Along the auroral zone at ionosphere altitudes the westward auroral electrojet is located in the morning sector and the eastward one is located in the evening sector, and they are short-circuited by reverse currents mainly across the polar cap [Chapman, 1935]. Between the electrojets in the auroral belt in the premidnight sector there exists a discontinuity in the ionosphere current [Harang, 1946].
To pass to a concept of the auroral oval, we need a new idea of spatio-temporal distribution of the geomagnetic field variations at high latitudes. Akasofu et al. , Fel'dshtein , and Fel'dshtein and Zaitsev  showed that during disturbed periods the westward electrojet is located in the auroral oval. The westward electrojet extends from the latitudes of the auroral belt in the morning sector to the evening sector to the pole of the eastward electrojet. As a result of the overlap of electrojets there appears a "slit" between them whose latitude increases as one passes from near-midnight hours to evening ones. In the models developed by Fel'dshtein  and Akasofu  for electrojets, this slit was called the "Harang discontinuity" [Heppner, 1972], and since then this term has been in general use. The model of the spatio-temporal distribution of the geomagnetic field variations given by Fel'dshtein and Zaitsev  appeared to be adequate to the general structure of geophysical phenomena at high latitudes during magnetospheric disturbances [Akasofu, 1977].
Rostoker [1992a, b] criticized the present idea of the Harang discontinuity. According to his new model of a three-dimensional current system during the active phase of a magnetospheric substorm the westward electrojet splits into two isolated segments with discontinuities in latitude and local time in the midnight sector: in the morning sector the electrojet is at the latitudes of the poleward auroral zone, and in the evening sector it is at the pole of the eastward electrojet. Both portions of the westward current are not connected by a current crossing the midnight sector. Their physical natures are also different because magnetic lines connect them to different parts of the magnetosphere. The existence of two active areas separated in latitude by approximately 1000 km was demonstrated by Rostoker [1992a] through a Viking image of auroral fluorescence in the ultraviolet on September 19, 1986, at 1633-1637 UT.
However, no investigations have been carried out for the character of the current system in the vicinity of the overlap between the westward and eastward electrojets. For the substorm under consideration, there is a unique opportunity to carry out such investigations using the data from three chains of magnetic stations which were in operation in that period in the region of the overlap between electrojets along the geomagnetic meridians close to each other (~110o, ~145o, and ~157o).
The goal of this paper is to give a comprehensive analysis of variations in the geomagnetic field and associated ionospheric currents in this region and of a relationship between the currents and auroral emissions during the active phase of the magnetospheric substorm (on the basis of simultaneous observations) on September 19, 1986.
Figure 1 presents a sequence of two Viking images of auroral emission in ultraviolet on September 19, 1986, for the northern high-latitude ionosphere. The domain of the most intense emission is hachured; that of the less intense emission is dotted. In the sector between evening and premidnight hours the most intense emission takes place in two domains separated in latitudes. The magnetic observatories, whose positions are shown by crosses, cover this region rather densely. According to Rostoker [1992a, b], just here the latitudinal discontinuity in the westward electrojet is expected to be at the meridian where the westward and eastward electrojets "meet" each other at auroral latitudes.
According to auroral activity index AE [Data Book 20, 1991] the magnetospheric substorm started at ~1410 UT with a fast smooth increase in the westward electrojet, which lasted till 1645 UT, the highest intensity being of ~1000 nT. At that moment the magnetic field due to the auroral electrojets amounted up to ~1350 nT, after which it decreased to ~250 nT at 1800 UT. This substorm had no specific initiation phase. Possibly this can be explained through the fact that the magnetic field was not perfectly quiet before the substorm onset.
Figure 1 shows the auroral emission distribution at the substorm development phase. In this phase the most intense emission takes place at certain moments at the poleward and equatorward boundaries of the oval, a diffuse emission being between them [Starkov and Fel'dshtein, 1971].
To characterize the ionospheric currents from the magnetic field variations two procedures were used: a mapping of disturbance vectors in the horizontal plane, and a calculation of the electrojet latitudinal structure from the data of a meridional chain of stations. The first procedure has long been in use (see, for instance, Akasofu ); the second one became popular recently [Kotikov et al., 1987].
