Generation of magnetic and particle Pc5 pulsations during ...€¦ · Table 1. Ground stations....

doi: 10.1098/rspa.2010.0079 published online 26 May 2010Proc. R. Soc. A

V. Pilipenko, O. Kozyreva, V. Belakhovsky, M. J. Engebretson and S. Samsonov stormsduring the recovery phase of strong magnetic Generation of magnetic and particle Pc5 pulsations

Supplementary data

pa.2010.0079.DC1.htmlhttp://rspa.royalsocietypublishing.org/content/suppl/2010/05/26/rs

"Data Supplement"

Referencespa.2010.0079.full.html#ref-list-1http://rspa.royalsocietypublishing.org/content/early/2010/05/26/rs

This article cites 53 articles

P<P Published online 26 May 2010 in advance of the print journal.

Subject collections (29 articles)atmospheric science �

Articles on similar topics can be found in the following collections

Email alerting service herethe box at the top right-hand corner of the article or click Receive free email alerts when new articles cite this article - sign up in

articles must include the digital object identifier (DOIs) and date of initial publication. priority; they are indexed by PubMed from initial publication. Citations to Advance online prior to final publication). Advance online articles are citable and establish publicationyet appeared in the paper journal (edited, typeset versions may be posted when available Advance online articles have been peer reviewed and accepted for publication but have not

http://rspa.royalsocietypublishing.org/subscriptions go to: Proc. R. Soc. ATo subscribe to

This journal is © 2010 The Royal Society

on July 14, 2010rspa.royalsocietypublishing.orgDownloaded from

http://rspa.royalsocietypublishing.org/content/suppl/2010/05/26/rspa.2010.0079.DC1.html

http://rspa.royalsocietypublishing.org/content/early/2010/05/26/rspa.2010.0079.full.html#ref-list-1

http://rspa.royalsocietypublishing.org/cgi/collection/atmospheric_science

http://rspa.royalsocietypublishing.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=royprsa;rspa.2010.0079v1&return_type=article&return_url=http://rspa.royalsocietypublishing.org/content/early/2010/05/26/rspa.2010.0079.full.pdf

http://rspa.royalsocietypublishing.org/subscriptions

http://rspa.royalsocietypublishing.org/

Proc. R. Soc. Adoi:10.1098/rspa.2010.0079

Published online

Generation of magnetic and particle Pc5pulsations during the recovery phase of strong

magnetic stormsBY V. PILIPENKO1,2,*, O. KOZYREVA2, V. BELAKHOVSKY3,

M. J. ENGEBRETSON4 AND S. SAMSONOV5

1Space Research Institute, Moscow, Russia2Institute of the Physics of the Earth, Moscow, Russia

3Polar Geophysical Institute, Apatity, Russia4Augsburg College, Minneapolis, MN, USA

5Institute of Cosmophysics and Aeronomy, Yakutsk, Russia

The dynamics of intense ultra-low-frequency (ULF) activity during three successivestrong magnetic storms during 29–31 October 2003 are considered in detail. Thespatial structure of Pc5 waves during the recovery phases of these storms isconsidered not only from the perspective of possible physical mechanisms, but as animportant parameter of the ULF driver of relativistic electrons. The global structureof these disturbances is studied using data from a worldwide array of magnetometersand riometers augmented with data from particle detectors and magnetometerson board magnetospheric satellites (GOES, LANL). The local spatial structure isexamined using the IMAGE magnetometers and Finnish riometer array. Though ageneral similarity between the quasi-periodic magnetic and riometer variations isobserved, their local propagation patterns turn out to be different. To interpret theobservations, we suggest a hypothesis of coupling between two oscillatory systems—a magnetospheric magnetohydrodynamic (MHD) waveguide/resonator and a systemconsisting of turbulence + electrons. We propose that the observed Pc5 oscillationsare the result of MHD waveguide excitation along the dawn and dusk flanks of themagnetosphere. The magnetospheric waveguide turns out to be in a meta-stable stateunder high solar wind velocities, and quasi-periodic fluctuations of the solar wind plasmadensity stimulate the waveguide excitation.

Keywords: ultra-low-frequency waves; magnetic storms; relativistic electrons;magnetometer; riometer

*Author for correspondence ([email protected]).

Electronic supplementary material is available at http://dx.doi.org/10.1098/rspa.2010.0079 or viahttp://rspa.royalsocietypublishing.org.

One contribution to a Special feature ‘Geospace effects of high speed solar wind streams’.

Received 9 February 2010Accepted 23 April 2010 This journal is © 2010 The Royal Society1


mailto:[email protected]

http://dx.doi.org/10.1098/rspa.2010.0079

http://rspa.royalsocietypublishing.org


2 V. Pilipenko et al.

1. Introduction: mechanism of Pc5 waves and their impacton particle dynamics

Geomagnetic ultra-low-frequency (ULF) Pc5 waves (with typical periods ofa few minutes) are a persistent component of the disturbed magnetosphere.Traditionally, it was supposed that the main source of Pc5 pulsations isthe Kelvin–Helmholtz instability (KHI) of the magnetopause engulfed by thesolar wind flow (Yumoto & Saito 1980; Kivelson & Pu 1984). This notionis based on well-established statistical observations of strong growth of Pc5intensity with the increase of the solar wind velocity (Engebretson et al.1998; Mathie & Mann 2001). Jumps or quasi-periodic variations of the solarwind pressure may be an additional source of Pc5 oscillations (Kessel et al.2004). Statistical examination of apparent frequencies of solar wind densityand magnetospheric oscillations showed that certain discrete frequencies in thePc5 band occur more often than do other frequencies in the solar wind andmagnetosphere (Viall et al. 2009). These results argued for the existence ofinherent frequencies in the solar wind that directly drive global magnetosphericoscillations at the same discrete frequencies. In addition, magnetosphericpulsations in the Pc5 frequency band can be effectively excited by fluxes ofenergetic protons with non-Maxwellian distributions in energy or space (Pilipenko1990). However, the latter oscillations are small-scale in the direction along thegeomagnetic shells and they are nearly totally screened by the ionosphere fromground magnetometers.

