Since the first global image of the theta aurora from the DE 1 spacecraft was reported by Frank et al. [1982], studying this type of polar cap phenomenon has been of great interest. Unlike the usual quasi-circular aurora pattern, the theta aurora is characterized as a ring of luminosity termed the ``auroral oval'' and a ``bar'' across the polar cap from noon to midnight. The transpolar arc or ``bar'' can be as few as a hundred kilometers wide and its luminosity may be comparable to or substantially less than the average emissions from the auroral oval. This peculiar auroral pattern is a quasi-steady feature that can last for several hours. The sun-aligned arc moves slowly across the polar cap in the direction of the IMF By component in the northern hemisphere [Frank et al., 1985, 1986; Craven and Frank, 1991]. The motion is in the opposite direction in the southern hemisphere for the same sign of By [Craven and Frank, 1991; Craven et al., 1991]. Such auroral images occur under quiet geomagnetic conditions in the northern and southern polar caps [Frank et al., 1982, 1985, 1986; Nielsen et al., 1990].
Signatures of inverted V precipitating electrons were sometimes observed in the transpolar arcs by the DE 2 spacecraft [Frank et al., 1986]. The authors attributed these to the parallel potential drops above the ionosphere. From ion composition observations Peterson and Shelley [1984] concluded that plasmas in the transpolar arc originate in both the ionosphere and solar wind. Based on the similarity of spectral characteristic of the plasmas associated with transpolar arcs and the poleward portion of the nightside auroral oval, several authors argue that the transpolar arc is on closed field lines connected to the distant plasma sheet or plasma sheet boundary layer [Frank et al., 1982, 1986; Peterson and Shelley, 1984]. Menietti and Burch [1987] reached a similar conclusion from detailed analyses of electron distributions observed by DE 1. Filamentary energetic plasma structures, occasionally observed in the magnetotail lobes, are possible source regions of the polar cap arcs [Huang et al., 1987]. The composition of these structures, measured by the ISEE spacecraft, was consistent with origins in the plasma sheet or plasma sheet boundary layer. Assuming that theta auroras exist simultaneously in both hemispheres, Huang et al. [1989] showed that a theta aurora was observed in the northern polar cap while its associated plasmas were detected in the southern lobe of the magnetotail. Images from the DE 1 and Viking spacecraft showed theta auroral arcs can coexist in both hemispheres, aligned parallel to one another but on opposite sides of the magnetic pole [Craven et al., 1991]. They related the location of the arcs to the poleward edge of the "horse-collar aurora" seen for contracted polar cap situations [Hones et al., 1989].
Several models of the polar cap morphology have been introduced for theta auroras. From the plasma and field measurements by the DE 1 and DE 2 spacecraft, Frank et al. [1982, 1986] concluded that the polar cap consisted of two open lobe cells bifurcated by a closed region which mapped to the plasma sheet boundary layer or the distant plasma sheet. The ionospheric convection is typical of four-cell patterns observed with positive IMF Bz. Such a polar cap configuration has been numerically simulated by Toffoletto and Hill [1989, 1990] for steadily northward IMF conditions. Using particle measurements from the DMSP spacecraft, Meng [1981] proposed that the sun-aligned polar arcs were poleward extensions of the dawn and dusk sides of the auroral oval observed during the geomagnetically quiet times. Murphree et al. [1982] reached a similar conclusion by analyzing auroral images from ISIS 2 during periods of northward IMF. These results required a highly contracted polar cap and a dawn-dusk asymmetry or tilting of the plasma sheet. Based on detailed analyses of ion compositions measured by DE 1 during the theta auroral event reported by Frank et al. [1982], Peterson and Shelley [1984] favored the latter interpretation.
Reiff and Burch [1985] proposed a global convection model in which three-cell convection patterns develop when the IMF has northward and large By components. The theta auroral arc is depicted at the central convection reversal of the dawn NBZ current region [Iijima et al., 1984]. On a theoretical basis, assuming the magnetic field merging occurs in the cusp region and reconnection near the flanks of the magnetotail with mostly northward IMF, Kan and Burke [1985] proposed the transpolar arc develops on sunward convecting, closed field lines in the central polar cap. The dynamo region for the theta auroral arc develops on tailward convecting, closed field lines in the plasma sheet that are magnetically connected to the central polar cap. A northward IMF merging model containing lobe cells, merging cells, and viscous cells was used to explain dual theta auroral arcs [Burch et al., 1992]. It involved dayside merging both at the high latitudes on open field lines and at lower latitudes on closed field lines as well as tail reconnection. They argue that dual theta auroral arcs are located on sunward convecting, closed field lines in the central polar cap with a narrow channel of antisunward flows in-between. From the observations acquired with DMSP, ISEE, and Viking particle instruments, Henderson et al. [1996] interpreted that open field lines can partially intrude into the region between the horse collar arcs and the auroral oval. In contrast to the merging driven process, Lundin et al. [1991] investigated the possible role of viscous interactions and plasma diffusion in the low-latitude boundary layer [e.g., Axford and Hines, 1961; Eastman et al., 1976]. They argue that the theta aurora maps to the boundary between the open and closed magnetic field lines. It connects to the low-latitude boundary layer on the dayside and the plasma sheet boundary layer on the nightside. In this configuration, low-latitude boundary layer expands into the polar cap which is considerably reduced in size.