So let us briefly present the calculation procedure for current densities from the data of the meridional chain. The calculations through variations in H or Z components of the magnetic field can be carried out independently of one another. According to results of the computer modeling, the use of the H component is the better choice because its variations are caused mainly by natural ionospheric currents, and boundary effects influence them only slightly [Kotikov et al., 1991]. Thus we shall present the analysis procedure for the H component.
The latitudinal interval above the plane meridional chain of magnetometers is being broken up into 50 bands with half widths b within an infinitely thin layer at the altitude h = 115 km above the ground. The current density j in terms of amps per meter is uniform in latitude within each band, and we assume the currents to be infinitely long. The values of the horizontal component dH from the current element with the dX width in the point with the x coordinate counted from the center of the current band on the ground are
where dH is in nT and m = 4p x10-7 H m-1.
The field from the current band with the coordinates in the interval from (x-b) to (x+b) is determined by the formula
For N observational points and N current systems we have N equations with N terms similar to (1). In the matrix form the equation are written in the form H = Aj, where A is a square matrix N x N and its coefficients are the coefficients at j.
If there are N observational points along the meridional chain, one can obtain fairly detailed current distribution. Actually there are only a few stations along the meridian. If one needs a detailed meridional section of j, an ill-posed inverse problem should be solved. The latter is solved by the interpolation method [Tikhonov and Arsenin, 1974] in which a minimum of the specially formed interpolation functional F is sought for. In our case it has the following form
The functional F is continuous in terms of j and has continuous derivatives. To find the solution, we set the first functional derivatives with respect to the vector j equal to zero. The coefficient matrix of a set of equations which we obtain through differentiation with respect to j is constant for a fixed distribution of observational points and thus can be predetermined.
At the final stage, we solve the equations of the following type (the mth one):
where Aim are the elements of the initial matrix A. The physical sense of the interpolation parameter a is in its role as a regulator, which determines the balance between the minimum of the total current and the minimum of the discrepancy between the observed and calculated field values. The parameter a is constant for a fixed distribution of the observational points and is determined from the requirement for minimum of the discrepancy between the experimental field values and rated ones.
At 1410 UT (this moment is not shown in Figure 2) the DT values were not higher than 20 nT, and spatial orientations of the vectors were random. The disturbance with respect to the AE index was the lowest at that moment, and the DT value characterizes the uncertainty in choice of a level from which the quiet field is counted. At 1525 UT (see Figure 2a), westward and eastward currents of approximately the same intensity appear within the auroral zone (the directions and intensity of the current being determined by the vector obtained with the vector DT rotation through 90o clockwise). As one passes to the evening hours, the latitude of the region, where westward currents are located, increases. The boundary between two electrojets becomes clearly pronounced and situated to the north of Scandinavia, between Amderma and Zhelanie Cape, Kamennyi Cape and Sayakha (dashed line in Figure 2).
At 1633 UT (see Figure 2b) the disturbance intensification is followed by a sharp intensification of the westward current which exceeds the eastward current in intensity and by a displacement of the boundary between the currents of opposite directions (the dashed line in Figure 2b) toward the earlier hours of local time. At the meridional chains, as the pole is approached, one can observe a broadening of the westward electrojet, which is a distinctive feature of the near-midnight sector during magnetosphere substorms. The westward current reaches its highest intensity in the vicinity of the auroral bulge [Akasofu, 1968]. The fact that intense westward currents exist at meridional chains at 145o and 157o at 2100-2300 MLT does not confirm the assumption of Rostoker [1992a] that the westward current does not cross the midnight meridian.
Comparing the DT distribution and auroral emission (see Figures 1 and 2b), we can see that the westward current is located within the regions of intense discrete auroral forms, and the eastward current is located in the region of a weaker diffusive emission equatorward of the discrete forms. The westward current poleward of the eastward one cannot be considered as an isolated formation, but is a portion of the wide westward current which flows from the nighttime sector along the band of intense auroral emission. Apparently, this portion of the westward electrojet is just the ionospheric part of a three-dimensional loop arising at the substorm development phase. The conclusion that the westward electrojet crosses the near-midnight meridian and extends to the evening sector is confirmed also by the DT distribution at 1637 UT (see Figure 2c).
