Multi-spacecraft THEMIS observat

Document Sample
Multi-spacecraft THEMIS observat Powered By Docstoc
					1    Tail reconnection triggering substorm onset.


3    V. Angelopoulos(1), J. P. McFadden(2), D. Larson(2), C. W. Carlson(2), S. B. Mende(2), H.

4    Frey(2), T. Phan(2), D. G. Sibeck(3), K.-H. Glassmeier(4), U. Auster(4), E. Donovan(5), I. R.

5    Mann(6), I. J. Rae(6), C. T. Russell(1), A. Runov(1), X. -Z. Zhou(1), L. Kepko(7)


7       1. IGPP/ESS, UCLA, Los Angeles, CA USA

8       2. Space Sciences Laboratory, University of California at Berkeley, CA USA

9       3. Code 674, NASA/GSFC, Greenbelt, MD, USA

10      4. TUBS, Braunschweig, D-38106, Germany

11      5. Dept. of Physics and Astronomy, University of Calgary, Calgary, Canada

12      6. Department of Physics, University of Alberta, Edmonton, Alberta, Canada

13      7. Space Science Center, University of New Hampshire, Durham, NH, USA


1    Abstract

2    Magnetospheric substorms explosively release solar wind energy previously stored in

3    Earth’s magnetotail, encompassing the entire magnetosphere and producing spectacular

4    auroral displays. Despite 40 years of research, it is still unclear whether a substorm is

5    triggered by a disruption of the electrical current flowing across the near-Earth

6    magnetotail, at ~10RE (RE = Earth Radius = 6374km) or by the process of magnetic

7    reconnection typically seen further out in the magnetotail, at ~20-30RE. Here we report

8    on the most comprehensive dataset exploring substorm timing collected to-date.

9    Reconnection at 20RE was observed 1.5min prior to auroral intensification, at least 2min

10   prior to near-Earth current disruption, and about 3min prior to substorm expansion onset.

11   These observations show that substorm onsets are likely initiated by tail reconnection.



2    1. Introduction

3       Substorms are global reconfigurations of the magnetosphere involving solar wind

4    energy storage in Earth’s magnetotail and an abrupt conversion of that energy to particle

5    heating and kinetic energy1, 2 . Because phenomena related to substorm onset are initially

6    localized (1-2 RE) in space3,4 but expand quickly to engulf a large portion of the

7    magnetosphere, fortuitous (and thus unoptimized) conjunctions between previous single

8    satellite missions have been unable to pinpoint the exact location of the substorm trigger

9    in space. This has led to diverging theoretical efforts to explain substorm onset 5,6 . The

10   key observational question is whether substorm phenomena are triggered at the first 1-2

11   minutes by a near-Earth dipolarization (current disruption) process, at around 10RE, or by

12   the process of magnetic reconnection at around 20-30RE. The THEMIS mission7,8 was

13   designed specifically to address this question. The mission employs five identical

14   satellites (hereafter termed probes) on orbits enabling recurrent probe alignments parallel

15   to the Sun-Earth line (probes within YGSM±2RE from each other). The probes can thus

16   monitor tail phenomena simultaneously at ~10 RE and at ~20-30RE downtail, while

17   mapping magnetically over a network of ground-based observatories (GBOs) which can

18   determine the meridian and time of substorm onset on the ground. Here we present the

19   first timing results from THEMIS for an isolated substorm, on February 26 th 2008.

20      The ground signatures of auroral substorms consist of a rapid auroral intensification,

21   a breakup of auroral forms into smaller filaments, a poleward expansion and a Westard

22   surge of the most intense auroral arcs. Those signatures are all within 1-2 minutes of each

23   other, and have often been used interchangeably as synonymous to substorm onset. Onset

1    is also commonly identified by ground magnetic signatures of currents associated with

2    auroral arc intensification, including abrupt increases in the auroral electrojet (AE) inde x9

3    and irregular pulsations in the 40-150s range, called Pi2 pulsations10 . Both high (auroral)

4    latitude and mid- latitude magnetometer stations are used for that purpose and both are

5    also known to coincide with auroral onset to within 1 minute 11 .

