Multi-instrument data sets from the ground and satellites at both low and high altitude have provided new results concerning substorm onset and its source region in the magnetosphere. Twenty-six out of 37 substorm onset events showed evidence of azimuthally spaced auroral forms (AAFs) prior to the explosive poleward motion associated with optical substorm onset. The azimuthal wavelengths associated with these onsets were found to range between 132 and 583 km with a mean value of 307 _+ 115 km. The occurrence rate increased with decreasing wavelength down to a cutoff wavelength near 130 km. AAFs can span 8 hours of local time prior to onset and generally propagate eastward in the morning sector. Onset itself is, however, more localized spanning only about 1 hour local time. The average location of the peak intensity lbr 80 onsets was 65.9 + 3.5 CGMIat, 22.9 _+ 1.2 Mlt, whereas the average location of the AAF onsets was at 63.8 _+ 3.3 CGMIat, 22.9 _+ 1.1 Mlt. AAF onsets occur during time periods when the solar wind pressure is relatively high. These low-latitude wavelike onsets appear as precursors in the form of long-period magnetic pulsations (Pc 5 band) and frequently occur on the equatorward portion of the double oval distribution. AAFs brighten in conjunction with substom onset leading to the conclusion that they are a growth phase activity causally related to substorm onset. Precursor activity associated with these AAFs is also seen near geosynchronous orbit altitude and examples show the relationship between the various instrumental definitions of substorm onset. The implied mode number (30 to 135) derived from this work is inconsistent with cavity mode resonances but is consistent with a modified flute/ballooning instability which requires azimuthal pressure gradients. It is suggested that this instability exists in growth phase but that an additional factor exists in the premidnight sector which results in an explosive onset. The extended source region 'and the distance to the open-closed field line region constrain reconnection theory and local mechanisms for substom onset. It is demonstrated that multiple onset substorms can exist for which localized dipolarizations and the Pi 2 occur simultaneously with tail stretching existing elsewhere. Further, the tail can be less stretched at geosynchronous orbit during the optical auroral onset than during the precursor pseudobreakups. These pseudobreakups can be initiated by auroral streamers which originate at the most poleward set of arc systems and drift to the more equatorward main UV oval. Observations are presented of these AAFs in conjunction with low-and high-altitude particle and magnetic field data. These place the activations at the interface between dipolar and taillike field lines probably near the peak in the cross-tail current. These onsets are put in the context of a new scenario for substorm morphology which employs individual modules which operate independently or couple together. This allows particular substorm events to be more accurately desc...

During the later stages of the auroral substorm the luminosity distribution frequently resembles a double oval, one oval lying poleward of the normal or main UV auroral oval. We interpret the double oval morphology as being due to the plasma sheet boundary layer becoming active in the later stages of the substorm process. If the disturbance engulfs the nightside low‐latitude boundary layers, then the double oval configuration extends into the dayside ionospheric region. The main UV oval is associated with the inner portion of the central plasma sheet and can rapidly change its auroral character from being diffuse to discrete. This transition is associated with the substorm process and is fundamental to understanding the near‐Earth character of substorm onset. On the other hand, the poleward arc system in the nightside ionosphere occurs adjacent to or near the open‐closed field line boundary. This system activates at the end of the optical expansion phase and is a part of the recovery phase configuration in substorms where it occurs. These two source regions for nightside discrete auroral arcs are important in resolving the controversy concerning the mapping of arcs to the magnetosphere. The dayside extension of this double oval configuration is also investigated and shows particle signatures which differ considerably from those on the nightside giving clues to the magnetospheric source regions of the aurora in the two local time sectors. Near‐Earth substorm onsets are shown to be coupled to processes occurring much further tailward and indicate the importance of understanding the temporal development of features within the double oval. Using “variance images,” a new technique for the investigation of these dynamics is outlined.

The poleward arc system of a double oval distribution is shown to activate at the end of the optical expansion phase signifying the beginning of substorm recovery. The velocity dispersed ion signature (VDIS) can exist coincident with this discrete aurora developing on the most poleward oval. Although the VDlS is usually associated with ion beams in the plasma sheet boundary layer, it is demonstrated that the ionospheric signature is not beamlike but distributed in pitch angle. At the time when the double oval begins to form, the magnetic field in the magnetotail lobe becomes less flared and can show Pc 5 period oscillations. Similar pulsations also exist in the ionosphere associated with the most poleward oval and with stationary surge formation. Theoretical considerations link this phenomenon with a wave source tailward of xGSE = −30RE and fast mode evanescent waves propagating earthward in the tail lobe region. In this case the magnetotail appears to act like a waveguide and the plasma sheet boundary layer as a resonance region. This implies that the coupling of this fast mode wave is with the plasma sheet boundary layer and not with dipolar like field lines. The implications of this for the reconnection model of substorms are discussed.

The Tsyganenko long magnetospheric model (1987) has been used in conjunction with ultraviolet images taken by the Viking spacecraft to investigate the relationship of the auroral distribution to different magnetospheric regions. The model describes the large-scale structure of the magnetosphere reasonably well for dipole tilt angles near zero, but it appears to break down at higher tilt angles. Even so, a wide variety of auroral configurations can be accurately described by the model. It appears that the open-closed field line boundary is a poor indicator of auroral arc systems with the possible exception of high-latitude polar arcs. The auroral distribution typically called the "oval" maps to a region in the equatorial plane quite close to the Earth and can be approximately located by mapping the model current density maximum from the equatorial plane into the ionosphere. Although the model may break down along the flanks of the magnetotail, the large-scale auroral distribution generally reflects variations in the near-Earth region and can be modeled quite effectively. pause (D. Stern, private communication, 1989) but rather it uses observations of field vectors close to the magnetopause, which as will be seen still results in a reasonable approximation. Before such an empirical model is used extensively, however, it is necessary to ensure that results from it appea• reasonable in a global context. Thus questions such as "ar( the ionospheric projections of boundaries in general agreement with actual ionospheric data?" need to be investigated before the model can be trusted to provide a basis ol comparison between various magnetospheric theories which require precise knowledge of certain regions of the magnetosphere. The next section of this paper deals with the large-scale structure of the magnetosphere derived from the model. Then using images of the northern hemisphere auroral distribution taken in 1986 by the UV imager on board the Viking spacecraft in combination with the Tsyganenko [1987] and International Geomagnetic Reference Field (IGRF) 1985 magnetic field models, we relate the auroral distribution observed under a variety of geophysical conditions to corresponding regions in the magnetosphere and demonstrate that the model is appropriate. METHOD OF MAPPING AND DEFINITION OF REGIONS In order to make comparisons between the auroral images and the model calculations, the auroral data were first transformed from the view as seen from the satellite onto a plane in IGRF 1985 eccentric dipole coordinates (at an altitude of 120 km) [Wallis et al., 1982]. Each transformed image covered 60 ø magnetic latitude with 12 MLT at the top center of the image and with dusk (18 MLT) to the center left. This coordinate system was selected because it orders magnetic field data reasonably well without distorting the spatial relations implied by latitude and local time changes. A rectangular grid of points spaced at about 33km intervals based on this image format was used to trace field lines into the magnetosphere from the...

No abstract