C. Deehr and D. Lummerzheim

Geophysical Institute, University of Alaska Fairbanks

Abstract. A combination of all-sky imagers (ASTV) and meridian-scanning photometers (MSP) was used to identify the optical signature of the growth phase, onset, expansion, and recovery of 33 auroral substorms in the Alaskan sector. The discrete auroral arc that brightens at auroral substorm onset was found to be poleward of the diffuse aurora that contains the H emission by a distance of between 10 and 300 km. The average onset time was 2215 magnetic local time (MLT) and the average geographic latitude of the onset arc and the H arc was 64.6° and 63.6°, respectively. The poleward crossover of the peak H emission occurs shortly after substorm onset, for substorm intensification after 2100 MLT. The peak H emission crosses back to the equatorward position during substorm recovery between 2100 and 0100 MLT. After 0100 MLT the peak H emission remains poleward of the electron-trapping boundary for the rest of the night. During the growth phase the peak H emission moves equatorward more quickly than does the onset arc and monotonically doubles in total intensity during the equatorward motion, in a manner quite unrelated to the fluctuations in brightness of the onset arc. In the onset arc a reduction of 427.8-nm N₂⁺ and 557.7-nm [OI] emissions just prior to onset has been reported earlier and dubbed auroral fading. We find that the 630.0-nm [OI] emission rises considerably relative to the others, and we conclude that the fading observed in the few minutes prior to onset is the ionospheric signature of a pulse of > 10¹⁰ electrons cm² s-eV^-1 with energies less than 100 eV. Thus the observations reported here do not support the argument that the substorm onset begins from within the proton precipitation, but the discovery of the soft electron pulse during auroral fading” just prior to onset does support those substorm onset theories requiring a pulse of Alfven speed electrons poleward of the electron trapping boundary and in the region of upward field-aligned current.

1. Introduction

1.1. Ground-Based Observations of Magnetospheric Processes

In the years since the definitive description of the auroral substorm, the emphasis shifted from ground-based observation, which formed the basis of the original work [Akasofu, 1964], to space-based images and field and particle measurements. It has become apparent, however, that the valuable overview provided by the satellite data lacks some of the temporal and spatial resolution available from the ground. Thus any complete description of the vast scale of the auroral process, from the ionosphere to the solar wind, will require the coordinated use of orbital, suborbital, and ground-based observations.

A renewed examination of ground-based optical observations is presented here, with particular emphasis on aeronomical studies of hydrogen emissions, using two-dimensional detectors, data-handling techniques, and observational schedules developed over the years as a part of the National Science Foundation (NSF) Coupling, Energetics, and Dynamics of Atmospheric Regions (CEDAR) initiative. The morphology of the hydrogen emission, which is the locus of greatest proton energy flux, will be compared with that of the electron-induced aurora in the context of the auroral substorm.

1.2. Hydrogen Emissions as a Measure of Proton Precipitation

Historically, the discovery [Vegard, 1939] and subsequent measurement of Doppler shift [Meinel, 1950] of the hydrogen Balmer emissions in the aurora were the first evidence that the aurora was produced by precipitating energetic particles. The "hydrogen aurora" is diffuse because the incoming protons undergo charge-exchange collisions and thereby spend half the time as neutral hydrogen that crosses the magnetic field lines. Davidson [1965] modeled the spread of the H emission in the atmosphere due to this process using a Monte Carlo trace of a thin sheet of precipitating protons. The resulting hydrogen emission was concentrated in the 110-km altitude region and spread more than 50 km poleward and equatorward of the original point of entry [see also Johnstone, 1972; Synnes et al., 1998]. Because of the resulting diffuse nature of the H emissions, their use as an ionospheric signature of proton precipitation has been slow to evolve, compared to the more extensive documentation of the bright, discrete auroral forms associated with keV electron precipitation [Vallance Jones, 1974].

Because of instrumental limitations, both the photoelectric and the spectrographic descriptions of the hydrogen morphology were restricted in temporal and spatial resolution. There was agreement, however, that the locus of brightest H emissions, corresponding to greatest proton energy flux, was equatorward of the discrete electron aurora in the evening and poleward in the postmidnight period [Wiens and Vallance Jones, 1969]. The crossover occurred in the N-S oriented electron arcs that develop after the passage of the westward traveling surge [Montbriand, 1971; Fukunishi, 1975]. The evening arc was observed sporadically to consist almost entirely of proton-induced emissions and was thereby referred to as the “hydrogen arc.