At the substorm recovery phase, as the disturbance becomes weaker, the boundary between the electrojets returns to its position before the substorm onset. At 1745 UT (see Figure 2d) the Harang discontinuity is to the north of Amderma and between Sayakha and Belyi Island. The intensities of the eastward and westward currents estimated through the disturbance vector become approximately equal again. The westward current in the evening sector to the pole of the eastward one and the current in the near-midnight sector compose a unified westward electrojet. In the evening this electrojet intensity decreases markedly.
The substorm development at 1525 UT (see Figure 3a) is followed by the appearance of the westward current with intensity of ~0.2 A m-1 at 70o >= F >= 65o at near-midnight hours (the meridian L = 157o) and at later evening hours (the meridian L = 145o). In addition to this current remaining at the same intensity, eastward currents appear, their highest intensities being of about 0.4 A m-1 at F = 75o and 62-65o. At the meridian L = 110o the eastward current with the highest intensity of 0.5 A m-1 takes place at F = 63-67o, and the westward current with ~0.15 A m-1 takes place at F = 74-76o. So at the beginning of the substorm active phase the single westward current in the near-midnight sector at evening hours gives way to two currents, a westward one and an eastward one, which are close to each other, the distance between the oppositely directed currents increasing when passing to earlier hours of local time.
At 1633 UT the westward electrojet is significantly extended (see Figure 3b), and there appears a fine structure in the form of the ionospheric current local maxima with intensities up to 1.5 A m-1. These maxima are spatially connected with bright forms of auroras whose positions are presented by dashed rectangles at each meridian. Kotikov et al.  obtained similar coincidence between the maxima in the ionospheric current latitudinal distribution and the bright auroral forms. At the meridian L = 110o, which crosses both the westward electrojet and the eastward one, only westward current coincides with the bright fluorescence, and the eastward current is to the equator of the bright auroral forms.
The main peculiarities of the interrelation between the emission and ionospheric currents, which were noticed at 1633 UT, can be seen also at 1637 UT (see Figure 3c) are: the filamentary structure of the westward electrojet with the highest currents of ~1.0 A m-1, the linkage between the intense emission and the westward current maxima, and the eastward current location equatorward of the bright discrete forms of the auroral emission. During the substorm recovery phase at 1745 UT (see Figure 3d) the auroral current intensities decrease, and the westward electrojet in the premidnight sector is displaced to higher latitudes giving way to the eastward current.
Table 2 presents integral intensities of equivalent ionospheric currents crossing the associated meridian. During the substorm development phase an abrupt increase in westward currents up to 800-850 kA occurs at L=157o, the eastward current intensity of about 150-200 kA being relatively constant at L = 110o. The westward current intensity decreases to ~530 kA at L = 145o and to 140-190 kA at L = 110o. Apparently, in the near-midnight sector there appears a partial branch of the westward current from the equatorial segment of the auroral belt to the magnetosphere. In the near-pole segment of the auroral belt the westward current crosses the midnight meridian and connects to the magnetosphere at the west edge of the auroral bulge deep in the evening sector. This current system differs much from that suggested by Starkov and Fel'dshtein , i.e., the westward current at L = 110o is not an isolated structure formation, but an extension of the westward electrojet from the nighttime sector to the evening one along the band of increased intensity of the auroral emission.
Rostoker [1992a] considered the results of the ionospheric currents model developed by Kamide et al.  as an additional argument for the existence of the westward electrojet discontinuity giving rise to two current portions which do not cross the midnight meridian. According to Kamide et al.  the ionospheric current distribution at 2300 UT on March 31, 1979, was such that the westward currents in the evening and morning sectors were in fact isolated from each other. However, these results might be caused by the limited data the region of the supposed westward electrojet discontinuity. This assumption is confirmed by the observed distribution of the equivalent current vectors at that moment, which is given by Kamide et al. . In several hours a large enough number of observational points comes to the region of the proposed discontinuity due to the Earth's rotation (for instance, at 0320 UT on April 1, 1979). According to Kamide et al. , in this case the westward current flowing from the night sector to the evening one is continuous. Thus the westward electrojet is a continuous formation in its near-pole segment, notwithstanding the branch of the ionospheric current portion to the magnetosphere in the near-midnight sector during the active substorm phase. We should conclude that its nighttime sector is projected on the plasma layer in the magnetotail within a large longitudinal interval, and the westward electrojet intensification and its fine structure in the near-pole segment are associated with processes at the outer periphery of the central part of the plasma sheet. Thus we do not need to associate the westward electrojet in the evening sector with processes at the edge of the low-latitude boundary layers, which processes cause the substorm generation [Rostoker and Eastman, 1987].