The dynamics of waves during magnetic storms is observed to be closelyrelated to that of particles, and various kinds of inter-relationships can occur:resonant (Pokhotelov et al. 1986) or non-resonant (Mager & Klimushkin 2007)excitation of magnetohydrodynamic (MHD) waves by instabilities of energeticparticles, modulation of ionospheric fields and currents by quasi-periodic particleprecipitation (Engebretson et al. 1991), modulation of the ring current particlesby MHD waves (Southwood & Kivelson 1981), and energization of the equatorial(Elkington et al. 1999) and auroral (Keiling et al. 2002) electrons by ULFturbulence. To reveal the wave–particle interaction effects from ground-basedobservations, magnetometer data should be supplemented with ionosphericriometer observations, indicating the level of energetic particle precipitation intothe ionosphere.

ULF modulation of precipitating magnetospheric electrons has been revealedby simultaneous magnetic and riometer observations of cosmic noise absorption(CNA; Olson et al. 1980; Kleimenova et al. 1997; Weatherwaxet al. 1997; Poschet al. 1999). The increased CNA may be related to increased intensity ofprecipitating electrons or to an increase in the rigidity of their spectrum. ULFwaves detected by ground magnetometers, however, are not always accompaniedby riometric pulsations (Spanswick et al. 2005). The opposite situation, whenCNA modulation is not accompanied by ground geomagnetic pulsations, is morerare (Krishnaswamy & Rosenberg 1987).

The most currently accepted mechanism for the modulation of electronprecipitation by ULF waves was suggested by Coroniti & Kennel (1970):a compressional wave component b� modulates the growth rate of theelectron–cyclotron very-low-frequency (VLF) instability. However, quite ofteninvestigations of electron precipitation associated with Pc5 pulsations did

Proc. R. Soc. A



Pc5 pulsations during the storm recovery 3

not reveal either b� (Nose et al. 1998) or background VLF turbulence(Paquette et al. 1994). Therefore, other mechanisms should be invoked, such asperiodic acceleration and precipitation of electrons by the parallel electric field(E�) of a dispersive Alfven wave (Stasiewicz et al. 2000), or modulation of the field-aligned potential drop upon Alfven wave interaction with the auroral accelerationregion (Fedorov et al. 2004).

While a general association between geomagnetic storms and relativisticelectron enhancements at geosynchronous orbit has been known for some time(Reeves 1998), the wide variability of the electron response and the puzzling timedelay (approx. 1–2 days) between storm main phase and the peak of the responsehas frustrated the identification of responsible mechanisms. Since the solarwind does not directly contact the electrons in question, some magnetosphericintermediary must more directly provide the energy to the electrons. ULF wavesin the Pc5 band have emerged as a possible energy reservoir for relativisticelectrons (Rostoker et al. 1998). The acceleration mechanisms require seedelectrons of a few hundred kiloelectron volts, which are usually supplied bysubstorms and subsequently energized by Pc5 waves. These electron events arenot merely a curiosity for scientists, but can have disruptive consequences forgeosynchronous spacecraft (Pilipenko et al. 2006).

Some enhancement in electron energies beyond levels expected from conservingadiabatic invariants at geosynchronous orbit occurs rapidly (within a fewhours) at the onset of a magnetic storm, but there is also a slower additionalacceleration, so that peak fluxes are often seen only after a number of days.Considerable evidence has been accumulated favouring the existence of a ULFcontribution to the later, slower energization of electrons (Baker et al. 1998;Mathie & Mann 2001; O’Brien et al. 2001; Romanova & Pilipenko 2008).These studies showed that long duration elevated Pc5 wave power during therecovery phase appeared to discriminate better than any geomagnetic indexbetween those storms that do and do not produce relativistic electrons. Mainphase intensity did not appear to be an important indicator of subsequentelectron behaviour. This led to proposals for an energization mechanism basedon resonant interaction of drifting electrons with coherent Alfven oscillations(Elkington et al. 1999; Liu et al. 1999). This drift-resonance mechanism is in facta revival of the old idea of a ‘geosynchrotron’ (see references in Pokhotelov et al.(1999)). Thus, any model of relativistic electron dynamics should incorporateadequate information about the ULF wave characteristics during magnetic storms(Perry et al. 2005).

However, it is still unclear whether Pc5 pulsations during strong storms arephysically the same as common Pc5 waves, but are just more intense, or not. Themost intense Pc5 waves, with amplitudes exceeding by an order of magnitudethan those of common Pc5 pulsations, were observed during the recovery phaseof severe magnetic storms (Kleimenova & Kozyreva 2005; Kleimenova et al.2005). These anomalously strong pulsations were referred to as global Pc5pulsations (Potapov et al. 2006) because they were observed simultaneouslyin the morning and evening sectors over a wide range of latitudes. Theseglobal Pc5 pulsations were found to be associated with high-speed solar windstreams (HSS). It may be expected that such intense Pc5 waves will producea significant modulation of energetic particles and can be an effective driver ofelectron acceleration.

Proc. R. Soc. A




Table 1. Ground stations. IAGA, International Association of Geomagnetism and Aeronomy;MLT, mean local time.