With the development of better interference filters and detector sensitivity, meridian-scanning photometers (MSP) indicated that H emissions changed brightness and position at least as quickly as the electron induced emissions during the course of the substorm. This tendency led Vallance Jones et al. [1982] to order the data by aligning the onset times, in an attempt to extract a consistent pattern in the relationship between the H emission and the more well known electron-induced aurora. There were still inconsistencies that were evident between the evening and the morning substorms. It was also evident from the Vallance Jones et al. data that the initial brightening occurred as Akasofu [1964] had defined it, in the most equatorward, discrete electron arc and that this arc was poleward of the hydrogen arc. The former study was not so clear on the fate of the protons after midnight, but it generally agreed that the H emission remained poleward of the morning aurora. Oguti [1973] reported that the mesoscale electron and proton aurora were not co-located, and he suggested that these regions may be associated with the upward and downward field-aligned currents (Region 1 and Region 2, respectively).

In a fortunate pass of the satellite ESRO-1A over the Tromso all-sky camera, Deehr et al. [1971] and Deehr and Egeland [1972] interpreted particle and photometer observations to show that the onset of the substorm shown in the Tromso all-sky camera occurred poleward of the hydrogen arc and the stable trapping boundary. A similar latitudinal distribution was measured by Romick and Sharp [1967] in a coordinated ground-based and satellite experiment over Alaska.

In the meantime, low-altitude, polar-orbiting satellite particle measurements confirmed the general morphology and showed that the number flux of precipitating electrons is ~100 times that of protons but that the energy flux is less than 10 times as great. In fact, there are reported some occasions when the proton energy flux [1985] significantly exceeded that of the electrons (D. Evans, private communication, 1985). Proton and electron precipitation is observed over the entire zone, but areas of greatest energy flux (corresponding to strongest optical emissions) are only rarely colocated.

The purpose of this study is to confirm the previous observations of the relationship of proton and electron aurora at a somewhat better resolution in time and space. In addition, we examine this relationship in the critical period just prior to substorm onset.

2. Instrumentation

The instruments used in this study are based at the Poker Flat Research Range (PFRR) and are operated mainly in support of sounding rocket campaigns during the dark of the moon in the months of December through March. Thirty-three substorms were chosen for this study from the data in the 1990-1993 seasons. Magnetic local time (MLT) at PFRR is ~11 hours behind UT and 2 hours behind Alaska standard time.

The MSP is a set of spatially scanning, 3-inch diameter telescopes with optical interference filters at the auroral wavelengths (558-nm [OI], 428-nm (N₂⁺), 630-nm [OI], 486-nm (H_b), and 732-nm [O II]) which tilt on and off the wavelengths of interest so as to allow the removal of any other sources of light from the signal. The filters are from 0.3 to 0.7 nm wide; the field of view is 1°; one meridian scan takes 2-s, (0°-180°), and the electronic integration time is usually 1/5 to 1/10 of the time it takes to observe 1° (10 m).

There is a 2-s interval between sky scans, and two sky scans of on and off wavelength positions are averaged, so each recorded data scan is 16-s long. The N-S magnetic meridian scans are separated according to wavelength and plotted as gray scales of intensity as a function of time with north at the top and time increasing to the right (Plate 1).

The all-sky imager (ASTV) is an unfiltered Westinghouse silicon-intensified target (SIT) with a 40-mm photocathode used with a Nikon "Fisheye" 8-mm focal length, f/2.8 lens. The 30 frames per second analog signal is recorded on 1/2 inch VHS cassette magnetic tape, which is sometimes run in time-lapsed mode. The pictures are oriented so that magnetic south is at the top and east is to the right (Figure 1).

3. Observations

There is sufficient darkness to run the ASTV and MSP at Poker Flat Research Range during the months of September through April. The optical instrumentation at PFRR is normally run in support of rocket campaigns and more rarely during ground-based programs during the dark of the moon in the period December through March. The MSP data from rocket campaigns during the years 1990 through 1993 were the best available data at that time. These were scanned for examples of auroral intensifications. This survey led to a list of 33 records of substorm onsets, both from prior activity and from little or no activity. These examples are listed in Table 1.

