M. Brittnacher 2, G. Parks 2, G. D. Reeves 3, R. R. Anderson 4, and K. Yumoto s

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL 104, NO A10, PAGES 22,807-22,817, OCTOBER 1, 1999 On relative timing in substorm onset signatures K Liou, C-I Meng, T Y Lui 1, P T Newell M Brittnacher 2, G Parks 2, G D Reeves 3, R R Anderson 4, and K Yumoto s Abstract We investigate the timing of popular substorm onset signatures to understand their temporal relationship Proxies for substorm onsets include auroral breakups, high-latitude magnetic bays, low-latitude Pi2 bursts, dispersionless injections at geostationary orbits, and auroral kilometric radiation We use the auroral breakup, identified with Polar UVI images, as a common reference time frame to calibrate the others Results, illustrated by two well-defined auroral substorms, unambiguously indicate that none of the four frequently used substorm onset proxies can provide a consistent timing of substorm onset This inconsistency in substorm onset timing is attributed as a consequence of emporal and spatial limitations on each observational technique A delay between the proxy identifiers and the auroral breakup is found to be typical It is therefore strongly suggested from this study result that a common reference time frame for substorm onset is necessary, and we propose it should be auroral breakups We argue that there is a need for an intercalibration of magnetospheric substorm phenomenology by using a unified definition of the substorm onset 1 Introduction The initiation of an explosive substorm expansion process, or more simply "onset," is one of the most widely studied topics in the field of magnetospheric physics From observational considerationsubstorm onsets can be determined by many substorm associated features For instance, auroral breakups have been widely used as the onset of auroral substorms [Akasofu, 1964] The sharp decrease in magnetic negative bays at high latitudes are used for polar magnetic substorms [Akasofu and Meng, 1969; Meng and Akasofu, 1969] Transient Pi2 magnetic pulsation ( s), particularly at low and middle latitudes, is frequently used as an onset indicator for magnetic substorms [Rostoker, 1968; Saito et al, 1976; Nagai et al, 1998] The sudden increase of electron and/or proton fluxes at energies from tens of kev to hundreds of kev, known as "dispersionless plasma injections," near the inner edge of Applied Physics Laboratory, Johns Hopkins University, Laurel, Maryland 2Geophysics Program, University of Washington, Seattle 3Los Alamos National Laboratory, Los Alamos, New Mexico 4Department of Physics and Astronomy, University of Iowa, Iowa City 5Department of Earth and Planetary Sciences, Faculty of Science, Kyushu University, Hakozald, Fukuoka, Japan Copyright 1999 by the American Geophysical Union Paper number 1999JA / 99 / 1999J A $ 0900 plasma sheet is also a popular onset identifier [Saito et al, 1976; Belian et al; Reeves et al, 1997] The intensification of auroral kilometric radiation (AKR), radio waves in the frequency range from about 30 to 800 khz, has recently become an additional tool for timing the onset [Slavin et al, 1993; Murata et al, 1995] There is no doubt that multiple measurement tech- niques are advantageous in understanding substorm processes Understanding the relative timing of different substorm signatures occurring at different locations and times can lead us to a coherent global picture of the substorm process Furthermore, a reliable timing of substorm onset can provide a common time flame for substorm observations in various parts of magnetosphere and model interpretations For example, recent statistical studies, based on Pi2 pulsations, on the evolution of substorm features in the magnetotail have established the notion that the earthward flow was enhanced at least 20 min prior to the onsets and the tailward flow associated with plasmoids started about 2 min before the Pi2 onsets [Machida et al, 1998] and magnetic reconnection takes place 2 min earlier than the dipolarization [Miyashita et al, 1998] However, Lui et ai [1998] examined plasma flow relative to auroral breakups and arrived at conclusions contrary to these two studies In particular, Liou et al [1998] found a I - 3 min typical delay in Pi2 onset compared to auroral breakups Therefore the use of Pi2 signals to time onsets and implications from this type of studies need to be thoroughly checked with auroral breakups The main difficulty in determining substorm onsets probably stems from their transient and localized na- 22,807