6       Auroral substorm phenomena were incorporated into the phenomenological model of

7    the magnetospheric substorm12,13 , in the early 1970’s. Prior to substorm onset, during the

8    substorm growth phase, stable arcs intensify and move equatorward while the

9    magnetotail plasma sheet thins and the cross-tail current increases13, 14, 15 . During the

10   expansion phase auroral arcs typically advance towards the poleward edge of the auroral

11   oval and a current wedge develops in space at around 12RE composed of field-aligned

12   currents into and out of the ionosphere13 . Current wedge formation is also referred to as

13   dipolarization, as the field becomes more dipole- like, or as current disruption because it

14   is consistent with a disruption (reduction) of the duskward cross-tail current5 . Further

15   downtail, fast tailward flows threaded by Southward magnetic fields, or Earthward flows

16   threaded by Northward fields are observed near substorm expansion onset, and have been

17   interpreted as evidence for magnetic reconnection16, 17 . The expansion is followed by

18   substorm recovery, during which auroral forms progress at the poleward boundary of the

19   auroral oval for hours, until they eventually reduce in intensity and move equatorward,

20   often starting another substorm sequence. Arc intensification alone does not necessarily

21   constitute a substorm, even though it may involve the same undelying physics as

22   substorms18 . It is the sequence of growth phase, expansion and recovery of the aurora that

23   constitutes a bona fide substorm process. Due to the gradual intensification of auroral

1    arcs at growth-phase it is typically easier to identify a substorm onset by its poleward

2    expansion. For the above reasons poleward auroral expansion onset has been most

3    commonly used as the demarcation of substorm onset. However, when high temporal and

4    fine spatial resolution instrumentation is used, there is a need to differentiate and monitor

5    separately all the aforementioned observational determinations of onset.


7    2. Timing substorm onset

8       Figure 1 shows the THEMIS probe locations at 4:50UT on Feb 26, 2008. All probes

9    were less than 1RE from the nominal neutral sheet. Figure 2(a) shows AET H, a proxy of

10   the Auroral Electrojet (AE) index computed from the magnetic perturbations at twenty

11   THEMIS auroral ground-based observatories (GBOs)19 deployed from Eastern Canada to

12   Western Alaska (several also part of the CARISMA magnetometer array). A sudden

13   AET H increase to 200nT indicates an isolated onset; based on the inflection point in AET H

14   we determine the substorm onset as: 04:54:00UT. Panels (b) and (c) demarcate the

15   substorm onset as determined by the Pi2 magnetic pulsations in the East-West component

16   (Y) at station Gillam (high latitude Pi2 onset) and in the North-South component (X) at

17   station Carson City, NV (mid- latitude Pi2 onset). Data were band-pass filtered in the 10-

18   120s range (contains the 40-120 s Pi2 band but is including higher frequencies to reduce

19   aliasing from filter response at the Pi2 band). Onset was determined, again by visual

20   inspection, as the inflection point in the waveform at the time of the first increase in

21   signal amplitude above background. Onset times are inserted as vertical dashed lines in

22   each panel and are summarized in Table 1.

1       A movie of image mosaics from the THEMIS ASIs is shown in Movie S1 20 . High-

2    resolution stills from that movie are shown in Figure 3(f) and (g) for times 04:50:03UT

3    and 04:53:03UT respectively. Gillam (bottom left) had partially cloudy skies, reflecting

4    the moon from low in the horizon and obscuring the North-East and South-West views

5    (the latter is reflection on the dome). However, these white light imagers have (by design)

6    sufficient sensitivity and dynamic range to discern auroral luminosity changes even in the

7    presence of clouds and the moon. It is evident in panels (e) and (f) that a relatively stable