The example of the MSP and ASTV data shown in Plate 1 and Figure 1 was chosen because the substorm onset took place in the field of view of the instruments, and it contains all of the elements of a typical, isolated substorm. This onset occurred near the local zenith at ~1145 UT (0030 MLT) on February 1, 1990, after a period of little activity indicated by

the poleward position of the precipitation zone which moved equatorward starting at ~1115 UT. It is important to note that the substorm onset occurred in the PFRR meridian in this case (no brightening to the east or west) and that the increase in intensity of the poleward arc was accompanied by the development of rays, folds, and a secondary band prior to poleward motion. The 557.7-nm [OI] and 427.8-nm (N₂⁺) emissions are present also in the H arc, owing to the excitation by secondary electrons created in the process of the proton interaction with the atmosphere [Rees, 1982; Strickland et al., 1993; Lummerzheim et al., this issue].

The plan of this work, then, is to illustrate the morphology of the substorm using the event shown in Plate 1 and Figure 1. We will then examine the other 32 substorms to establish the statistical occurrence frequency of the substorm characteristics of interest.

4. Growth Phase

4.1 Latitudinal Morphology

There are generally three precipitation regions separable in latitude during this period. Consider the aurora shown in Plate 1 and Figure 1:

1. The first region consists of poleward arcs and bands characterized by higher values of the 630-nm/558-nm intensity ratio. At 1140 UT, the 630-nm/558-nm emission ratio of 0.2 with 200 R of 428-nm emission indicates incoming electrons with a peak energy (E_p) of 0.3 keV [Rees and Luckey, 1974].

2. The second region consists of an arc at the trapping boundary with a lower 630-nm/558-nm intensity ratio, indicating higher electron E_p. At 1140 UT the 630-nm/558-nm emission ratio of 0.04 with 1200 R of 428-nm emission indicates 10-keV electron precipitation.

3. The third region consists of the “hydrogen arc,” a diffuse arc, extending over 100-km of latitude on the equatorward side. At 1140 UT the 486-nm/428-nm emission ratio was near 0.5, indicating ~3 mW m² of 5-keV protons [Rees, 1982].

The energetic arc at the trapping boundary is not usually evident in the optical data. This latitudinal morphology seems to be present in most of the satellite passes, however, and marks the poleward edge of the diffuse aurora [Romick and Sharp, 1967; Deehr et al., 1973, 1976].

4.2. Equatorward and Westward Motions

A typical optical ionospheric signature of the growth phase is shown in the MSP elevation angle and intensity versus time plot in Plate 1. At 1115 UT the precipitation region around 35^o above the north horizon begins to move equatorward at a rate of ~100 m s^-1. This motion usually coincides with a similar equatorward motion of the dayside aurora, and it is initiated by a southward turning of the interplanetary magnetic field (IMF) B_Z [cf. Sandholt et al., 1986].

Typically, the arcs rise and fall in brightness with periods between 150 and 600 s and there is simultaneous westward motion of luminosity and structure at ~1.5 km s^-1 [Xu et al., 1993]. In the example in Figure 1, the soft electron arc referred to as poleward arcs and bands above developed from east to west across the sky between 1135 and 1140 UT.

For the 33 storms in this study the time elapsed between the beginning of the growth phase and the onset was typically 30–45 min. During this time the H arc moved equatorward but in a manner different from that of the electron arcs. The H emission due to a thin sheet of incoming protons is normally ~100 km across in latitude [Davidson 1965], but about midway through the growth phase, it appeared to broaden in latitude and move more quickly equatorward than the discrete electron-induced arcs.

The H arc is shown in Plate 1 as 486.1-nm H_b emission in the February 1, 1990, event between 1100 and 1200 UT. The apparent, sudden equatorward motion of the H arc starting at ~1125 UT may be characterized as moving more quickly equatorward compared to the smooth equatorward trend of the electron emissions. This may be because of a latitudinal thickening of the proton sheet, possibly associated with the longitudinal passage of some structure, or a sudden movement of the source region.

The reality of the departure of the H emission from the electron arcs is better illustrated by a plot of the actual latitudinal variation with time of the emission regions shown in Figure 2. We measured the location of the arcs graphically

using an observer sheet featuring distance along a curved Earth versus altitude to scale [Chamberlain 1961, p. 108]. Angles from the observer to the maximum of intensity of the arcs were scaled to the assumed altitude of 110 km and referred to the geographic latitude of the geomagnetic conjugate. The latitudinal motion of the H emission region during growth phase expansion is obviously different from that of the electron arcs. It is difficult to record when it occurs out of the zenith, but it is identifiable in 10 of the 33 cases in Table 1.

