctive auroral oval in the nightside are explained quite well by confeatures seen in the convection under a uniform conductivity. sidering the Hall current closure process of the region-1 FAC. This effect is remarkable in the crescent/round cell configuration and Harang discontinuity. Although the modification process by the ionosphere is mainly responsible for the Harang discontinuity, the position of the original convection shear The analysis of the dependence of nightside convection on IMF By and ionosphericonductivity shows that the Harang discontinuity is attributed partially to the structure of magnetospheric driver but mainly to the effect of nonuniform auroral caused by the magnetospheric driver itself does not severely conductivity. Thus the cause of the Harang discontinuity is differ from that of the Harang discontinuity. Strictly speaking, surely the synthesized effect of magnetospheric driver and therefore, the Harang discontinuity is a synthesized effect of several processes in the M-I system, although the most impornonuniform ionospheric conductivity. It is shown from the present simulation that not only global tant contribution is from the auroral conductance effect. For features but also fine signatures seen in the results of Heppner IMF By+, convection reversal seen by the radar observations and Maynard [1987] can be interpreted as the manifestations of in the postmidnight may be a migration of the protruding dynamical processes operating in the M-I convection system. It tongue structure under a bin-averaging approach. It is con- is concluded therefore that satellite observations summarized cluded from the present investigation that satellite observa- by adopting the pattern-recognition-based approach [Heppner tions summarized by adopting the pattern-recognition-based and Maynard, 1987; Weimer, 1995] represent the convection approach represent the convection pattern under a quasi- pattern under a quasi-stationary condition better than do the stationary condition better than do the radar observations radar observationsummarized by using the bin-averaging apsummarized by using the bin-averaging approach. proach [Leonard et al., 1995; Rouhoniemi and Greenwald, 1996]. 4. Conclusion The relationship between the IMF and the ionospheric electric potential (convection pattern) has been the topic of previous experimental and theoretical discussions by many authors. The main intention aimed at by them is to understand the dependence of convection on the IMF in the S-M-I coupling scheme. An essential problem is to clarify the relative roles played by the magnetospheric driver and the ionospheric modification. While the conceptual models hitherto investigated are based on the geometrical treatment of antiparallel merging, the simulation study can reproduce the convection pattern maintaining a self-consistency in the whole M-I system [Fedder et al., 1998; Tanaka, 1999]. This paper incorporates the ionospheric loading effect on the magnetosphere into the models in a more rigorous manner, by reproducing the ionospheric convection from the numerical simulation adopting a highresolution in the ionosphere and a sufficient self-consistency in the S-M-I system. From this method we can discuss the ionospheric loading effect on the convection in a more realistic manner. In this paper the main attention is paid to investigating the breaking of mirror symmetry with respect to the IMF By. For the southward IMF condition during the growth-phase interval, the two-cell convection pattern retains mirror symmetry in the dayside although IMF-By-dependent flow deflection is seen on newly opened field lines. From the comparison of convection patterns obtained under uniform and realistic ionospheric conductivity conditions, it is confirmed that the mirror symmetry of convection is broken mainly on the nightside by the nonuniform-ionospheric-conductivity effect. As a result, the round/crescent cell pattern in the Northern Hemisphere is enhanced for IMF By+ losing mirror symmetry, and cell cen- ters of convection are shifted to the earlier MLTs. It is more adequate to say that these convection patterns are more or less caused by the synthesized effect of more than one process rather than by a single elementary process. With a realistic ionospheric conductivity, the antisunward flow in the nightside polar cap is nearly aligned with the noon- Acknowledgment. Hiroshi Matsumoto thanks A. Nishida and J. Raeder for their assistance in evaluating this paper. References Atkinson, G., and D. Hutchison, Effect of the day night ionospheric conductivity gradient on polar cap convective flow, J. Geophys. Res., 83, 725, Caudal, G., and M. Blanc, Magnetosphericonvection during quiet or moderately disturbed times, Rev. Geophys., 26, 809, Cowley, S. W. H., J.P. Morelli, and M. Lockwood, Dependence of convection flows and particle precipitation in the high-latitude dayside ionosphere on the X and Y components of the interplanetary magnetic field, J. Geophys. Res., 96, 5557, Crooker, N. U., Dayside merging and cusp geometry, J. Geophys. Res., 84, 951, Crooker, N. U., Mapping the merging potential from the magnetopause to the ionosphere through the dayside cusp, J. Geophys. Res., 93, 7338, de la Beaujardiere, O., V. B. Wickwar, and J. H. King, Sondrestrom radar observations of the effect of the IMF By component on polar cap convection, in Solar Wind-Magnetosphere Coupling, edited by Y. Kamide and J. A. Slavin, p. 495, D. Reidel, Norwell, Mass., Erickson, G. M., R. W. Spiro, and R. A. Wolf, The physics of the Harang discontinuity, J. Geophys. Res., 96, 1633, Fedder, J. A., S. P. Slinker, and J. G. Lyon, A comparison of global numerical simulation results to data for the January 27-28, 1992, Geospace Environment Modeling challenge event, J. Geophys. Res., 103, 14,799, Foster, J. C., J. M. Holt, R. G. Musgrove, and D. S. Evans, Solar wind dependencies of high-latitude convection and precipitation, in Solar Wind-Magnetosphere Coupling, edited by Y. Kamide and J. A. Slavin, p. 477, D. Reidel, Norwell, Mass., Harang, L., Polar geomagnetic storms, J. Geophys. Res., 51,353, Harel, M., R. A. Wolf, P. H. Reiff, R. W. Spiro, W. J. Burke, F. J. Rich, and M. Smiddy, Quantitative simulation of a magnetospheric substorm, 1, Model logic and overview, J. Geophys. Res., 86, 2217, Heppner, J.P., and N. C. Maynard, Empirical high-latitude electric field models, J. Geophys. Res., 92, 4467, Holt, J. M., R. H. Wand, J. V. Evans, and W. L. Oliver, Empirical models for the plasma convection at high latitudes from Millstone Hill observations, J. Geophys. Res., 92, 203, Iijima, T., Field-aligned currents in geospace: Substance and significance, in Magnetospheric Current Systems, Geophys. Monogr. Ser., vol.