8    arc extended across Gillam and Sanikiluaq at 04:50:03UT, and that by 04:53:03UT an arc

9    intensification had taken place at Gillam. The spot that brightened (see arrow) was

10   initially ~100km in width (note: the all sky camera field of view is ~800km across the

11   sky, mapped to 110km in altitude, in these images). The onset arc had been intensifying

12   gradually for several minutes prior to 04:50UT, while developing sub-structures: i.e.,

13   small (50-100km scale) filaments which moved along the arc and/or died down. The

14   filaments were observed to intensify and die down at various times in stations both to the

15   East of Gillam (at Sanikiluaq) and to the West (at Rabit Lake, not shown), and on one

16   occasion (at Rabit Lake at ~04:50UT) were accompanied by transie nt enhancements of

17   ULF power. Since onset was localized within the field of view of Gillam, but the

18   filaments were developing for several minutes and over a 2hr magnetic longitude range,

19   we assume that the filaments were not directly related to the localized substorm trigger.

20   The arc intensification can be seen in panel (d). Using the inflection point of the intensity

21   increase we determine the optical onset to be at 04:51:39UT. The expansion arc was at

22   67.8o geomagnetic latitude when it intensified and advanced poleward of 68.2o at

23   04:52:21UT, the time of substorm expansion onset. The latter was corroborated by a

1    similar procedure at station Rankin Inlet, whose field of view overlaps that of Gillam.

2    Examination of the East-West (Y) component at THEMIS mid- latitude stations Pine

3    Ridge, SD (UT of midnight = 07:25UT) and Shawano, WI (UT midnight at 06:05UT),

4    located on either side of Gillam, reveals that opposite sign deflections were observed at

5    these stations. Thus, a substorm current wedge analysis is also co nsistent with our

6    inference from optical data that the onset meridian was located near Gillam.

7       Using the T96 magnetospheric model21 we can project probes P1-P5 along magnetic

8    field lines to the ionosphere near the West coast of Hudson Bay, i.e. very close to station

9    Gillam. The probes’ magnetic footpoints were organized East-to-West in longitude as

10   shown in Figure 2(f) and (g) and were within 1hr of MLT of the optical onset location,

11   indicated by the arrow in panel (g). Therefore, THEMIS probes were ideally positioned

12   to examine relative timing of substorm signatures on the ground and in space.


14      3. Timing magnetotail phenomena

15      Figure 3 shows a 30- min sequence of magnetic field and particle measurements from

16   the THEMIS probes P1, P2, P3 and P5. (Probe P4, to the West of P3 saw similar

17   signatures as P3 but with a 20s delay which we interpret as a spatial effect from

18   Westward expansion). The (negative) Bx decrease on probes P1 and P2, between 04:45-

19   05:01UT in panels (a) and (g) indicates that the two probes started to move away from

20   the neutral sheet. We interpret this as growth-phase plasma sheet thinning. The plasma

21   sheet ion density and average energy (obtained from the data shown on panels (f) and (l))

22   was on the order of 1particle/cc and 1keV respectively. These are conditions of a “cold-

23   dense plasma sheet” following prolonged intervals of northward interplanetary field 22 .

1        P1 observed tailward flows (Vz<0) accompanied by southward (Bz<0) and duskward

2    (By >0) excursions of the magnetic field, followed by Earthward flows and opposite

3    polarity magnetic field perturbations. P2 observed Earthward flows of the same character

4    as P1 and approximately at the same time. These signatures are expected from a

5    reconnection site first located Earthward, then retreating (or reappearing) tailward of

6    probe P1 at 05:01UT. The By signatures (By >0 tailward of the reconnection site; By <0

7    Earthward of it) are also classical Hall signatures of reconnection23, 24, 25 .

8        Prior to the onset of the fast tailward flows, both probes observed two component

9    ions: a 500eV component, commensurate with the cold plasma sheet prior to the event