4.3. Changes in Auroral Intensity Just Prior to Onset

The sequence of ASTV pictures of the initial brightening and poleward expansion on February 1, 1990, is shown in Figure 1. Of the two electron-induced arcs, the equatorward arc moved slowly equatorward and remained at a roughly constant intensity, while the poleward arc moved in from the east around 1135 UT, brightened and began the poleward expansion at ~1144:50 UT. The traditional nonlinear increase in intensity of the poleward arc at onset was preceded by the development of curls and folds. One of the curls intensified, and a second band appeared as a part of the poleward expansion.

The new observation to be reported here is not visible in the ASTV, but it is in the MSP data. An enlargement of Plate 1 is shown in Figure 3, which displays the intensity as a function of time in the four emissions for a portion of the sky near the magnetic zenith from 1130 to 1150 UT. The location of the electron-induced arc at the trapping boundary is marked with a dotted line. The onset arc is enclosed between two dashed lines. The magnetic zenith is shown as a horizontal line with its intersection of the equatorward arc marked by a diamond. The same pattern of lines is copied onto the H_b panel to show the relative location of electron and proton precipitation. Careful photometry of the peak emission of the poleward arc (between the dashed lines) was carried out from its appearance at 1137 UT to its intensification at onset near 1144 UT. The brightness of the electron-induced emissions and their ratios are shown as a function of time in Figure 4. Note that the intensity of 630-nm [OI] exceeds that of 557.7-nm [OI] from 1138 to 1143 UT, prior to the explosive increase of the 557.7-nm [OI] emission at 1144:50 UT.

The average energy of the incoming electrons producing the onset arc may be estimated from the 630-nm/427.8-nm and 630-nm/558-nm emission ratios [Rees and Luckey, 1974] shown in the second panel of Figure 4. The intensity ratios indicate that the average energy decreased from >1 keV to < 100 eV around 1138 UT. In addition, the rising intensity of the emissions indicates that the electron number flux rose at the same time to well above 10¹⁰ cm² s until the onset at 1145 UT. This rise is too slow to be due to a velocity dispersion in the energetic electrons prior to the rapid increase of 558-nm emission associated with onset. The magnitude of the increase is beyond any possible ionospheric heating. A similar increase in the 630-nm/558-nm emission ratio was noted in 16 of the 33 substorms in this study.

The change of intensity for the hydrogen emission is more difficult to document, owing to the diffuse nature of the hydrogen arc. The meridian-scanning photometer data for the H emission during the growth phase is shown in the top panel of Figure 5. The bottom panel of that Figure is a plot of the hydrogen emission integrated over the meridian between 0^o and 180^o and 20^o to 150^o elevation angle measured from south. This integration interval was chosen to exclude any brightness increase due to the increased path length near the horizon, but to include the entire hydrogen arc. It is apparent that the intensity of hydrogen emission over the meridian is constant, except for the period between 1115 and 1130 UT, when it increases monotonically to nearly twice its original intensity. This is also the period of the sudden equatorward movement of the hydrogen arc. Thus the change in H intensity occurs during the equatorward motion that appears faster than that of the electron-induced arc.

5. Onset

5.1 Relative Latitudinal Location of H and Electron Arcs at Onset

Special attention has been directed toward the morphology of hydrogen emission in the growth phase and onset [Samson et al., 1992a, 1992b]. This was prompted by results from the Canadian Auroral Network for the OPEN Program Unified Study (CANOPUS) chain of MSPs, which indicated that the substorm onset begins with the brightening of an electron arc from within the diffuse aurora and the main proton precipitation called the hydrogen arc. Samson et al. [1992a, 1992b] has since referred to the onset arc brightening from within the hydrogen emission but poleward of the peak proton precipitation. The importance of this point rests in the attempt by Samson et al. to establish the onset of the substorm on closed, dipolar, field lines.

The temporal and spatial resolution of the CANOPUS chain of MSPs has unfortunately been compromised to match that of the satellites it supports (>50 km, >1 min) and to ease communication of data. Thus the CANOPUS system can not resolve the question if the electron arc that brightens at onset is closer than 50 km from the hydrogen arc, the diffuse aurora, and the boundary for stable trapping.