IAGA CGM CGM midnight observationsstation code latitude longitude MLT UT sampling step

Abisko ABK 65.2 102.3 UT+2.2 21:24 M, R (10 s)Ivalo IVA 65.0 109.8 UT+2.7 20:56 M, R (10 s)Sodankyla SOD 63.8 107.7 UT+2.6 21:02 M, R (10 s)Oulu OUL 61.0 106.6 UT+2.5 21:06 M, R (10 s)Hankasalmi HAN 58.6 105.0 UT+2.4 21:12 M (10 s)Dombås DOB 59.3 90.2 UT+1.4 22:17 M (10 s)Narsarsuaq NAQ 66.3 43.9 UT−2.3 02:07 M (20 s)Meanook MEA 62.2 305.4 UT+15.8 08:12 M (5 s)McMurray MCM 63.4 308.6 UT+16.0 07:59 M (5 s)Tixie TIX 65.7 196.9 UT+8.5 16:04 M (1 s)Chokurdakh CHD 64.8 212.4 UT+9.6 15:09 M (1 s)Zyryanka ZYK 59.9 217.5 UT+9.9 14:48 M (1 s)

In this paper, we consider global Pc5 pulsations of the geomagneticfield and riometric absorption during the recovery phases of the strong,complex magnetic storms that occurred during 29–31 October 2003 (the‘Halloween storm’), and examine their spatial structure and relation tovariations of energetic particle fluxes at geosynchronous orbit and interplanetaryparameters. These observations raise many important questions about thepossible excitation mechanisms of global Pc5 magnetic pulsations and theirimpact on particle dynamics.

2. Observational facilities

The interplanetary parameters are characterized by 1 min magnetic and plasmadata from the advanced composition explorer (ACE) satellite. The data havebeen time-shifted to account for propagation effects from the satellite to themagnetopause. However, both the density and velocity of the solar wind was sostrong that the ACE instrument was completely saturated and only data in thesearch mode (approx. 33 min time resolution) are available from approximately12.40 universal time (UT) on 28 October through approximately 00.50 UT on31 October (Skoug et al. 2004). The OMNI data are practically absent forthis storm, except for very short time intervals, so they have not been used inthis study.

To describe the global structure of H-component (North–South) magneticdisturbances, data from selected stations from the IMAGE, CARISMA/CANMOS, CPMN, GIMA and Greenland Coastal Arrays of magnetometershave been used. Their corrected geomagnetic (CGM) coordinates and samplingrates are given in table 1. The local structure of magnetic and precipitationdisturbances is examined using data from magnetic and riometer stations inFinland and Canada. These stations may be grouped in gradient pairs to revealpropagation effects in the latitudinal and longitudinal directions.

Proc. R. Soc. A




0 MLT

6 MLT18 MLT

12 MLT

MEA

DWS

CHD

ZYKTIX

IVAABK DOB

OULSOD

HAN

NAQ

MCM

ISLPBQ

GOES-12

LANL-01A

LANL-97A LANL-02A

GOES-101990-095

1991-080

1994-084

50

60

70

80

90

Figure 1. Map showing the relative location of magnetic and riometer stations (open triangles), andthe geomagnetic projection of geosynchronous satellites (asterisks and open squares) in geomagneticcoordinates for the moment 05.00 UT.

Table 2. Geosynchronous satellites.

satellite CGM longitude MLT

1990-095 34.16 UT−02.11991-080 266.79 UT−10.31994-084 216.39 UT+10.1LANL-97A 175.27 UT+06.8LANL-01A 80.07 UT+01.2LANL-02A 141.01 UT+04.1GOES-10 296.4 UT−09.6GOES-12 356.5 UT−05.0

The ground observations are augmented by 1 min data from the SOPA particledetectors on board the LANL satellites, which measure electron fluxes in theenergy range of 50–26 MeV. In addition, two other geostationary spacecraft,GOES-12 and GOES-10, have been employed to identify magnetospheric ULFwave activity. The GOES fluxgate magnetometers measure, with a 1 min cadenceand 0.2 nT sensitivity, three components of the magnetospheric magnetic field:Hp is northward parallel to satellite spin axis and perpendicular to its orbit, Heis Earthward along the satellite–Earth line, and Hn is perpendicular to both Hpand He, and directed eastward. The satellite locations are given in table 2. Thelocation of key ground stations and the geomagnetic projections of geostationarysatellites are shown in figure 1 for the 05.00 UT epoch.

Proc. R. Soc. A




3. The space weather event of 29–31 October 2003

During 29–30 October 2003 (the ‘Halloween super-storm’) a series of threeextremely large magnetic storms occurred with intensity |Dst| ∼ 180 nT onapproximately 12.00 UT 29.10 (day 302; figure 2a); |Dst| ∼ 380 nT onapproximately 24.00 UT 30.10 (day 303; figure 2b); and |Dst| ∼ 400 nT onapproximately 00.00 UT 31.10 (day 304; figure 2c). This series of storms wascaused by an extremely fast coronal mass ejection with solar wind velocityV up to approximately 2000 km s−1 and plasma density enhancements up toN ∼ 80 cm−3. During the storm period the solar wind flow was very irregular,and steep increases of N occurred, especially evident on 31 October, 05.30 UTand 11.00 UT. The solar origin of these storms and various aspects of spaceweather associated with them were considered in Panasjuk et al. (2004).