10   and a 10keV component which appeared gradually. Both probes also observed relatively

11   low temperature electrons (100-200eV) of decreasing flux. These will be discussed again

12   later as further evidence of plasma sheet thinning. However, despite the plasma sheet

13   thinning, P1 and P2 remained in the plasma sheet throughout most of the interval shown,

14   based on the flux levels and average energy of ions and electrons. With the appearance of

15   the fast tailward flows on P1 the average energy of the ions and electrons increased to

16   10keV and 1keV respectively. After the tailward retreat of the reconnection site at

17   05:01UT, both probes observed even hotter plasma (20keV ions, 2keV electrons) and

18   entered very close to the neutral sheet as evidenced by the near- zero transitions of Bx at

19   around 05:02UT. This is evidence of plasma heating at the reconnection outflow and

20   plasma sheet dipolarization at 05:02UT at 22RE, the distance of P1.

21       If the flows are due to reconnection, the current sheet should resemble a slingshot- like,

22   standing Aflvén wave25 . We examined the correlation between the measured ion flows,

23   Vi , and the flows predicted from reconnection outflow, VA ~ ±B·Ni-1/2 on P1 and P2

1    (VA : Alfvén speed, B: magnetic field, N i : the ion density), in order to evaluate the shear

2    stress balance. The correlation coefficients obtained for the Earthward and Tailward

3    flows on P1 and for the Earthward flows on P2 were 0.79, 0.61 and 0.86; whereas the

4    slopes were –0.53, 0.32 and 0.51 respectively. The slopes are likely underestimates

5    because we have not yet included energetic particles in the velocity determination,

6    however the results are robust, showing that the flows are consistent with Alfvénic

7    acceleration, as expected from magnetic reconnection.

8       P3 was near the neutral sheet. It observed fast (>400km/s) Earthward flows starting at

9    around 04:52:27UT, followed by a transient increase in the Northward component of the

10   magnetic field (transient dipolarization) at 04:53:05UT. The onset of fast flows was

11   followed by a more permanent dipolarization at 04:54:40UT, signifying the development

12   of a substorm current wedge in near-Earth space. The transient dipolarization at

13   ~04:53:05 UT is interpreted as the first indication of a substorm current wedge at P3.

14      P5, near geosynchronous altitude, saw an energy-dispersed ion injection of the 50-

15   200keV ions (Figure 3(s)). The flux increase at the lowest energy (highest flux) channel

16   is an exception, as it responds to the local plasma and is correlated with convective

17   velocity changes measured at the same time. The energetic particle dispersion (more

18   energetic particles drifting faster than lower energy particles) is consistent with a

19   duskward drift of those particles to the location of P5 after an injection near midnight.

20   Such dispersed injections are classical signatures of substorms observed by

21   geosynchronous orbit satellites27 .

22      We now turn to detailed timing of the various signatures in space. Figure 4 shows the

23   few minutes prior to substorm onset (04:48:00-04:53:30). The tailward flows on P1

1    started at 04:52:30UT and were preceded by Northward convective flow (Vz>50km/s)

2    starting at 04:50:28UT and an accompanying Southward deflection of the magnetic field

3    (Bz<0). Note that the x-component of the magnetic field was de-trended with a 6min

4    running average in panels (a) and (f) . We interpret these signatures as evidence of the

5    onset of reconnection inflow towards the neutral sheet starting at 04:50:28UT. Ion

6    velocity distributions in the XY despun spacecraft plane (+X = Earthward,

7    +Y=dawnward since P1’s spin axis is pointing south) are shown in Figure 4(c). It is

8    evident that there are two components to the ion distribution: a relatively isotropic

9    component below a few keV and a duskward / tailward streaming component above a

10   few keV. These are the cold and the hot components seen earlier in the spectra of Figure

11   3(d) . Similar behavior was found on probe P2. Energetic ion angular spectrograms on

12   both P1 and P2 (not shown) corroborate the duskward / tailward streaming of the

13   particles above a few keV. The anisotropy of the few keV ions intensified by 04:51:14UT

14   (see distribution function sampled then); even 1keV ions started to exhibit pronounced

15   duskward drift. The gradient scale is approximately the gyroradius of a 1keV proton in

16   the local field (~20nT) i.e., ~600km. This is evidence that the plasma sheet was very thin.