The onset shown in Plate 1 and Figure 1 occurred near the PFRR zenith at ~1145 UT on February 1, 1990, after a period of little activity, indicated by the poleward position of the precipitation zone which moved equatorward at ~1115 UT. As was shown in section 4, the “hydrogen arc” advanced equatorward more quickly than the electron precipitation until it was well separated (75 km) from the onset arc at the time of poleward expansion. The relationship of the H arc to the onset arc may be seen in detail in the 486.1-nm H_b emission panel in Plate 1 and Figure 3.

The blue arrow in Plate 1 indicates the elevation angle where the 558-nm [OI] emission brightens just before the breakup. The bottom panel shows that the 630-nm [OI] emission brightens at a slightly higher elevation angle due to the higher altitude of the emission region. The black arrow in the H_b panel indicates the elevation angle where the peak hydrogen equatorward of the electron arc. Figure 3 is a magnification of this section from Plate 1. The dotted and solid lines mark the location of the brightening arc and diffuse aurora (trapping boundary), respectively. These locations are copied to the H_b panel and again clearly show the relative location of the hydrogen arc to the electron arcs.

Because the H emission due to a thin proton sheet is theoretically [Davidson, 1965] ~100 km broad in latitude, the onset arc can still be within the H emission and poleward of the incoming protons. In the 33 substorms chosen for this study the latitudinal extent of H emission is usually ~100 km, which supports the results of the Monte Carlo models and indicates that the typical hydrogen arc is produced by a thin sheet of protons.

Assuming an altitude of 110 km for the maximum intensity of the arcs (E_p=5 to 10 keV), we have measured the location of the arcs at onset. This was accomplished graphically for the 33 substorms using the same method described in section 4 for Figure 2. The resulting plot (Figure 6) shows that the electron-induced arc brightens and moves poleward at onset and is always poleward of the H arc at onset, sometimes by as much as 300 km. This result is in agreement with the observations of Vallance Jones et al. [1982].

5.2 Magnetic Local Time and Latitude of Onset

The Magnetic local time (MLT) of onset in Figure 6 does not seem to be grouped around magnetic midnight but ranges over several hours, mostly before magnetic midnight. This is in agreement with the satellite image data of Craven and Frank [1991] which ranged from 2000 to 0002 hours with an average of 2250 MLT.

Figure 7 shows the onset occurrence in each half-hour interval for the 33 events in this study. Note that the resulting occurrence frequency is sharply peaked near 2200 MLT; more than half of the onsets occurred within an hour of that time.

The distribution in latitude ranged between 60° and 70° magnetic latitude (see Table 1), but the average latitude of H emission at onset was 63.6°, and the average electron onset arc was at 64.6° magnetic latitude with a standard deviation of 1.24° and 1.28°, respectively.

6. Expansion and Recovery

6.1 Latitudinal Motion of the Peak H Emission in MLT

Vallance Jones et al. [1982] observed that H emissions were differently distributed in the morning and the evening. There appear to be three zones in MLT in which the latitudinal motion of the H arc is different during a substorm expansion and recovery. A typical example is illustrated in Plate 2 by an MSP trace from January 31, 1990, at Poker Flat Research Range that shows three major intensifications in the three different regions of MLT. The first intensification occurs at around 0800 UT (2050 MLT). Note that the H emission does not expand poleward. The second intensification is at 1000 UT (2250 MLT), and the peak in the H emission crosses over to the poleward side within a few minutes. The recovery of the latter substorm begins just before 1200 UT, when the electron aurora and the peak in the H emission return to the equatorward side of the auroral zone. This motion is the growth phase of a new substorm which begins near 1230 UT (0135 MLT). After the poleward expansion in this case, the peak in the H emission does not return to the equatorward side, and no new substorms are seen that night.

We examined the crossover of the peak H emission, both poleward and equatorward for the 33 substorms, and found that poleward expansion does not occur during intensification before ~2100 MLT. Both poleward and equatorward crossovers occur between 2100 and 0100 MLT, and few onsets occur after approximately 0100 MLT, when the H emission is invariably poleward of the electron arcs.

6.2 Spatial and Temporal Relationship of the Peak H Emission and Electron Aurora

The expansion phase has been characterized by the stepwise poleward expansion of electron arcs moving progressively westward. In agreement with spectrographic studies [Wiens and Vallance Jones, 1969; Montbriand, 1971; Oguti, 1973; Fukunishi, 1975], we find that the H emission does not participate in the westward traveling surge. It typically expands poleward after the development of N-S oriented electron arcs, a few minutes after the initial expansion, as shown in Figure 1.