Different from the steady state models, Newell and Meng [1995] proposed a time dependent model for theta aurora formation. The authors argue that theta auroras exclusively occur when the IMF turns southward after a prolonged northward interval. They suggest that after the turning, IMF Bz must stay negative at least 10 mins for the sun-aligned auroral arc that appears on the dawn or dusk side of the oval to evolve into the transpolar arc. The dawn-dusk motion of the transpolar arc is in the direction of IMF By and halts if IMF turns northward again. They also argue that theta auroras can last for several hours only if the IMF Bz component changes sign every hour or two. In this model, the IMF Bz component plays the most significant role. They claim IMF measurements from IMP 8 show southward turnings before the occurrence of theta auroras. However, in the theta auroral event reported by Huang et al. [1989], IMF Bz observed by ISEE 3 was never negative during the hour and a half before the first appearance of the transpolar arc (see Figure 1 of Huang et al. [1989]). Furthermore, in the event reported by Craven et al. [1991], the first southward turning of the IMF following two hours of northward Bz, occurred several minutes after DE 1 observed a polar arc (cf. Figures 1 and 5 of Craven et al. [1991]). The Bz component was negative for only about 3 mins and a second southward turning of the IMF followed about 10 mins later. Nevertheless, it should be noted that the IMF By changed sign several minutes before the occurrence of the transpolar arcs in both events. We investigate the hypothesis that the change in IMF By triggered these two theta auroral events. During the refereeing process of our paper we were first made aware of the unpublished work of Cumnock et al. (in press), who discussed five events in which the transpolar arc developed after IMF By changed sign while Bz remained positive; these events will be shown to fit naturally within the framework of the model independently developed in this paper.
After the Polar spacecraft was launched, with the Wind spacecraft already in orbit, long duration and simultaneous observations of the northern polar cap and interplanetary conditions became available. In this paper, we present a model to explain the magnetic topology required for the formation of theta auroras and preliminary results of theta aurora observations from three particle instruments (Hydra, CEPPAD, and TIMAS), the electric field instrument (EFI), and the VIS imager on Polar, the magnetometers (MFI) on Wind, and SuperDARN ground-based radars. Although we find evidence to support the model of Newell and Meng [1995], there are also counter examples. Our model generalizes that of Newell and Meng, not necessarily requiring southward turning of the IMF and flipping the sign of Bzto generate and maintain theta auroras. The key to the formation of theta auroras lies in the consequences of "merging" boundary conditions on the dayside magnetopause changing with IMF By and Bz.
As summarized in Figure 1, theta auroras can be generated from five different initial and boundary conditions (two of them are the mirror images of the other two).
The first, associated with IMF Bz > 0 and By 
0, was proposed by Kan and Burke [1985], as shown in Figure 1a.
They consider the formation of theta auroral arcs that begin as bulges from the midnight auroral oval, then step across the polar cap into the noon sector.
Craven et al. [1986] observed a transpolar arc showing a similar evolution.
The auroral arc started ~30 mins after a sudden commencement when the IMF Bz was positive.
The IMF By, however, appeared to be the dominant component, unlike the suggestion of Kan and Burke [1985].
The second (third), associated with IMF Bz > 0 and By > 0 (By < 0), followed by a southward turning of the IMF, was proposed by Newell and Meng [1995].
Figure 1b shows the configurations of the second condition.
Configurations for By < 0 are the mirror images of Figure 1b.
The fourth and the fifth conditions are associated with By changing sign while Bz remains positive, after a prolonged period of northward IMF.
The example shown in Figure 1c is for a dawnward turning of By; a duskward turning of By produces a mirror image of this case.
The cases presented in Figures 1b and 1c represent the most commonly observed genesis of theta bar, from along the poleward boundary of the dawn or dusk side auroral oval.
Note that the figure is used here only to illustrate the distinct initial and boundary conditions that can generate the theta auroras.
A detailed description of our model is presented in the next section.
Please send questions, comments, or suggestions about the paper to:
Shen-Wu Chang
Department of Physics and Astronomy, The University of Iowa, Iowa City, IA 52242
Phone:(319)335-3828; Fax:(319)335-1753;
swc@space-theory.physics.uiowa.edu