17      Electron velocity space distribution functions (Figure 4(d)) exhibit a bidirectional

18   anisotropy prior to 04:50:54UT, which intensified in the ensuing few minutes. Low

19   energy (50-300eV) electrons are seen streaming towards the reconnection site (opposite

20   to the field direction, i.e., approximately Earthward). Electron distribution function

21   spectra at 0o , 90o and 180o pitch angles are shown at 04:50:35-38UT and 04:52:20-23UT

22   as functions of energy in panel (k). They demonstrate that 50-300eV electrons were

23   indeed streaming opposite to the field (Earthward, towards the reconnection site), while

1    400-2000eV electrons were streaming along the field direction, i.e., away from the

2    reconnection site. Such electron streaming is a signature of reconnection due to the Hall

3    current system17 . The above observations at P1 show that tail reconnection had started by

4    04:50:28UT at the location of P1.

5        P2 was closer to the lobe than P1 as evidenced in the magnetic field and ion energy

6    spectra it measured (Figure 3(g), 3(i)-(j)) 4. No direct connection of field lines at P2 to

7    the reconnection site was evident until after onset, at 05:00:08 UT (Figure 3(i)-(l)), when

8    intense ion and electron fluxes were seen. However, just like at P1 at 04:50:28UT,

9    observations at P2 show at 04:50:38UT the beginning of inflow towards the reconnection

10   site (Vz>0, Figure 3(g)) and the start of a positive deflection of Bz along with a bipolar Bx

11   signature. We interpret these as signatures of a nightside flux transfer event 28, 29 , marking

12   the arrival of signatures of tail reconnection from somewhere tailward of P2. The fact that

13   the deflection of Bz was Northward at P2 and southward at P1 suggests the establishment

14   of a reconnection topology between the two probes, i.e., between 17 and 22 RE.

15       P3 saw a slow ramp-up of the Earthward flow velocity (Vz>50km/s) at 04:52:27 UT.

16   The energetic ion anisotropy (not shown) was consistent with the Earthward motion of

17   the plasma at this time, following a prolonged period of gradient anisotropy within a thin

18   current sheet. The Earthward flows seen at P3 did not necessarily emanate from the

19   reconnection region, but may be nearby plasma which accelerated Earthward due to the

20   forces from the establishment of a reconnection topology further downtail. The first

21   signatures of dipolarization were seen at 04:53:05 UT.


23   4. Summary and discussion

1        Table 1 summarizes our results on timing. The observations at the outermost probes

2    make a compelling case for onset of tail reconnection at or prior to t1=04:50:28UT,

3    between P1 and P2. We can determine the downtail location of the source, x0, and the

4    time of reconnection onset, t0, by approximating the Alfvén speed near the reconnection

5    site as 500km/s (based on a local density measurement of 1 particle/cc and magnetic field

6    of 20nT). Noting that the reconnection pulse can thus travel a distance of 5RE (the P1-P2

7    inter-probe separation) in 60seconds, we obtain (t1-t0)+(t2-t0)=60s, resulting in

8    t0=04:50:03UT and x0=20RE. Time delays relative to that onset time are tabulated in

9    Table 1, third column.

10       It is evident that reconnection onset at 04:50:03UT preceded the onset of a uroral

11   intensification by 96 seconds. Arc intensification was followed, 20sec later, by high-

12   latitude Pi2 onset. The high latitude Pi2 onset may signify the arrival of the field aligned

13   current pulse generated by the reconnection flows in the tail 11 . It is unlikely that a shear