2 22,808 LIOU ET AL: BRIEF REPORT ture [eg, Akasofu and Meng, 1969; Ohtani et al, 1991; Angelopoulos et al, 1994; Lui et al, 1998] which cannot be easily detected with sparse observatories either on the ground or in space For example, ground-based all-sky cameras have been used to study auroral morphology for more than 40 years, and many of them are now operated automatically and can provide auroral images with temporal resolutions in the order of tens of seconds However, ground-based all-sky cameras can sometimes misidentify intensification of a substorm and the movement of a disturbance into the field of view as a newly occurring substorm [eg, Rostoker et al, 1980] Ground-based magnetometers have been continuously monitoring the Earth's magnetic fields for more than a century, and they are all over the globe providing local magnetic fields at different latitudes and longitudes with a time resolution of I rain or better A global chain of high-latitude magnetometer stations particularly those in the auroral zone have been widely used for study magnetic substorms for some time The sharp decrease in the H component (magnetic north) of the magnetic field in the auroral zone is believed to be caused by the extrusion of an enhanced westward electrojet across the midnight sector; therefore this technique strongly relies on the knowledge of the location of magnetometer station with respect to the location of westward electrojet [eg, Akasofu and Meng, 1969; Meng and Akasofu, 1969; Rostoker, 1968] Pi2 pulsations at low and/or middle latitudes are relatively easy to access when compared to many other data since they can be easily derived from magnetograms Many low- and mid-latitude magnetometers can provide I s resolution magnetic field data which theoretically can provide very accurate timing for substorm studies [eg, Machida et al, 1998; Miyashita et al, 1998] In reality a few tens of seconds is probably the norm because Pi2s are long period ( s) waves However, Pi2 pulsations alone cannot be used to identify a substorm onset definitively because one cannot tell whether a Pi2 signal is in association with a substorm onset or with an intensification of the auroral surge during an expansion phase [eg, Rostoker, 1968], and one may find difficulties in determining the start of Pi2 signals because some Pi2s start at small amplitudes and can grow over time Note that although all these shortcomings in timing the onset are pointed toward ground-based observations, in situ space observations that cannot simultaneously resolve the temporal and spatial issue will have the same drawback For instance, a dispersionless injec- tion can be timed with an accuracy of seconds (about 10 s for SOPA) which is certainly an advantage compared to say, Pi2s The typical local time extent of the injection is about 6 hours With four LANL satellites in orbit currently the probability that one of the LANL satellites is in the injection region at onset time is about 60-70% When there is no satellite in the injection region, it is still possible to get very good timing (+/- 30 s) of the injection (as opposed to onset) based on energy dispersion analysis However, the injections are typically delayed by a variable amount (0-10 min) compared to auroral onset Since we are interested in the sequence of events, the variable amount of the delay is the problem but not the delay itself In addition, the local time extent of the injection can be as little as 3 hours [Arnoldy and Moore, 1983] or even I hour reported by Reeves et al [1992]; therefore the chance of missing he onset of particle injections can be significant Intense auroral kilometric radiation is generated at 1 3 RE, geocentric, near the acceleration altitudes in the night and evening sectors [Gurnett and Frank, 1973; Green et al, 1979]; therefore it has the advantage of being an instantaneous remote measurement However precise timing of onsets can only be achieved by the right location of the spacecraft at the right time This is because AKRs propagate predominantly perpendicular to the magnetic fields Historically speaking, auroral breakups are the oldest and probably the most often used timing tool for substorm onsets Therefore we will use the auroral breakup as a common frame to calibrate other onset signatures by assuming that all the onset signatures have a common ultimate source Furthermore, auroral breakups will be determined by global imaging to avoid ambiguities in association with spatial and temporal limitations on the observation techniques Currently, there are two global auroral imagers, UVI and VIS, aboard the Polar spacecraft routinely taking northern hemispheric auroral images when the oval is viewable from Polar Global images by comparison give a more consistent indicator of substorm onset but the time resolution given by both UVI and VIS is comparatively poor (- 1 min on average) A bigger problem with global imaging is that the data set is more sparse Polar, like any other satellite, has problems with sometimes not being in the right place at right time However, for the purpose of this study these problems may not be as serious as it is thought because we can always find good events from millions of auroral images already taken by the two imagers In this study auroral images observed by UVI will be used to time substorm onsets We will use two well-defined auroral substorm events selected from many similar events to illustrate the need of a unified substorm reference time Some typical features that we found from many events such as inconsistency in proxy onset timing and time delay will be emphasized 2 Observations Data used in this paper consist of global auroral imagery acquired from the ultraviolet imager (UVI) on board the ISTP Polar satellite IT off et al, 1995], ground-based 1 s high-resolution magnetograms, used to monitor Pi2, from Kakioka Magnetic Observatory at low magnetic latitude, 1 min averaged magnetograms