12 24,516 TANAKA: IMF B, AND AURORAL CONDUCTANCE EFFECTS 118, edited by S. Ohtani et al., p. 107, AGU, Washington, D.C., Iijima, T., and T. A. Potemra, The amplitude distribution of fieldaligned currents at northern high latitudes observed by Triad, J. Geophys. Res., 81, 2165, Iijima, T., T. A. Potemra, L. J. Zanetti, and P. F. Bythrow, Large-scale Birkeland currents in the dayside polar region during strong northward IMF: A new Birkeland current system, J. Geophys. Res., 89, 7441, Leonard, J. M., M. Pinnock, A. S. Rodger, J. R. Dudeney, R. A. Greenwald, and K. B. Baker, Ionospheric plasma convection in the Southern Hemisphere, J. Atmos. Terr. Phys., 57, 889, 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, Lu, G., et al., Interhemispheric asymmetry of the high-latitude ionospheric convection pattern, J. Geophys. Res., 99, 6491, Lyons, L. R., G. Lu, O. de la Beaujardiere, and F. J. Rich, Synoptic maps of polar caps for stable interplanetary magnetic field intervals during the January 1992 geospacenvironment modeling campaign, J. Geophys. Res., 101, 27,283, Moses, J. J., G. L. Siscoe, R. A. Heelis, and J. D. Winningham, A model for multiple throat structures in the polar cap flow entry region, J. Geophys. Res., 93, 9785, Nishida, A., T. Mukai, T. Yamamoto, S. Kokubun, and K. Maezawa, A unified model of the magnetotail convection in geomagnetically quiet and active times, J. Geophys. Res., 103, 4409, Ogino, T., R. J. Walker, M. Ashour-Abdalla, and J. M. Dawson, An MHD simulation of the effects of the interplanetary 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, Richmond, A.D., and Y. Kamide, Mapping electrodynamic features of the high-latitude ionosphere from localized observations: Technique, J. Geophys. Res., 93, 5741, Rodger, A. S., S. W. H. Cowley, M. J. Brown, M. Pinnock, and D. A. Simmons, Dawn-dusk (Y) component of the interplanetary mag- netic field and the local time of the Harang discontinuity, Planet. Space Sci., 32, 1021, 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, Sonnerup, B. U. O., Theory of the low-latitude boundary layer, J. Geophys. Res., 85, 2017, Stern, D. P., The origin of the Birkeland current, Rev. Geophys., 21, 125, Tanaka, T., Finite volume TVD scheme on an unstructured grid system for three-dimensional MHD simulation of inhomogeneousystems including strong background potential fields, J. Comput. Phys., 111, 381, 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, Tanaka, T., Configuration of the magnetosphere-ionosphere convection system under northward IMF condition with nonzero IMF By, J. Geophys. Res., 104, 14,683, Tanaka, T., Field-aligned current systems in the numerically simulated magnetosphere, in Magnetospheric Current Systems, Geophys. Monogr. Ser., vol. 118, edited by S. Ohtani et al., p. 53, AGU, Washington, D.C., 2000a. Tanaka, T., The state transition model of the substorm onset, J. Geophys. Res., 105, 21,081, 2000b. Vasyliunas, V. M., Mathematical models of the magnetosphericon- vection and its coupling to the ionosphere, in Particles and Fields in the Magnetosphere, edited by B. M. McCormac, p. 60, D. Reidel, Norwell, Mass., 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, T. Tanaka, Communications Research Laboratory, Nukuikitamachi, Koganei-shi, Tokyo , Japan. (tanaka@crl.go.jp) (Received September 5, 2000; revised April 11, 2001; accepted April 26, 2001.)