Note in Figure 5 that the H emission signal is contaminated by some electron emission during the extremely bright poleward expansion. This is characterized by sharp negative signals when the contamination appears in the background filter and positive signals when it appears in the signal channel. Such contamination prevents the acquisition of detailed tracking of any H emission that may be associated with the poleward expanding arc with its red lower border and plethora of molecular emissions.

The increased time resolution afforded by the MSP over earlier studies shows that the H emission generally is found between the larger-scale regions of electron precipitation during this period. This was pointed out by Oguti [1973], Fukunishi [1975], and Viereck and Stenback-Nielsen [1985], who also found the H emission usually poleward of pulsating aurora after midnight but periodically venturing equatorward into the pulsations (area near 1430 UT marked in Figure 7). This invasion of the pulsating aurora was found by Viereck and Stenback-Nielsen to produce an increase in the time between pulsations which is proportional to the intensity of the H emission.

7. Summary

7.1 The Growth Phase

The H emission is in the diffuse aurora, which can have a significant electron-induced component (E_p >10 keV), sometimes appearing as thin filaments in the zenith and sometimes as a homogenous arc at the boundary of stable trapping on the poleward side of the diffuse region. Electron arcs and bands (E_p <5 keV) appear on the poleward side of the trapping boundary, and one of these (usually the most equatorward one) brightens at onset. Luminosity and structure pulsate and move westward at typical convection velocities in all of these forms. The latitudinal dispersion of the H emission (~100 km) is consistent with a thin proton sheet, but it broadens or moves equatorward more quickly that the electron arcs, indicating a possible expansion of the proton beam in latitude, owing to longitudinal motion or change in the source geometry. The 630-nm emission in the onset arc rises relative to other emissions during the few minutes just prior to the brightening at onset in a significant number of cases. This was shown to indicate a pulse of increased electron number flux of average energy <100 eV.

7.2 The Onset

The onset arc is located 10–300 km poleward of the peak proton precipitation inferred from the H emission in all 33 of the onsets in this study. The occurrence frequency of onset peaks sharply near 2300 MLT, and the average magnetic latitude of onset is 63.6º for protons and 64.6º for electrons.

7.3 The Expansive Phase

The electron arcs expand poleward with the development of the westward traveling surge. The H emission expands poleward after the formation of the N-S oriented arcs in those intensifications that occur after 2100 MLT and before 0100 MLT. If the crossover of the H emission is an indicator of the development of a current wedge and/or the convection reversal, then these events are limited in longitude from 2100 to 0100 MLT, with a significant peak in occurrence frequency just after 2200 MLT.

7.4 The Recovery Phase

Before 0100 MLT the H emission can return to the equatorward edge of the auroral zone, and the substorm sequence begins again. After 0100 MLT the H emission remains in the dark region poleward of the pulsating aurora and the diffuse electron aurora that is associated with the boundary of stable trapping. Brief incursions into the latter regions occur, but no new onsets take place.

8. Discussion and Conclusions

The electron trapping boundary, which defines the poleward edge of dipolar field lines, is at the poleward edge of the H emission region prior to the crossover of the H emission near midnight, and it is at the poleward edge of the discrete aurora in the post-midnight region. This is in agreement with observations from satellites and the ground for many years [Deehr et al., 1973, 1976; Romick and Sharp, 1967].

The discrete auroral arc that brightens at auroral substorm onset was found to be poleward of the diffuse aurora that contains the H emission by a distance between 10 and 300 km for the 33 intensifications studied here. The average onset time was just after 2200 MLT, and the average geographic latitude of the onset arc and the H arc was 64.6º and 63.6º, respectively. We conclude that the auroral substorm originates poleward of dipolar field lines and in the region of upward field-aligned current.

The poleward crossover of H emission occurs shortly after substorm onset. The H emission crosses back to the equatorward position during substorm recovery between 2100 and 0100 MLT. After 0100 MLT, H emission remains poleward of the electron trapping boundary for the rest of the night.