3 LIOU ET AL: BRIEF REPORT 22,809 Table 1 Station coordinates used in the data analysis Location Code Latitude Longitude Magnet!c Latitude Kotel'nyy KTN 7594øN 13771øE 7000 ø Tixie TIK 7159øN 12878øE 6576 ø Chokurdakh CHD 7062øN 14789øE 6479 ø Zyryanka ZYK 6575øN 15078øE 5974 ø Kakioka KAK 2923oN 14018øE 2183 ø obtained from four 210 ø magnetic meridian (210 ø MM) network stations [Yumoto et al, 1996], 10 s energetic plasma data from synchronous orbit particle analyzer The geomagnetic field observations from the four (SOPA) on board the LANL spacecraft [Reeves high-latitude 210 ø MM stations also showed a typical et al, 1997], and AKR from Polar plasma wave inves- pattern (negative bays) of an isolated substorm in the tigation (PWI) [Gurnett et al, 1995] The locations of H component (magnetic north) of the magnetograms the ground magnetometer stations are listed in Table 1 shown in Plate 23 Note that the vertical bar in each We have selected two typical, well-defined substorm events from Polar UVI images The choice of these events is because the 210 ø MM chain stations, Kakioka, panel of Plate 2 marks the time of auroral breakup, and the width of the bar indicates the uncertainty in determining the auroral substorm onset either due to fiand the LANL satellite are all located near the nite integration time and/or missing frames Magnetic midnight region The two events and summary results are listed in Table 2 21 Case 1:3 April 1996, UT Plate 1 shows a sequence of false color Polar UVI auroral images at the two Lyman-Birge-Hopfield bands (LBHs and LBH ) for a substorm interval from 1521:32 to 1548:31 UT on April 3, 1996 Note that not all the images taken by UVI are plotted except the time interval around the onset time The frame rate for UVI is 37 s, and the time tag is the center of the integration period Images are mapped to the AACGM coordinates [Baker and Wing, 1989], and only the nightside part of the northern hemisphere from 60 ø to 90 ø magnetic latitudes is shown with midnight placed at the center right and dawn at the top on the left corner The locations of the four 210 ø MM stations are marked with their initials respectively in each image A sudden auroral brightenlug, determined by the start of the increase of area-integrated auroral luminosity (photon rate) (see Figure 1), at 1531:20 UT + 18 s taking place in the premidnight region at 2300 MLT and 66 ø Mlat followed by an explosive surge expansion signifies the auroral breakup or substorm onset Note that the auroral luminosity was integrated over a fixed area that is large enough to include the auroral bulge but small enough to exclude other auroral activities seen before the onset; thus the increase of the area-integrated luminosity simply reflects the increase of total auroral energy input rate According to the IMF observations from the Wind spacecraft which is 77 RE upstream (not shown), the IMF Bz changed sign to negative at UT after about 1 hour in the northward position field observations from Tixie showed an immediate but gradual decrease in the H component of magnetic fields at the onset time indicated by the yellow vertical bar, signifying a nearby passage of westward electrojet in the ionosphere This is consistent with the auroral images as one can see from Plate 1 that Tixie was located right on the eastward edge of the onset arcs Although Chokurdakh was located about 1 hour MLT east of the Tixie, its H component of magnetic field shows a similar feature but slower in increase This gradual decrease in the H component is often interpreted as a growth phase signature In this case, however, the start of the slow decrease corresponds to the substorm onset A sharp edge of the negative bay observed from Kotel'nyy significantly lags behind the auroral breakup This delay is not surprising because Kotel'nyy was located at around 2350 MLT or - 1 hour east and 3 ø- 4 ø north of the location of auroral breakup at the onset time (see Plate 1) As the substorm evolved, the substorm bulge expanded westward, poleward, and eastward and finally reached Kotel'nyy station after Table 2 Delay of the Proxy Substorm Onsets Relative to the Auroral Substorm Onset / /1436 Onset hhmm (MLT, Mlat) Magnetic Bay Kotel'nyy Tixie Chokurdakh Zyryanka Kakioka Pi 2 Particle injection electron proton LANL 084 foot prints AKR Polar (Rs, MLT, Mlat) 1531:20 (2300, 66 ø ) 1436:05 (2030, 66 ø ) 8 min (2350, 700 ø) 7 min (2250) 0 min (2334, 658 ø) 12 min (2233) 0 min (0035, 648 ø) / (2334) 7 min (0054, 597 ø)? (2353) 1 min (0032, 290 ø) I min (2332) No 3 rain 2 rain 1 rain (2000, 656 ø ) (2100, 651 ø ),--,0 rain No (80, 2125, 773 ø) (71, 1143, 693 ø)