14   Alfvén wave, starting at ~500km/s, can travel from 20RE to the ionosphere in 96 seconds,

15   both due to the high density in the plasma sheet and due to the large distance to the

16   source. Conversely, kinetic Alfvén waves30 of an ion acoustic gyroradius scale may

17   exceed the local Alfvén speed by a factor of sqrt(2) and arrive faster. Those waves can

18   also accelerate electrons31 , which may result in visible aurora ahead of the wave. This

19   would explain our observations of arc intensification by 20s earlier than Pi2 onset and

20   would be consistent with other observations32 of “Alfvénic aurora” at onset.

21       Twenty one seconds after the high- latitude Pi2 onset the aurorae started to expand

22   poleward. This expansion started 6 seconds prior to the arriva l of an Earthward flow

23   perturbation at P3 and about 40 sec prior to the dipolarization at P3. Therefore the initial

1    poleward motion of the aurora cannot be caused by the near-Earth flux pileup of

2    reconnection flows. It is likely associated with the change in magnetic field mapping, as

3    reconnection at 20RE results in higher magnetic latitude flux being engulfed in the

4    reconnection process. Flux pileup and current wedge formation may, however, be

5    responsible for later stages of poleward arc expansion; careful modeling of the current

6    wedge currents in realistic magnetotail fields is needed to properly address this question.

7       Across the field lines the reconnection process started to affect the magnetosphere at

8    11RE (P3’s location) 144s after onset. The time delay relative to reconnection onset is

9    commensurate with the magnetosonic speed of a 1keV plasma sheet near the neutral

10   sheet, ~350km/s. The first evidence of intense dipolarization, interpreted as the

11   reconnected flux arrival at that same location, was seen 30s after the first indication of

12   Earthward flow and 3 minutes after reconnection onset at 20RE. The latter is in agreement

13   with the simultaneously observed plasma flow speed, ~400km/s. The onset of mid-

14   latitude Pi2 pulsations was observed near-simultaneously with the arrival of the

15   dipolarization at 11RE. Mid- latitude Pi2 onset has been interpreted previously as the first

16   evidence of the establishment of the substorm current wedge, and is predominantly an

17   integrated response to high latitude field-aligned currents of the current wedge11 . The

18   observed Pi2 onset at 04:53:05UT was consistent with such an interpretation.

19      As a final comment, in the event studied the plasma sheet was colder and denser than

20   typical, suggesting that the slower Alfvén and magnetosonic speeds may result in longer-

21   than-typical communication times between the various regions within the plasma sheet,

22   as well as between the tail and the ionosphere. This may have been responsible for the

1    easy temporal differentiation of the substorm signatures on the ground and in space

2    observed in this event.

3        In summary, using a THEMIS constellation alignment we have shown that tail

4    reconnection at ~20RE precedes the onset of substorm onset by 1-3 minutes, depending

5    on the reader’s choice of substorm onset identification, and results in the avalanche of

6    near-Earth tail and ionospheric events which constitute the magnetospheric substorm.

7    Reconnection starts deep in the plasma sheet, close to the neutral sheet, and expands

8    towards the plasma sheet boundary as it moves downtail, while heating the plasma and

9    dipolarizing the plasma sheet on the Earthward side. One surprising aspect of our

10   observations is how quickly the aurora intensifies in response to reconnection onset

11   (96sec). Wave-accelerated electrons by reconnection- generated kinetic Alfvén waves

12   may explain this tight coupling between the ionosphere and the reconnecting plasma

13   sheet.


15   Acknowledgme nts

16   This work was supported by NASA contract NAS5-02099. The work of KHG was

17   financially supported by the German Ministerium für Wirtschaft und Technologie and the

18   German Zentrum für Luft- und Raumfahrt under grant 50QP0402. Logistical support for

19   fielding and retrieval of the THEMIS-GBO and CARISMA data is provided by the CSA.