The relatively high resolution in time and space of the measurements reported here also led to the recognition of a drop in general auroral intensity during the few minutes prior to onset. This effect was mentioned originally by Snyder and Akasofu [1972] and Mende and Eather [1976] but was best described by Pellinen and Heikkila [1978] as a drop in brightness seen in all-sky camera images in the few minutes prior to onset. Their meridian-scanning photometer data in 427.8-nm N₂⁺ (1 N) emission confirmed the observation, but the 486.1-nm H_b channel was contaminated by electron emissions. Baumjohann et al. [1981] found that the auroral fading was accompanied by magnetic field change indicating

a drop in equivalent current, while the electric fields remained roughly constant. Morse and Romick [1982] used meridian scanning photometer data to describe the fading of the 427.8-nm N₂⁺ (1 N) and 557.7-nm [OI] emissions as part of a series of fluctuations prior to onset.

We have extended these earlier studies to include the 630-nm [OI] emission as a ratio with the other emissions, in order to document the changes in the incoming electron characteristic energy. This has led to the recognition of a pulse of increased electron number flux, with an average energy <100 eV, to occur during the auroral fading period in the few minutes just prior to substorm onset in the onset arc. This change corresponds to the sudden equatorward motion of the H arc in a significant number of cases, and it is a previously unobserved phenomenon associated with substorm onset.

The separation and very different temporal behavior of the onset arc and the peak proton precipitation reported here do not support theories of substorm onset requiring the onset arc to occur on closed, dipolar field lines inside the region of greatest proton energy flux as put forward by Samson et al. [1992a, 1992b] and Lyons and Samson [1992]. On the other hand, the discovery that the auroral fading just prior to onset is the signature of a significant pulse of low-energy electrons supports substorm onset theories requiring an Alfven wave or shock wave prior to onset.

Acknowledgments. The authors are indebted to C. Moore, D. Osborne, and T. Hallinan for the instrument operation at Poker Flat Research Range. The authors acknowledge the contributions from a number of colleagues and students to the work, most notably S. Akasofu, T. Hallinan, J. Kan, H. Nielsen, R. Smith, and B. Xu. This work was funded in part by grant ATM 92-24679 from the Atmospheric Sciences Division of the National Science Foundation and NASA contract NAG5-693.

Janet G. Luhmann thanks Gerald J. Romick and Fokke Creutzberg for their assistance in evaluating this paper.

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

Akasofu, S.-I., and A.L. Snyder, Comments on the growth phase of magnetospheric substorms, J. Geophys. Res., 77, 6275, 1972.

Baumjohann, W., R.T. Pellinen, H.T. Opgenoorth, and E. Nielsen, Joint two-dimensional observations of ground magnetic and ionospheric electric fields associated with auroral zone currents: Current systems associated with local auroral break-ups, Plant. Space Sci., 29, 431, 1981.

Chamberlain, J.W., Physics of the Aurora and Airglow, Academic, San Diego, Calif., 1961.

Craven, J., and L.A. Frank, Large-scale evolution of auroral dynamics, in Auroral Physics, edited by C.-I. Meng, M.J. Rycroft, and L.A. Frank, 273, Cambridge Univ. Press, New York, 1991.

Davidson, G.T., Expected spatial distribution of low-energy protons precipitated in the auroral zones, J. Geophys. Res., 70, 1061, 1965.

Deehr, C.S., and A. Egeland, Auroral morphology, Ann. Geophys., 28, 415, 1972.

Deehr, C.S., A. Egeland, and F. Søraas, Auroral morphology, in The Radiating Atmosphere, edited by B. McCormac, p. 125, D. Riedel, Norwell, Mass., 1971.

Deehr, C.S., et al., Particle and auroral observations from the ESRO 1/AURORAE satellite, J. Atmos. Terr. Phys., 35, 1979, 1973.

Deehr, C.S., J.D. Winningham, F. Yashuara, and S.-I. Akasofu, Simultaneous observations of discrete and diffuse auroras by the Isis 2 satellite and airborne instruments, J. Geophys. Res., 81, 5527, 1976.

Fukunishi, H., Dynamic relationship between proton and electron auroral substorms, J. Geophys. Res., 80, 553, 1975.

Johnstone, A.D., The spreading of a proton beam by the atmosphere, Planet. Space Sci., 20, 292, 1972.

Lummerzheim, D., M. Galand, J. Semeter, M.J. Mendillo, M.H. Rees, and F.J. Rich, Emission of O I (630-nm) in proton aurora, J. Geophys. Res., this issue.

Lyons, L., and J.C. Samson, Formation of the stable auroral arc that intensifies at substorm onset, Geophys. Res. Lett., 19, 2171, 1992.