4 ,,, - 22,810 LIOU ET AL- BRIEF REPORT LBHL O o D 0 0 O0 O0 0 0 O O 0 I ' '0 0 ' 15:20 5:24 5:28 5:32 1 $ 3 5:40 UT (hh:rnrn) Figure 1 Integrated photon flux over an area defined by 65 ø - 75 ø Mlat and MLT for a portion of substorm period between 1520 and 1540 UT on April 3, 1996 for Polar UVI The down-arrow indicates the time when the first auroral brightening in association with the breakup was identified 1539 UT There was little magnetic signature at the Zyryanka station This may be because the auroral bulge (electrojet) was too far ( 4 ø) from the station Plate 2b shows the H component magnetic fields observed from Kakioka There are several Pi2 bursts occurring during the one hour interval but only one corresponds to the substorm onset We associate the Pi2 burst which is closest in time to the breakup with substorm onset The onset time is then determined by the time at which the Pi2 pulse reaches maximum amplitude minus 1/4 of the Pi2 wave period This gives the Pi2 onset time as 1532:07 UT, about 1 min lag from the auroral breakup Interestingly, the start of the positive bay coincides with this Pi2 onset This may be because the Kakioka station was located at 0032 MLT near the onset meridian A brief and weak intensification of aurora took place in the premidnight region around 2200 MLT between 1510 and 1524 UT (partially shown) This brightening caused perturbed fields ( 20 nt) in the Kotel'nyy magnetogram and generated a weak Pi2 signal at Kakioka Although the Tixie station was located only 1 deg south of the aurora, it observed little response The geosynchronous energetic particle data are shown in Plates 2c and 2d for electrons and ions, respectively The energy channels plotted from top to bottom are E50-75 kev, E kev, E kev, E kev, and E kev for electrons and P kev, Pl kev, P kev, and P kev for protons A clear dispersionless particle injection (simultaneous increase in fluxes at different energies) can be seen in ion channels at 1534:22 UT :E 10 s The electron channels show a slow decrease in fluxes at all energy levels roughly starting at 1500 UT and a sharp decrease around 1532 UT, a few minutes before the sharp in- 0 crease in electron fluxes at 1534 UT There is no clear injection signature in electron channels, however, because the electron flux did not increase over its previous level The ionospheric footprint of the LANL satellite, calculated by T89 model [Tsyganenko, 1989], was 2000 MLT and 656 ø Mlat at 1531 UT, about 2 hour MLT from the westward edge of the first brightening arcs AKR is the highest-frequency plasma emission generated in the auroral zone and is closely related to the ac- celeration mechanisms of auroral particles [Green et al, 1979] Plate 2e shows Polar PWI plasma wave observations in the frequency range of khz In agreement with the magnetic field and optical observations, little AKR can be seen until the onset time around 1531 UT when enhanced wave emissions started The initial wave frequency was in the khz range After the onset, the plasma waves enhanced and the wave frequency extended toward lower frequencies around 100 khz in 4 min Since AKRs are generated near the local electron cyclotron frequency, this decrease in the wave frequencies indicates an upward moving source region Note that a weak AKR signal occurred around 1520 UT, which is consistent with the auroral image result 22 Case 2:3 May 1997, UT Plate 3 shows a sequence of false color Polar UVI auroral images at both LBHs and LBH1 bands for a substorm interval from 1427:20 to 1456:09 UT on 3 May 1997 By calculating the area-integrated photon flux (see Figure 2), we identify 1435:55 UT :E 18 s as the breakup time From Plate 3 the breakup is characterized by a sudden brightening of spatially localized arcs in the oval, taking place around 2030 MLT and with 1 i LBHL LBHS oo 0, 14:23 o DO O0 DO O0 DO O0 DO i, I I,, i I,, m I J, i i Ii,, i 14:27 14:,31 14:,35 14:39 14:4,3 UT (hh:rnnl) Figure 2 Integrated photon flux over an area defined by 65 ø - 75 ø Mlat and MLT for a portion of substorm period between 1423 UT and 1443 UT on May 3, 1997 for Polar UVI The down-arrow indicates the time when the first auroral brightening in association with the breakup was identified o oo

5 LIOU ET AL' BRIEF REPORT 22, :32 UT 1523:59 UT 1527:40 UT 1528:26 UT 1528:53 UT 1530:54 UT LBH' LBH LBHI _ :R L; s LBHs 1531:20 UT 1532:07 UT 1532:34 UT 1533:48 UT 1536:15 UT 1537:28 UT q LBH LBH! LBHI _ :Hs _ BHs _ LBHI 1538:42 UT 1541:09 UT 1542:23 UT 1543:36 UT 1546:04 UT 1548:31 UT / '"", f ß / LBHs LBHs~_ :HI LBHs -- LBHs Photons-cm-2-s -1 Polar Ultraviolet Image Day / o Plate 1 A sequence of nightside LBH auror d images from Polar ultraviolet imager showing auroral breakup at 1531:20 UT on April 3, 1996 Four 210 ø MM stations are marked in each image