During each storm recovery phase a global intensification of the Pc5 waveactivity was observed, namely 29 October 2003 (approx. 10.00–16.00 UT); 30October 2003 (approx. 05.00–08.00 UT); and 31 October 2003 (approx. 11.00–14.00 UT). The global spatial-temporal dynamics of Pc5 pulsations during thesestorms was considered in detail by Kleimenova & Kozyreva (2005). They noticedseveral uncommon, and still not well understood, features of Pc5 wave activity.The ULF wave activity substantially intensified (about an order of magnitudelarger than typical Pc5 pulsations) during the recovery phases of the 29 Octoberand 31 October storms, but not after the very intense 30 October storm. Theelevated level of Pc5 activity is evident from the behaviour of the ULF wavepower index on 29 October (figure 2a, bottom panel) and 31 October (figure 2b,bottom panel). This ULF wave power index, derived from the data from aworldwide array of ground stations, characterizes the global wave power intensityin the Pc5 band (Kozyreva et al. 2007). At the same time, the ULF indexindicates that the global Pc5 activity is substantially weaker on 30 October(figure 2b, bottom panel).

The global spatial structures of Pc5 waves were different for each storm: on29 October maximal intensity (f ∼ 3 mHz) was observed in two latitudinallynarrow regions in early morning and post-noon hours, whereas on 31 Octobervery intense Pc5 waves were observed throughout a wide dayside region fromapproximately 50◦–70◦. Pc5 pulsations along the IMAGE profile demonstratedtypical features of resonant structure (latitude-dependent spectral peaks) on 29October, whereas on 31 October these features were not evident. Moreover,unusually deep penetration of Pc5 pulsations towards lower latitude (downto approx. 50◦, or L ∼ 2.4) on 31 October demonstrated a local minimum ofamplitude and phase reversal in the H-component at approximately 58◦ latitude(Kleimenova & Kozyreva 2009).

(a) Relativistic electron response to magnetic storms

During the 2003 Halloween storm, the radiation belts were strongly distorted;even the slot region was observed to be filled following storm onset (Baker et al.2004; figure 3). Starting with the first shock at 01.31 UT on 28 October, therelativistic (2–6 MeV) electron flux increased across the outer radiation zone(L > 4) by about two orders of magnitude in approximately 24 h (from less than10 to greater than 103). The Dst excursion at approximately 24.00 UT 30 October

Proc. R. Soc. A




14(a)12108642

2000

4000

3000

2000

1000

0

–100

–200

–300

–4006050403020100

2.22.01.81.61.41.21.000.00 04.00 08.00 12.00 16.00 20.00 24.00

UT

ULF-index

Var(SYM-H)

SYM-H

AE

Bz

V

N

18001600140012001000800

40200

–20–40–60

Figure 2. The basic space weather parameters during three successive storms: (a) 29 October(b) 30 October and (c) 31 October. The first and second panels show solar wind density Nand solar wind velocity V , respectively, from ACE (for 29–31 October in low-resolution mode).The remaining panels show the north–south IMF component Bz as measured by ACE, the AEindex, the SYM-H index the variance Var(SYM-H) of the SYM-H index and the ULF wave powerground index.

Proc. R. Soc. A




14(b)121086420

2000

4000

3000

2000

1000

0

–100

–200

–300

–4006050403020100

2.22.01.81.61.41.21.000.00 04.00 08.00 12.00 16.00 20.00 24.00

UT

ULF-index

Var(SYM-H)

SYM-H

AE

Bz

V

N

18001600140012001000800

40200

–20–40–60

Figure 2. (Continued.)

was accompanied by an abrupt flux decrease to approximately 1 across theentire outer zone. Subsequently, the fluxes recovered over an approximately 3day period, peaking at approximately 103 on 04 November (day 308).

Proc. R. Soc. A




14(c)121086420

2000

4000

3000

2000

1000

0

–100

–200

–300

–4006050403020100

2.22.01.81.61.41.21.000.00 04.00 08.00 12.00 16.00 20.00 24.00

UT

ULF-index

Var(SYM-H)

SYM-H

AE

Bz

V

N

18001600140012001000800

40200

–20–40–60

Figure 2. (Continued.)

In the slot region (2.5 < L < 3), the electron flux increased from approximately102 to 103 in approximately 24 h starting on 29 October, and peaked atapproximately 6 × 103 several days later. The L-value where the maximumoccurred moved from Lmax ∼ 3.5–4.0 29 pre-October to Lmax ∼ 2.3 by 3 November.

Proc. R. Soc. A




SAMPEX: electrons 2–6 MeV

300 301 302 303 304 305 306day of year 2003

1

2

3

4

5

6

7

L v

alue

10–1

1

10

102

103

104

105

e (c

m–2

sr–

1 s–

1 )

Figure 3. The distribution of daily-averaged intensities of relativistic electron fluxes measured bySAMPEX during the Halloween storm, displayed as a function of day of the year in 2003 (horizontalaxis) and L (vertical axis).

This deep penetration of relativistic electrons is consistent with the statisticallinear relationship between the outer radiation belt inner boundary and Dst(Tverskaya 2000).

The rate of radial diffusion by intense ULF waves at the onset of the stormowing to drift resonant acceleration in the slot region was estimated to occurover a time scale of approximately 1 day (Loto’aniu et al. 2006). Though ULFwave-induced diffusion is an important element of the magnetospheric electrondynamics, to interpret the relativistic electron enhancement in the slot region,local acceleration and pitch–angle scattering in the inner radiation belt providedby plasmaspheric hiss and chorus should be invoked (Shprits et al. 2006).

4. ULF wave and riometer activity during 29 October and 31 October storms

We consider in detail the structure of magnetic and riometric Pc5 oscillationsduring two periods with intense ULF activity: the recovery phases of the 29October and 31 October storms. On 29 October, Pc5 activity started around08.00–09.00 UT after the end of the storm main phase and spread to the morning(MEA), noon (NAQ), afternoon (OUL) and night (ZYK) hours (figure 4a). On31 October, Pc5 pulsations were observed throughout both the morning flankof the magnetosphere, from early morning hours up to pre-noon hours (MEA,ABK, NAQ), and the evening flank, from post-noon to midnight hours (OUL,ZYK, MEA; figure 4b).