20   CARISMA is operated by the University of Alberta.


1    References

2       1. Akasofu, S. –I., Planet. Space Sci., 12, 273, (1964).

3       2. Axford, I., Phys. Chem. Earth (C), 23, 147, (1999).

4       3. Ohtani, S.-I. et al., AGU Monogr. Ser., 64, 131, (1991).

5       4. Angelopoulos, V. et al., Geophys. Res. Lett., 24, 2271, (1997).

6       5. Lui, A. T. Y., J. Geophys. Res., 101, 13067, (1996).

7       6. Baker, D.N. et al., 101, 12975, (1996).

8       7. Angelopoulos, V. The THEMIS Mission, Space Sci. Rev., in press, (2008).

9       8. Sibeck D.G., and V. Angelopoulos, Space Sci. Rev., in press, (2008).

10      9. Maynaud, P.N., Geophys. Monogr. Ser. 22, American Geophysical Union, (1980).

11      10. Saito, T., Geomagnetic pulsations, Space Sci. Rev. 10, 319, (1969).

12      11. Kepko, L. et al., J. Geophys. Res., 109, A04203, (2004).

13      12. Rostoker, G. et al., J. Geophys. Res., 85, 1663, (1980).

14      13. McPherron, R.L., Rev. Geophys., 17, 657, (1979).

15      14. Coroniti, F. V., J. Geophys. Res., 90, 7427, (1985).

16      15. Sanny, J., et al., J. Geophys. Res., 99, 5805, (1994).

17      16. Russell, C. T., and R. L. McPherron, Space Sci. Rev., 15, 205, (1973).

18      17. Nagai, T., et al., J. Geophys. Res., 103, 4419, (1998).

19      18. Aikio et al., J. Geophys. Res., 104, 12263, (1999).

20      19. Mende, S., et al., Space Sci. Rev., in press, (2008).

21      20. Materials and methods are available as supporting material on Science Online.

22      21. Tsyganenko, N. A., J. Geophys. Res., 100, 5599, (1995).

23      22. Fujimoto, M. et al., Space Sci., Rev., 80, 325, (1997).


2    23. Nagai, T., et al, J. Geophys. Res., 106, 25929, (2001).

3    24. Øieroset et al., Nature, 412, 414, (2001).

4    25. Runov A., et al., Geophys. Res. Lett., 30, 1579, (2003).

5    26. Sonnerup, B. U. Ö., et al., J. Geophys. Res., 92, 12137, (1987).

6    27. Birn, J., et al., J. Geophys. Res., 102, 2325, (1997).

7    28. Semenov, V. S., et al., J. Geophys. Res., 110, A11217, (2005).

8    29. Sergeev, V. A., et al., J. Geophys. Res., 23, 2183, (2005).

9    30. Lysak, R. L., and W. Lotko, J. Geophys. Res., 101, 5085, (1996).

10   31. Chaston et al., Geophys. Res. Lett., 34, L07101, (2007).

11   32. Mende, S. B., et al., J. Geophys. Res., 108, 1344, (2003).

12   33. Larson et al., Space Sci. Rev., in press, (2008).

13   34. Auster, U., et al., Space Sci. Rev., in press, (2008).

14   35. McFadden et al., Space Sci. Rev., in press, (2008).






8    Table 1. Summary of timing results during the Feb 26, 2008 04:53:45UT substorm onset,

9    in order of time sequence. The last column is the time delay assuming reconnection onset

10   at 04:50:03UT, at 20RE, which was arrived at based on our interpretation of data and an

11   estimate of an average Alfvén speed in the plasma sheet of 500 km/s.