6 22,812 LIOU ET AL- BRIEF REPORT (a) 210 MM /1531 UT -50 KTN (6994, 102) TIX (6567,19688) CHD (6467,21212) ZYK (5962,21672) (b) Kakioka H-mngnetometer! ß ' (d) LANL 1994 P75keV-400keV loooo loo lo ' ' ''--' '! ' ' ' 1 s 15:00 15:10 15:20 15:30 15:40 15:50 16:00 UT (H ) Plate 2 Substorm event on April 3, 1996, commencing at UT The top panel displays the x component (magnetic north) of magnetograms from four 210 ø MM chain stations The second panel shows low-latitude Pi2 pulsation observed from Kakioka The third and fourth panels show energetic electron and ion data observed by LANL , respectively The bottom panel shows Polar PWI electric field observations of AKR The vertical bar in each panel marks the time of auroral breakup, and the width of the bar indicates the uncertainty determining the auroral substorm onset

7 LIOU ET AL- BRIEF REPORT 22, :20 UT 1429:10 UT 1430:24 UT 1432:14 UT 1434:51 UT 1435:18 UT LB s_ LBHi 1436:05 UT 1436:32 UT LBHs 1437:55 UT 1438:22 UT LBH ' 1439:09 UT LBHI 1440:59 UT LBHI 1442:40 UT LBHs 1448:21 UT LBHs ' LI ¾11 LBHi 1450:01 UT 1451:52 UT 1454:29 UT ;Hs 1456:09 UT L Hi t : _ LBH LBHI LBHs Polar Ultraviolet Image Day / LBH LBHI LBHs Photons-cm-=-s ' Plate 3 A sequence of nightside LBH auroral images from Polar ultraviolet imager showing auroral breakup at 1436:05 UT on May 3, 1997 Four 210 ø MM stations are marked in each image

8 22,814 LIOU ET AL' BRIEF REPORT 100 ß a ' ' 'com p93 t UT _1,,,,, i,,,,, i,,,,,, i,_, i, ß ß KTN (6994,20102) TIX (6567,19688) CHD (6467,21212) ZYK ( 962,21672) 62 (b) Kakioka H-magnetometer (c) LANL 1994 E50keV-315keV ' :00 14:10 14:20 14:30 14:40 14:50 15:00 UT (Hour) Plate 4 Substorm event on May 3, 1997, commencing at UT (top) The x component (magnetic north) of magnetograms from four 210 ø MM chain stations (middle) low-latitude Pi2 pulsation observed from Kakioka The third and fourth panelshow energetic electron and ion data observed by LANL , respectively (bottom) Polar PWI electric field observations of AKR The vertical bar in each panel marks the time of auroral breakup, and the width of the bar indicates the uncertainty in determining the auroral substorm onset

9 LIOU ET AL' BRIEF REPORT 22,815 an east-west extent of less than 1 hour Interestingly, the poleward edge of the bulge expanded from 67 ø to 71 ø Mlat within 2 rain, implying an extremely fast poleward expansion speed of,075 km/s Similar to the first case all four ground magnetometer stations were located near midnight on the eastern side of breakup location The geomagnetic field observations from the three high-latitude 210 ø MM stations (no CHD data) shown in Plate 4a indicate a typical pattern (negative bays) of an isolated substorm in the H component of the magnetograms Magnetic field observations from Tixie which is about 2 hour in MLT east of the breakup location show an oscillating H component of magnetic field starting around 1436 UT The onset of the negative bay occurred at UT, about 12 rain delay from the breakup time The Kotel'nyy station, which was located at about the same local time as Tixie but 3 ø - 4 ø higher in latitude than the breakup latitude showed a sharp decrease in the H-component of magnetic field around 1443 UT at, about the time when the auroral bulge reached the station The Zyryanka station was located at - 3 hour in MLT east of the onset bulge and detected little magnetic signature The low-latitude magnetogram from Kakioka for the period from 1400 to 1500 UT is shown in Plate 4b For this event it is clear to see that the onset of the Pi2 pulsation occurred at :58 UT Alternately, it is reasonable to associate a larger, second Pi2 signal at UT with the substorm onset because it coincides with the start of the positive bay The geosynchronous energetic proton and electron flux data is shown in Plates 4c and 4d In this event a clear dispersionless particle injection (simultaneous increase in fluxes at different energies) can be seen in both electron and ion channels around 1436:55 UT q- 10 s, indicating the LANL satellite was located immediately Earthward of the injection center The LANL footprint at the auroral altitude was 2100 MLT and 651 ø Mlat, consistent with the auroral observations Note that the common signature of a sharp decrease in electron flux as seen in the first case is not clearly presented in this case Plate 4e shows the dynamic spectrum of the plasma waves for the frequency range between khz from Polar PWI for the period between 1400 and 1500 UT During this interval, the Polar spacecraft was located on the dayside around the noon In this event a relatively narrow, continuouspectrum of AKR was observed for the entire 1-hour interval There is evidence of enhanced AKR but it begins earlier (about 1425 UT) and continues for about 25 min This observed AKR may correspond to an intense auroral precipitation in the postmidnight region during the late recovery phase of a substorm which started before 1335 UT but faded away after UT The CANOPUS Fort Simpson data (not shown) also show a negative bay beginning around 1325 UT and reaching about -300 nt about 1500 UT Fort Simpson would be near