Comparison of global ULF wave activity with the AE index (figure 4) showsthe tendency mentioned earlier in Samson & Rostoker (1981) and Kleimenova &Kozyreva (2005): the excitation of Pc5 pulsations in the morning sector isaccompanied by substorm occurrence in the dusk–night sector.

Proc. R. Soc. A




4000(a) (b)

AE AE

OUL OUL

NAQ

NAQ

MEA MEA

ZYK ZYK

3000

2000

1000

1000

1500

500

1.35

1.301.301.32

1.281.261.241.221.201.18

1.10

1.05

1.00

1.34

1.32

1.30

1.28

1.26

0.95

1.251.20

1.151.10

1.2

1.1

1.0

0.9

1.30

1.25

1.20

1.15

05.00 10.00 15.00 20.00UT

05.00 10.00 15.0000.00UT

Figure 4. The AE index and the H component of the magnetic field at Oulu, Narsarssuaqand Meanook (all in 104 nT), and Zyryanka (relative units) during two storm intervals: (a) 29October, 05.00–24.00 UT and (b) 31 October, 00.00–19.00 UT. Triangles and asterisks denote localgeomagnetic noon and midnight, respectively.

Despite the global character of Pc5 magnetic activity, the periodic responsein CNA observations can be seen only when a riometer station happens tobe in the morning sector, e.g. at MCM on 29 October, and OUL, MCMand TIX on 31 October (figure 5). At the IMAGE profile only the morningburst of Pc5 waves (04.00–08.00 UT) was accompanied by intense oscillations

Proc. R. Soc. A




4000(a)

3000

2000

1000

12

10

8

6

4

20

10

8

6

4

2

0

0

5

10

15

1

2

3

4

1

0

23456

0

0.5

1.0

1.5

2.0

2.5

(b)

1000

1500

500

AE AE

OUL

MCM MCM

TIX TIX

OUL

05.00 10.00 15.00 20.00UT

05.00 10.00 15.0000.00UT

Figure 5. The AE index and riometer absorption observed at Oulu, McMurray and Tixie duringthe same storm intervals shown in figure 4: (a) 29 October, 05.00–24.00 UT and (b) 31 October,00.00–19.00 UT. Triangles denote local geomagnetic noon and asterisks denote local midnight.

of CNA, whereas the after-midnight and afternoon bursts (02.00–04.00 UT,11.00–14.00 UT) were not, even though the mean CNA levels were nearly thesame. At east-Siberian stations only one of several Pc5 bursts on 31 October,when these stations were in the early morning sector (approx. 05.00 MLT, approx.21.00 UT), was accompanied by enhancement and quasi-periodic variationsof CNA.

(a) ULF wave activity at geosynchronous orbit

During the periods of ground Pc5 activity enhancement on 29 October and 31October, GOES-10 was in the morning sector of the magnetosphere (figure 6).This satellite-detected Pc5 pulsations, most evident in the Hn component.

Proc. R. Soc. A




(a)

100 He He

Hn Hn

B B

MEA, H MEA, H

500

–50–100

50

0

–50

–100

50

0

–50

–100

–150

100

100

50

1.301.34

1.321.30

1.28

1.26

1.25

1.20

1.1505.00 10.00 15.00 20.00

UT

00.00 05.00 10.00 15.00

UT

150

200

100

50

150

200

500

–50–100–150

(b)

Figure 6. ULF activity at GOES-10 in geosynchronous orbit and on the ground at Meanook duringthe same storm intervals shown in figure 4: (a) 29 October, 05.00–24.00 UT, and (b) 31 October,00.00–19.00 UT. The upper three panels show the He component, Hn component and magnitude ofthe magnetic field at GOES-10 (all in nanoTerla), and the bottom panel shows the H component ofthe magnetic field at the near-conjugate station MEA (in 104 nT). Triangles and dark stars denotelocal geomagnetic noon and midnight, respectively.

This kind of transverse polarization, for which the radial component (roughlycorresponding to He) is much less than the azimuthal component (roughlycorresponding to Hn), refers to the toroidal mode Alfven oscillations with small m.

However, the compressional component, as evident from variations of total Bfield, is also significant and comparable to the transverse components. Therefore,this wave mode cannot be interpreted as mere Alfven field line oscillations. Asimilar, but less evident structure of ULF disturbance is observed on GOES-12.

(b) Latitudinal structure of magnetic and CNA oscillations

Spectra for these events (Kleimenova & Kozyreva 2005) showed a distinctivestructure of Pc5 waves: on 29 October there was a latitudinally dependent peakvarying from 3.2 mHz at 65◦ (IVA) to 5 mHz at 55◦ (NUR-TAR), whereas on

Proc. R. Soc. A




20

40

60

80

100

120am

plitu

de (

nT)

–60

–40

–20

0

20

40

60

phas

e (°

)

56 58 60 62 64 66latitude

60

70

80

90

100

110

ampl

itude

(nT

)

56 58 60 62 64 66latitude

–60

–40

–20

0

20

phas

e (°

)

Figure 7. The latitudinal distribution of Pc5 pulsation (H-component) amplitude and phase alongthe IMAGE array (a) from 12.10 to 12.50 UT 29 October 2003, with frequency peaking near3.8 mHz, and (b) from 11.00 to 11.40 UT 31 October 2003, with frequency peaking near 2.9 mHz.