Meinel, A.B., Evidence for the entry into the atmosphere of high-speed protons during aurora activity, Science, 112, 590, 1950.

Mende, S.B., and R.H. Eather, Monochromatic All-Sky Observations and Auroral Precipitation Patterns, J. Geophys. Res., 81, 3771, 1976.

Montbriand, L.E., The proton aurora and auroral substorm, in The Radiating Atmosphere, edited by B. McCormac, p. 366, D. Riedel, Norwell, Mass., 1971.

Morse, T.H., and G.J. Romick, The fluctuation and fading of auroral arcs preceding auroral substorm onsets, Geophys. Res. Lett., 9, 1065, 1982.

Oguti, T., Hydrogen emission and electron aurora at the onset of the auroral breakup, J. Geophys. Res., 78, 7543, 1973.

Pellinen, R.J., and W.J. Heikkila, Observations of auroral fading before breakup, J. Geophys. Res., 83, 4207, 1978.

Rees, M.H., On the interaction of auroral protons with the earth's atmosphere, Planet. Space Sci., 30, 463, 1982.

Rees, M.H., and D. Luckey, Auroral electron energy derived from ratio of spectroscopic emissions, 1, Model computations, J. Geophys. Res., 79, 5181, 1974.

Romick, G.J., and R.D. Sharp, Simultaneous measurements of an incident hydrogen flux and the resulting hydrogen Balmer alpha emission in an auroral hydrogen arc, J. Geophys. Res., 72, 4791, 1967.

Samson, J.C., L.R. Lyons, P.T. Newell, F. Creutzberg, and B. Xu, Proton aurora and substorm intensifications, Geophys. Res. Lett. 19, 2167, 1992a.

Samson, J.C., D.D. Wallis, T.J. Hughes, F. Creutzberg, J.M. Ruohoniemi, and R.A. Greenwald, Substorm intensifications and field line resonances in the nightside magnetosphere, J. Geophys. Res., 97, 8495, 1992b.

Sandholt, P.E., C.S. Deehr, A. Egeland, G.J. Romick, B. Lybekk, R. Viereck, and G.J. Romick, Signature in the dayside aurora of plasma transfer from the magnetosheath, J. Geophys Res., 91, 10,063, 1986.

Snyder, A.L., and S.-I. Akasofu, Observations of the auroral oval by the Alaskan meridian chain of stations, J. Geophys. Res, 77, 3419, 1972.

Strickland, D.J., R.E. Daniell Jr., J.R. Jasperse, and B. Basu, Transport-theoretic model for the electron-proton-hydrogen atom aurora, 2, Model results, J. Geophys. Res., 98, 21,533, 1993.

Synnes, S.A., F. Søraas, and J.P. Hansen, Monte-Carlo simulation of proton aurora, J. Atmos. Sol. Terr. Phys., 60, 1695, 1998.

Vallance Jones, A., Aurora, Geophys. Astrophys. Monogr., vol. 9, D. Reidel, Norwell, Mass., 1974.

Vallance Jones, A., F. Creutzberg, R.L. Gattinger, and F.R. Harris, Auroral studies with a chain of meridian-scanning photometers, 1, Observations of proton and electron aurora in magnetospheric substorms, J. Geophys. Res., 87, 4489, 1982.

Vegard, L., Hydrogen showers in the auroral region, Nature, 144, 1089, 1939.

Viereck, R.A., and H.C. Stenbaek-Nielsen, Pulsating aurora and hydrogen emissions, J. Geophys. Res., 90, 11,035, 1985.

Wiens, R.H., and A. Vallance Jones, Studies of auroral hydrogen emissions in West Central Canada, III, Proton and electron auroral ovals, Can. J. Phys., 47, 1493, 1969.

Xu, B.-L., J.C. Samson, W.W. Liu, F. Creutzberg, and T.J. Hughes, Observations of optical aurora modulated by resonant Alfven waves, J. Geophys. Res., 98, 11,531, 1993.

__________

C. Deehr and D. Lummerzheim, Geophysical Institute, University of Alaska, 903 Koyukuk Ave. N, P.O. Box 757320, Fairbanks, AK 99775-7320. (cdeehr@gi.alaska.edu; lumm@gi.alaska.edu)

(Received February 20, 2000; revised May 8, 2000; accepted June 5, 2000.)

Copyright 2001 by the American Geophysical Union.

Paper number 2000JA002010.