local dawn at this time There are a few "microscale" enhancement and broadening of the AKR band within the 25 rain interval at about 1436, 1441, and 1445 UT, which correlate well with the three low-latitude Pi2 events shown in Plate 4b The first one may be associated with the auroral substorm onset 3 Discussion and Conclusion Comparisons of substorm onset observations made with the space-borne Polar UV imager and groundbased magnetometers from high latitudes indicated that geomagnetic disturbances detected by auroral zone stations are dependent strongly on their instantaneous locations with respect to the expanding auroral bulge It is also suggested that small geomagnetic disturbances may occur during the progress of an auroral substorm expansion The high-latitude magnetic bay can be delayed up to tens of minutes, even if the high-latitude auroral zone geomagnetic observatory is in the midnight sector The sharp decrease in geomagnetic bays occurring simultaneously with an auroral surge expanding overhead was documented some 30 years ago by Meng [1965] and Akasofu and Meng [1967] A schematic diagram of the magnetic signatures before and after the onset of substorm for stations located around the surge was nicely given by Rostoker et al [1980] This study provides a convincing evidence in terms of temporal and spatial relation between the global view of auroral substorm bulge and geomagnetic negative bays It is also important to note that the arrival of the auroral bulge do no necessarily produce a sharp negative bay Small temporal and spatial structures within the substorm current system can affect the ground magnetometer measurements substantially as well For example, in the second event the aurora seems to be located just overhead of Tixie station at UT, but the negative bay started at UT Although the Tixie H-magnetometer data did indicate a-50 nt negative bay at 1400 UT as the aurora expanding toward the station but it recovered and then oscillated for a couple of times On the basis of the auroral images, Tixie was most likely located at the western edge of a westward substorm current system Therefore the oscillatory feature in the observed H magnetic fields can be produced by a current oscillating north-south about the station or a westward current system which was turned on and off during this time interval We have shown that the combination of the positive bay and Pi2 burst is not sufficien to identify substorm onset We have also found the auroral onset to be slightly ahead of the low-latitude Pi2 pulsation by 1 rain from the two events studied, which is consistent with a delay of 1-3 rain reported by Liou et al [1998] We believe this delay is typical although an extensive statistical analysis is necessary to be certain This consistent delay of Pi2 onset relative to au-

10 22,816 LIOU ET AL: BRIEF REPORT roral breakups may suggest a cause-and-effect relationship between the two phenomena and favors the model of auroral zone standing Alfv n waves [eg, Baumjohann and Glassmeier, 1984, and references therein] In this model, standing Alfvdn waves associated with fieldaligned currents on auroral field lines can launch compressional waves radially inward, which then couple to lower latitude field lines as Pi2 signals observed on the ground [Rostoker, 1967; $akurai and McPherron, 1983; Baumjohann and Glassmeier, 1984] If the onset arc is initiated by upward field-aligned currents associated with standing Alfv n waves, the time delay in Pi2 then corresponds to the propagation time for the compressional waves arriving at Earth Dispersionless energetic particle injections are found to lag the auroral breakup by I - 3 min according to the two cases studied Although dispersionless particle injections were observed simultaneously in both electron and ion channels for the two events, there is a level of difference in the injection fluxes For the first event, ion injections were more pronounced than electron injection but the opposite held for the second event This observation is consistent with the characteristic of the "injection periphery" [Reeves et al, 1991] Birn et al [1997] also showed that there can be delay of the dispersionless electron and ion injections with respect to one another and that this delay is local time dependent Note that, however, our result also indicates that the energetic particle injection is centered at the auroral breakup longitudes not the midnight It also suggests that the LANL satellite was not located at the center of the injection front If the injection front has local time structure such that it fans out from the center of the injection, more delay should be expected as one move further out azimuthally from the center of injection On the other hand, it has been reported by Reeves et al [1996] that dispersionless injections measured by both CRRES and LANL at different L shells the difference in the dispersionless injection time could be as much as 5 min for satellites at the same MLT but I RE apart in radius