31 October two latitude-independent spectral peaks at 2.8 and 3.6 mHz wereobserved. However, the absence of a regular shift of central frequency with latitudedoes not yet prove the absence of resonant effects, caused by magnetosphericAlfven wave excitation by an external source. The presence of the resonant effectscan be inferred with confidence from gradient analysis (Baransky et al. 1995):peculiar behaviour of spectral amplitude and phase along longitude.

We estimate the latitudinal distribution of amplitude and phase alongthe geomagnetic longitude approximately 107◦ from approximately 66◦ (KIL)through ∼54◦ (TAR) during the intervals with highest Pc5 activity on 29 October(12.10–12.50 UT) and 31 October (11.00–11.40 UT). The phase shifts betweenstations have been estimated using cross-spectral analysis. The global latitudinaldistribution of amplitude and phase at a selected frequency is shown in figure 7.

The Pc5 activity on 29 October shows a typical resonant pattern at f ∼ 3.8 mHz(figure 7a). The amplitude distribution has a localized peak at approximately61◦ latitude. At the same range of latitudes, the phase variation has a steepgradient of approximately 100◦, indicating apparent poleward propagation. Athigher latitudes, in a non-resonant region, the phase variation is much moresmooth. At lower latitudes, far away from the resonant region, irregular phasebehaviour is observed. However, the phase estimates at those stations with verylow spectral power are not very reliable.

On 31 October, the amplitude phase is distinct from the expected resonantstructure (figure 7b). The latitudinal distribution of amplitude at f ∼ 2.9 mHz hasa wide maximum centred at approximately 66◦. A secondary peak is observed at

Proc. R. Soc. A




75(a)

(b)

70

65

60

55

latit

ude

50

45

75

70

65

60

55

latit

ude

50

4500.00 04.00 08.00 12.00 16.00 20.00 24.00

20

40

60

20

40

60

NAL (76.1°)

HRN (74.0°)

KIL (65.8°)

HAN (58.6°)

HLP (50.6°)

BEL (47.3°)

BFE (52.0°)

TAR (54.4°)LOV (55.7°)NUR (56.6°)

DOB (59.3°)OUJ (60.9°)RVK (62.2°)PEL (63.5°)MUO (64.6°)

SOR (67.2°)

LYR (75.1°)

NAL (76.1°)

HRN (74.0°)

KIL (65.8°)

HAN (58.6°)

HLP (50.6°)

BEL (47.3°)

BFE (52.0°)

TAR (54.4°)LOV (55.7°)NUR (56.6°)

DOB (59.3°)OUJ (60.9°)RVK (62.2°)PEL (63.5°)MUO (64.6°)

SOR (67.2°)

LYR (75.1°)

UT

Figure 8. Latitude–UT diagrams showing the temporal variations of the latitudinal distributionof Pc5 power in the 2–6 mHz range along the IMAGE profile on (a) 29 October 2003 and(b) 31 October 2003.

lower latitudes, separated by a local minimum at approximately 59◦. The phasevariation exhibits a rapid change in the region of the amplitude minimum byapproximately 100◦. At the same time, the phase is nearly steady in the regionof the amplitude maximum.

(c) Dynamics of ULF power latitudinal structure

To examine the distribution of ULF wave power across L-shells, we appliedthe technique of F-MLT diagrams, developed in Kleimenova & Kozyreva (2005).These diagrams, shown in figure 8, present the hourly integrated spectral powerof the H-component of the magnetic field in the 2–6 mHz band as a function oflatitude F for the whole day.

On 29 October (figure 8a), a burst of spectral power is observed duringthe storm onset. Regular Pc5 activity lasts from approximately 09.00 UT toapproximately 22.00 UT. During this period the epicentre of Pc5 power gradually

Proc. R. Soc. A




shifts towards lower latitudes from approximately 65◦ to approximately 58◦. Theequatorward boundary of Pc5 activity also shifts to unusually low latitudes, fromapproximately 55◦ to approximately 50◦. It is worth mentioning that the auroralelectrojet, as determined from the IMAGE magnetometer profile, shows a similartendency—a gradual shift to lower latitudes (not shown). Thus, the equatorwardboundary of the Pc5 activity moved to the energization region of relativisticelectrons in the slot region L < 3.

On 31 October (figure 8b) no equatorward latitudinal shift of Pc5 power can beseen; all ULF activations occur nearly at the same latitude, approximately 60◦–65◦. A stable minimum of pulsation power remains at approximately 56◦. Thisminimum corresponds to the latitude of the phase-reversal of H component. Aweaker, but noticeable Pc5 power extends to lower latitudes beyond the minimum,up to approximately 45◦, and even lower.

(d) Azimuthal propagation

An azimuthal propagation pattern has been revealed by a comparison of Pc5variations between longitudinally separated stations. As a typical example, wepresent the analysis of H-component data from the HAN and DOB stationsat geomagnetic latitude F ∼ 59◦, separated by DL = 14.8◦ in longitude (690 km)between 05.30 and 08.00 UT on 31 October, when the IMAGE array is on themorning side. The wavelet spectra show that the pulsation activity is composedof wave packets at two frequencies: f ∼ 2.8 mHz (T ∼ 360 s) and f ∼ 4.6 mHz(T ∼ 220 s) (figure 9a,b). The data have been narrowband-filtered around thosefrequencies, and cross-correlation analysis has been applied to each wave packet,detected simultaneously at both stations. The estimates corresponding to sub-intervals, when a wave packet was clearly evident at one station only, havebeen omitted. The wave packets demonstrate small regular time shifts Dt ∼10–20 s, indicating that pulsations propagate westward, i.e. anti-Sunward. Thecorresponding azimuthal velocities and wavenumbers m have been estimated as

V = DLRE cos F

Dtand m = 360f [Hz]Dt

DL, (4.1)

respectively, and are shown for each frequency in the two bottom panels infigure 9. The estimated values vary from packet to packet in the range m ∼ 0.5–2.5and V ∼ 40–160 km s−1.