2. The westward electrojet in the evening sector is an extension of the intense westward current to the near-midnight sector and is aligned with the band of an intense auroral emission close to the near-pole boundary of the auroral oval. The suggestions by Rostoker [1992a] on the existence of a longitudinal discontinuity in the westward electrojet, for its dissociation into two independent portions and for an absence of the westward currents crossing the midnight meridian along the auroral oval, were not confirmed. The continuity of the electrojet between nighttime and evening hours within the high-latitude segment of the auroral oval argues for the existence of a common current source within the entire longitudinal interval where this electrojet portion is located. Apparently, the electrojet is projected by magnetic field lines on the periphery of the plasma layer in the magnetotail on each side of midnight.
3. The integrated current intensity within the westward electrojet decreases significantly from nighttime to evening hours. This can be associated with a magnetic field aligned branch of a part of the ionospheric current near midnight at the equatorial side of the oval from the ionosphere to the magnetosphere.
Akasofu, S.-I., Physics of Magnetospheric Substorms, 599 pp., D. Reidel, Norwell, Mass., 1977.
Akasofu, S.-I., S. Chapman, and C.-I. Meng, The polar electrojet, J. Atmos. Terr. Phys., 27, 1275-1293, 1965.
Chapman, S., The electric current systems of magnetic storms, Terr. Magn., 40, 349-363, 1935.
Data Book 20, Auroral electrojet indices, World Data Center C2 for Geomagnetism, May 1991.
Fel'dshtein, Ya. I., Aurora morphology and geomagnetism, Polyarnye siyaniya i svechenie nochnogo neba, 10, 121-125, 1963.
Fel'dshtein, Ya. I., and A. N. Zaitsev, Current system of Sd-variations at high latitudes for winter MGG, Geomagn. Aeron., 5 (5), 1123-1125, 1965.
Harang, L., The mean field of disturbance of polar geomagnetic storms, Terr. Magn., 51, 353-370, 1946.
Heppner, J., The Harang discontinuity in auroral belt currents, Geophys. Norv., 29, 105-120, 1972.
Kamide, Y., et al., Numerical modeling of ionospheric data for the CDAW-6 intervals, Rep. UAG 88, 197 pp., World Data Center A for Sol.-Terr. Phys., Boulder, Colo., 1983.
Kotikov, A. L., Y. O. Latov, and O. A. Troshishev, Structure of auroral electrojets by the data from a meridional chain of magnetic stations, Geophysica, 23 (2), 143-154, 1987.
Kotikov, A. L., et al., Structure of the auroral zone phenomena by the data of meridional chains of stations: Magnetic disturbances in the nighttime auroral zone and auroras, J. Atmos. Terr. Phys., 53 (3-4), 265-277, 1991.
Kotikov, A. L., et al., Filamentary structure of the westward electrojet in the midnight sector auroral distribution during substorms: Comparison with Viking auroral observations, J. Atmos. Terr. Phys., 55 (14), 1763-1774, 1993.
Rich, F. J., and N. C. Maynard, Consequences of using simple analytical functions for the high-latitude convection electric field, J. Geophys. Res., 94 (3), 3687-3701, 1989.
Rostoker, G., Mapping of regions of auroral activation in the ionosphere to the magnetotail, in Proceedings of the International Conference on Substorms, ICS 1, pp. 71-74, Kiruna, Sweden, March 1992a.
Rostoker, G., Phenomenology of the auroral oval and mapping to the magnetotail, Inst. Geophys. Meteorol. Space Phys. Geosci. Rep., 1 (1), 30-32, 1992b.
Rostoker, G., and T. E. Eastman, A boundary layer model for magnetospheric substorms, J. Geophys. Res., 92, 12,187-12,201, 1987.
Starkov, G. V., and Ya. I. Fel'dshtein, Auroral substorm, Geomagn. Aeron., 11 (3), 560-563, 1971.
Tikhonov, A. N., and V. Ya. Arsenin, Procedures for Incorrect Problems Solution, 224 pp., Nauka, Moscow, 1974.