Therefore, unless the injection front actually starts at L = 66 RE, one expects there to be several minutes of delay between substorm onset and geosynchronous injection onset The first event which showed a longer time delay is the characteristic of the geosynchronous observations near the western edge of the injection region where the time delay between sub- storm onset and dispersionless injection is expected to be largest On the other hand, the second event, with 1 min delay, indicates that the LANL satellite was probably located at the same MLT as the injection center and therefore can be used to determine the most earthward injection front for this event With a typical plasma flow speed of km/s [Russell and McPherron, 1973] the earthward injection boundary is estimated to be I - 2 RE tailward of geosynchronous orbit The AKR observations showed a mixed result The first event matched the auroral breakup but the second event did not; actually, there was no apparent wave enhancement in AKR band Although there is evidence of weak enhancement and broadening of AKR band at the auroral onset time, one cannot be certain about this association because there are many of them before and after the onset time The Polar spacecraft was located in the midnight sector ( 80 RE, 773 ø Mlat, and 2125 MLT) for the first event and in the noon sector (,- 71 RE, 693 ø Mlat, and 1143 MLT) for the second event This result, in agreement with previous observational results from Voots et al [1977], suggests that dayside is not a good location for detecting AKR Indeed, AKR is more closely associated with intense electron acceleration in the nightside sector than with diffuse precipitation [eg, Green et al, 1979] However, when suitably located, AKR does identify onsets without a noticeable delay We have illustrated that popular substorm onset indicators other than auroral breakups can lead to inconsistent results and are subject to delays relative to auroral breakups Assuming that all of these substorm onset signatures originate from a common source, a common reference frame is necessary and the auroral breakup should be the natural one for the substorm reference frame The high variability of the time delay, relative to auroral breakups, from many major substorm onset identifiers suggests a need for an extensive study Finally, we recommend intercalibrating various substorm phenomena with a common reference time frame Acknowledgments We thank R P Lepping for the Wind IMF data and M Tezuka and K Takahashi for the 1 s Kakioka magnetometer data We also wish to thank the magnetometer teams of the 210 ø MM stations K Yumoto is the principal investigator for the 210 ø MM magnetometer network R D Belian is the principal investigator for the SOPA on board the LANL D A Gurnett is the principal investigator of Polar PWI This work was supported by NASA grant NAG to the Johns Hopkins University Applied Physics Laboratory The work of M Brittnacher and G Parks was supported by the NASA grant NAG The work of R R Anderson at the University of Iowa was supported by NASA grants NAG and NAG and NASA contract NAS , all with Goddard Space Flight Center Janet G Luhmann thanks John V Olson and another referee for their assistance in evaluating this paper References Akasofu, S-I, The development of the auroral substorm, Planet Space Sci, 12, 273, 1964 Akasofu, S-I, and C-I Meng, Intense negative bays inside the auroral zone I The evening sector, J Atmos Terr Phys, 29, 965, 1967 Akasofu, S-I, and C-I Meng, A study of polar magnetic substorms, J Geophys Res, 7, 293, 1969 Angelopoulos, V, C F Kennel, F V Coroniti, R Pellat, M G Kivelson, R J Walker, C T Russell, W Baumjohann, W C Feldman, and J T Gosling, Statistical characteristics of bursty bulk flow events, J Geophys Res, 99, 21257, 1994

11 LIOU ET AL: BRIEF REPORT 22,817 Arnoldy, R L, and T E Moore, The longitudinal structure of substorm injections at synchronous orbit, J Geophys Res, 88, 6213, 1983 Baker, K B, and S Wing, A new magnetic coordinate system for conjugate studies at high latitudes, J Geophys Res, 94, 9139, 1989 Birn, J, M F Thompson, J E Borovsky, G D Reeves, D J McComas, and R D Belian, Characteristic plasma properties during dispersionless substorm injection at geosynchronous orbit, J Geophys Res, 102, , 1997 Baumjohann, W, and K-H Glassmeier, The transient response mechanism and Pi 2 pulsations at substorm onset - Review and outlook, Planet Space Sci, 32, , 1984 Belian, R D, D N Baker, E W Hones Jr, P R Higbie, S J Bame, and J R Asbridge, Timing of energetic proton enhancements relative to magnetospheric substorm activity and its implication for substorm theories, J Geophys Res, 86, , 1981 Green, J L, D A Gurnett, and R A Hoffman, A correlation between auroral kilometric radiation and inverted V electron precipitation, J Geophys Res, 8J, 5216, 1979 Gurnett, D A, and L A Frank, Observed relationships between electric fields and auroral particle precipitation, J Geophys Res, 78, 145, 1973 Gurnett, D A, et al, The Polar