The azimuthal propagation patterns at the dusk flank of the magnetosphereat the geomagnetic latitude F ∼ 65◦ are determined using magnetic data fromCHD and TIX stations, though the wave activity at this flank is more irregularand strongly influenced by substorm activity. The central frequency of the waveactivity at the dusk flank, f ∼ 3.4 mHz, is different from those at the morningflank (not shown). Nonetheless, the observed time lags indicate anti-Sunward(eastward) propagation at dusk flank, too.

The average values at the morning flank of Dt = 15 s at f = 4.6 mHz correspondto m ∼ 1.7 and V ∼ 50 km s−1. These estimates exceed somewhat the velocitiesof common Pc5 pulsations, V ∼ 14 km s−1, at sub-auroral latitudes, obtained byOlson & Rostoker (1978). Thus, global Pc5 pulsations, similar to common Pc5pulsations, propagate azimuthally in the anti-Sunward direction at both flanksof the magnetosphere. The large dispersion between various wave trains in the

Proc. R. Soc. A




HAN(a)

(b)

(c)

(d)

DOB

20.00fr

eque

ncy

(mH

z)

14.8210.998.146.034.473.312.461.821.351.00

20.00

freq

uenc

y (m

Hz)

azim

utha

l wav

enum

ber

velo

city

(km

s–1

)

14.8210.998.146.034.473.312.461.821.351.00

0

–0.5

–1.0

–1.5

–2.0

–2.5

–3.0

–160

–140

–120

–100–80

–60–40

05.30 06.00 06.30 07.00

UT

08.0007.30

Figure 9. (a,b) Wavelet spectra, (c,d) azimuthal wavenumbers m and velocities V for magneticvariations at longitudinal profile HAN and DOB for 31 October. (c) Asterisk, 2.8 mHz; diamond,4.6 mHz.

same series of global Pc5 pulsations obtained here indicates that the azimuthalvelocity is not determined by the solar wind velocity, which remains constantwithin 10 per cent around 1000 km s−1 during the whole interval.

Riometric Pc5 variations are less coherent between separated stations whencompared with magnetic signals, and they do not demonstrate a consistentpropagation pattern between IVA and ABK. The cross-correlation coefficientbetween riometer records at IVA and ABK for the whole interval 05.30–07.00 UTon 31 October achieves a peak value R ∼ 0.33 at lag Dt ∼ 20 s, whereas formagnetic pulsations R ∼ 0.48 and Dt ∼ 15 s. Thus, on average both magnetic andriometer pulsations propagate tailward, though with different apparent velocities.

Proc. R. Soc. A




1.21

1.17

1.13

1.0904.00 04.30 05.00 05.30

UT06.00 07.00

0.140

0.723 abso

rbtio

n (d

B)

mag

netic

fie

ld (

× 1

04 n

T)

1.446

2.170

06.30

Figure 10. Superposition of Pc5 magnetic (H-component) and riometer pulsations at Abisko duringthe interval 04.00–07.00 UT on 31 October. Black solid line, magnetic field; grey solid line,absorbtion.

(e) Coherency between CNA and magnetic variations

Some similarity between magnetic and CNA Pc5 pulsations is seen at manystations during the intervals with enhanced ULF activity in the morningsector (e.g. at ABK on 31 October, figure 10). The cross-correlation coefficientfor the entire interval 05.10–07.00 UT on this day reaches peak values attime shifts approximately 60 s for ABK, and approximately 80 s for IVA. Theobserved phase leading of CNA variations with respect to magnetic pulsationsis traditionally understood as an indication that the near-equatorial region ofthe magnetosphere is the modulation region, from which energetic electrons andmagnetic disturbances propagate towards the ionosphere with different velocities.

However, closer examination of the overlaid magnetic (H-component) andriometer variations (figure 10) shows that they are synchronized only for arelatively short time interval (one to two periods) at approximately 05.40 UT,and beyond this interval correspondence between them is lacking. This visualimpression is supported by the cross-correlation coefficient estimated in a runningwindow: R between band-filtered (150–600 s) records reaches significant valuesonly for certain time intervals (e.g. 05.35–05.50 UT). Thus, CNA Pc5 pulsationsare not simply a replica of magnetic Pc5 pulsations; their spatial structures bothin latitudinal and longitudinal directions are not identical. Only for relativelyshort periods do their waveforms become synchronized.

5. Dynamics of energetic particles at geosynchronous orbit

Let us compare the ground ULF wave and CNA activity with the variations ofinterplanetary parameters and dynamics of energetic particles at geosynchronousorbit. At the time of Pc5 activation at 05.00 UT on 31 October LANL

Proc. R. Soc. A




satellites covered the morning flank of the magnetosphere from early morninghours through noon, LANL-90 (approx. 02 MLT), LANL-01 (approx. 05 MLT),LANL-02 (approx. 09 MLT), LANL-97 (approx. 12 MLT), and the evening flankfrom afternoon to dusk hours: LANL-91 (approx. 18 MLT), LANL-94 (approx.14 MLÒ) (figure 1).

Figure 11 shows the fluxes of electrons Je with energies E = 50–75 keV atthe LANL satellites along the electron magnetic drift shell. A sudden dropof Je near noon hours at LANL-02, LANL-97 and LANL-94 at approximately06.00 UT is caused by the passage of these satellites beyond the compresseddayside magnetopause. In the early morning hours, LANL-90 detects acloud of injected electrons drifting across the morning side at approximately06.00 UT. The injection onset coincides with the enhancement of waveactivity and CNA at Finnish stations. During the drift the electron cloudspreads out as evident from the comparison of LANL-90 and LANL-01. Theelectron fluxes in the morning sector (LANL-01) experience intense fluctuations(DJe/Je ∼ 1) in the Pc5 frequency range for all energy channels, in linewith the activation of magnetic pulsations and riometer variations in themorning sector.