                                                                  Infe rred delay (seconds
                  Event                 Observed Time (UT)
                                                                    since 04:50:03 UT)
           Reconnection onset             04:50:03 (inferred)               TRx =0
       Reconnection effects at P1              04:50:28                       25
       Reconnection effects at P2              04:50:38                       35
         Auroral intensification               04:51:39                    TAI=96
         High latitude Pi2 onset               04:52:00                      117
        Substorm expansion onset               04:52:21                   TEX=138
       Earthward flow onset at P3              04:52:27                      144
          Mid-latitude Pi2 onset               04:53:05                      182
           Dipolarization at P3                04:53:05                   TCD =182
       Auroral Electroject Increase            04:54:00                      237


1   Figure 1. Projections of THEMIS probes in X-ZGSM plane along with representative field

2   lines and neutral sheet location in Geocentric Solar Magnetospheric (GSM) coordinates.

3   Times refer to the time delays in Table 1.


6                               T ON=96s      Magnetotail lobe

                                 P5                          TRx=0
                                            P4 P3            x
    ZGSM [RE]

                Neutral sheet                                    P1
                                           TCD=182s     P2
                      Plasma sheet

                XGSM [RE]

1    Figure 2. (a) THEMIS AE index, AET H, computed from THEMIS high latitude GBOs.

2    (b), (c) Band-pass filtered (10s – 120s) ground magnetometer signals from Gillam (East-

3    West component) and from Carson City, NV (North-South component), in nanoTeslas.

4    (d): Integrated auroral intensity from Northern- half of Gillam station field of view

5    (arbitrary units). (e) Latitude of poleward- most extent of auroral luminosity at station

6    Gillam. (f) & (g): Composite images (mosaics) from THEMIS All Sky Imagers (ASIs)

7    around the location of substorm onset, over a continental outline. Stations used in both

8    stills are: Gillam (bottom left), Rankin Inlet (top left) and Sanikiluaq (bottom right). The

9    red line is the midnight meridian and color inserts indicate ionospheric footpoints of

10   THEMIS probes (same symbols as in Figure 1) using the T96 mapping model21 . The

11   arrow indicates optical onset location.



1    Figure 3. Overview of magnetic field and particle data for 30minutes, at 3sec resolution,

2    around the 04:54:00UT substorm. There are 6 panels per probe arranged from top-to-

3    bottom for probes P1, P2, and P3, plus one (bottom) panel showing only 30-200keV ion

4    spectra on P5, from the SST instrument33 . The six panels per probe are (from top to

5    bottom): Magnetic field measured by the FGM instrument34 ; Ion flow velocity measured

6    by the ESA instrument35 ; Energy spectra of 0.005-2000keV ions from the SST and ESA

7    instruments (next two panels); Energy spectra of 0.005-2000keV electrons from the same

8    instruments. All energy spectrograms show omni-directional differential energy flux

9    (eflux) in units of eV/(cm2 ·s·str·eV). The abrupt change in eflux for each species exists

10   because the data below 25keV and above 30keV were obtained by two instruments (ESA

11   and SST respectively) with different instrument geometric factors. Magnetic field and

12   velocity are in GSM coordinates; X, Y, Z components are shown in blue, green and red

13   respectively.


1    Figure 4. Data from P1, P2 and P3, during the first few minutes prior to substorm onset.

2    Panels (a), (f) and (h) show the magnetic field as in Figures 2 and 3, except on P1 the

3    resolution is now 4 samples/s. Note that on P1 and P2 the XGSM component (but not the

4    others) was de-trended (high-pass filtered) by subtracting a 6min running average, to

5    reveal details. Panels (b), (g) and (i) show the ion velocity as in Figures 2 and 4. For

6    dotted lines, see text. Panel (d) and (f) show the ion and electron velocity distribution

7    functions near the spin plane (+X= Earthward and is to the right of the page;

8    +Y=dawnward and is positive to the top of the page); units are in particles/(cm3 km3 /s3 )

9    and X, Y velocity planes are in km/s. Panel (e) shows the energy flux spectra of electrons

10   along (0 Deg.), opposite (180 Deg.) and perpendicular (90 Deg.) to the magnetic field for

11   the times indicated (in minutes and seconds).