plasma wave instrument, Space Sci Rev, 9'1, , 1995 Liou, K, C-I Meng, A T Y Lui, P T Newell, M Brittnacher, G Parks, and M Nosd, A fresh look at substorm onset identifiers, in Proceedings of the Fourth International Conference on Substorms, edited by S Kokubun and Y Kamide, pp , Kluwer Acad, Norwell, Mass, 1998 Lui, A T, et al, Plasma and magnetic flux transport associated with auroral breakups, Geophys Res Lett, 25, , 1998 Machida, S, Y Miyashita, A Ieda, A Nishida, T Mukai, Y Saito, T Yamamoto, and S Kokubun, Time evolution of the Earth's magnetotail associated with sub- storm onset: GEOTAIL observations, in Proceedings of the Fourth International Conference on Substorms, edited by S Kokubun and Y Kamide, pp , Kluwer Acad, Norwell, Mass, 1998 Meng, C-I, Polar magnetic and auroral substorms, MS thesis, Univ of Alaska, Fairbanks, May 1965 Meng, C-I, and S-I Akasofu, A study of polar magnetic substorms, 2, Three-dimensional current system, J Geophys Res, 7J, 4035, 1969 Miyashita, Y, S Machida, A Nishida, T Mukai, Y Saito, and S Kokubon, Temporal changes in the near and middistant magnetotail associated with substorms obtained by GEOTAIL observations: variations in total pressure and electric field, in Proceedings of the Fourth International Conference on Substorms, edited by S Kokubun and Y Kamide, pp , Kluwer Acad, Norwell, Mass, 1998 Murata, T, H Matsumoto, H Kojima, A Fujita, T Na- gai, T Yamanoto, and R R Anderson, Estimation of tail reconnection lines by AKR onsets and plasmoids entries observed with GEOTAIL spacecraft, Geophys Res Lett, 22, , 1995 Nagai, T, M Fujimoto, Y Saito, S Machida, T Terasawa, R Nakamura, T Yamamoto, T Mukai, A Nishida, and S Kokubun, J Geophys Res, 103, , 1998 Ohtani, S, K Takahashi, L J Zanetti, T A Potemra, R W McEntire, and T Iijima, Tail current disruption in the geosynchronous region, in Magnetospheric Substorms, Geophys Monogr Ser, vol 64, edited by J Kan, T A Potemra, S Kokubon, and T Iijima, p 131, AGU, Washington, DC, 1991 Reeves, G D, R D Belian, and T A Fritz, Numerical tracing of energetic particle drifts in a model magnetosphere, J Geophys Res, 96, 13,997, 1991 Reeves, G D, G Kettmann, T A Fritz, and R D Belian, Further investigation of the CDW 7 substorm using geosynchronous particle data: Multiple injections and their implications, J Geophys Res, 97, 6417, 1992 Reeves, G D, R W H Friedel, M G Henderson, A Lorth, PS McLachlan, and R D Belian, Radial propagation of substorm injections, in Substorms $, fur Space Agency Spec Publ, SP-389, , 1996 Reeves, G D, R D Belian, T C Cayton, M G Henderson, R A Christensen, PS McLachlan, and J C Ingraham, Using Los Alamos geosynchronous energetic particle data in support of other missions, in Satellite-Ground Based Coordination Source Book, edited by M Lockwood, M N Wild, and H J Opgenoorth, fur Space Agency Spec Publ, SP-1198, , 1997 Rostoker, G, The polarization characteristics of Pi-2 micropulsations and their relation to the determination of possible source mechanisms for the production of nighttime impulsive micropulsation activity, Can J Phys, , 1967 Rostoker, G, Macrostructure of geomagnetic bays, J Geo- phys Res, 70 ø, 4217, 1968 Rostoker, G, S-I Akasofu, J Foster, R A Greenwald, Y Kamide, K Kawasaki, A T Y Lui, R L McPherron, and C T Russell, Magnetospheric substorms- Definition and signatures, J Geophys Res, 85, 1663, 1980 Russell, C T, and R L McPherron, The magnetotail and substorms, Space Sci Rev, 15, 205, 1973 Saito, T, K Yumoto, and Y Koyama, Magnetic pulsation Pi 2 as a sensitive indicator of magnetospheric substorm, Planet Space Sci, 24, 1025, 1976 Sakurai, T, and R L McPherron, Satellite observations of Pi 2 activity at synchronous orbit, J Geophys Res, 88, , 1983 Slavin, J A, M F Smith, E L Mazur, D N Baker, E W Hones Jr, T Iyemori, and E W Greenstadt, ISEE 3 observations of Traveling Compression regions in the Earth's magnetotail, J Geophys Res, 98, 15,425, 1993 Tort, M R, et al, A far ultraviolet imager for the international solar-terrestrial physics mission, Space Sci Rev, 71, 329, 1995 Tsyganenko, N A, A solution of the Chapman-Ferraro problem for an ellipsoidal magnetopause, Planet Space Sci, 0ø7, , 1989 Voots, G R, D A Gurnett, and S-I Akasofu, Auroral kilometric radiation as an indicator of auroral magnetic disturbances, J Geophys Res, 82, 2259, 1977 Yumoto, K, and the 210 ø MM Magnetic Observation Group, The STEP 210 ø magnetic meridian network project, J Geomagn Geoelectr, J8, , 1996 R R Anderson, Department of Physics and Astronomy, University of Iowa, Iowa City, IA M Brittnacher and G Parks, Geophysics Program, University of Washington, Seattle, WA K Liou, T Y Lui, C-I Meng, and P T Newell, The Johns Hopkins University Applied Physics Laboratory, Laurel, MD (kan'liøu@jhuapl'edu) G D Reeves, Los Alamos National Laboratory, Los Alamos, NM K Yumoto, Department of Earth and Planetary Sciences, Faculty of Science, Kyushu University, Hakozaki, Fukuoka, Japan (Received December 16, 1998; revised March 19, 1999; accepted April 30, 1999)