Simultaneously with the electron injection, energetic protons (E = 50–400 keV)were injected on the dusk side (not shown). The cloud of injected protons driftedgradually from dusk hours (LANL-91) to early morning hours (LANL-01). Theproton fluxes Jp also demonstrated intense fluctuations with DJp/Jp ∼ 1 in thePc5 frequency band in all energy channels from 75–113 to 400–670 keV.

It is necessary to mention that the waveforms of the quasi-periodic Pc5variations are different for electron and proton detectors, moreover, the waveformsin different energy channels are not similar. These quasi-periodic particlevariations do not follow exactly either magnetic pulsations or CNA variations.This sheds some doubt on these coupling processes as merely the modulation ofparticle fluxes by MHD waves.

6. Quasi-periodic solar wind variations and ground response

The enhancements of the solar wind density N are irregular and exhibit significantquasi-periodic fluctuations with time scales from a few to ten minutes. Variationsof N almost completely determine the variations of the solar wind dynamicpressure P, because V varies noticeably only on much larger time scales.

Comparison of the N variations during the interval 04.00–08.00 UT on 31October with H-component pulsations at ABK shows a similarity in theirdynamics in time (figure 12). Spectra (not shown) for the period 06.10–07.10UT of N at ACE have spectral peaks at 2.2, 3.3 and 4.1 mHz, whereas on theground one observes close spectral peaks of magnetic fluctuations, but shiftedin frequency to 2.5, 3.9 and 4.7 mHz. Even a better correspondence between thedynamics of N (t) fluctuations and ground pulsations at ABK can be seen for theinterval 10.00–14.00 UT on 31 October. Each wave packet detected at ABK hasa counterpart in N (t) variations. Similar to the above interval, the spectral peaksat ABK (3.0 and 3.6 mHz) are shifted to higher frequencies when compared withspectral peaks of N fluctuations (2.7 and 3.3 mHz). The comparison of LANL

Proc. R. Soc. A




1500

1000

500

2.34 × 106

1.22 × 106

8.16 × 105

4.08 × 105

3.36 × 105

2.24 × 105

1.12 × 105

1.84 × 105

1.23 × 105

6.15 × 104

9.11 × 104

4.55 × 104

1.37 × 105

1.5 × 105

1.0 × 105

5.0 × 104

0

0

02.01.5

1.00.5

01.212

1.177 H ABK

ABK

Je

LANL-1991

Je

LANL-1994

Je

LANL-1997

Je

LANL-02

Je

LANL-01

Je

AE

LANL-1990

R

1.142

1.10703.00 05.00 07.00 09.00

UT

11.00 13.00

6.10 × 101

1.56 × 106

7.79 × 105

6.09 × 101

2.03 × 101

Figure 11. Solar wind density N , AE index, fluxes Je (particle/(keV cm2 str s)) of electrons withE = 50–75 keV measured at six LANL satellites at geosynchronous orbit, and riometer absorptionand magnetic field variations (H-component) at Abisko during the interval 03.00–13.00 UT on 31October 2003. Triangles and asterisks denote local geomagnetic noon and midnight, respectively.

electron fluctuations and magnetic and CNA pulsations at the same MLT onthe ground for intervals with elevated Pc5 activity shows that they are evidentlyrelated, but not coherent.

Proc. R. Soc. A




4

3

2

1

60

50

40

30

20

10

8

6

4

2

1.20

1.18

1.16

1.14

1.12

04.00 05.00 06.00

ABK

N

N

N

ACE

time shift = 26 min

time shift = –19 min

H-component

GEOTAIL

WIND

07.00UT

08.00

Figure 12. Comparison of simultaneous variations of the solar wind density N (cm−3) as measuredby ACE (time-shifted by 26 min), GEOTAIL and WIND (time-shifted by −19 min), and the H-component (104 nT) at ABK during the interval 04.00–08.00 UT on 31 October 2003.

To verify further a possible correspondence between irregular enhancementsof the solar wind plasma density and magnetospheric Pc5 activity, it would beinstructive to make an inter-comparison of all 3 days during the Halloween storm.However, owing to the sparse time resolution and rather poor reliability of the

Proc. R. Soc. A




ACE data on 29–30 October it is unfortunately not possible to present convincingevidence of possible pressure fluctuations that could be responsible for the ULFwave stimulation. However, we have inspected other ground-based magnetic data,which are often used as a proxy for changes in the solar wind dynamic pressure.The SYM-H (Dst) index is known to respond closely to such changes accordingto the empirical relation D(Dst) = 7.24

√Pdyn[nPa] (O’Brien & McPherron 2000).

The variability Var(SYM-H) of the detrended series of SYM-H index is shown infigure 2. Fluctuations of SYM-H show that during 30 October the solar wind isindeed more steady than during 29 October and 31 October. The low level ofplasma fluctuations on 30 October may be one reason for the relatively weakexcitation of the magnetospheric ULF wave response during this storm.

Thus, coordinated analysis of the solar wind variations, particle fluxes in themagnetosphere and magnetic field and CNA pulsations on the ground shows thatPc5 waves on the morning flank occur simultaneously with the injection of theelectron and proton clouds. However, though the solar wind velocity is very highall the time, the excitation of Pc5 pulsations occurs not steadily, but only duringspecific time intervals. Therefore, the Pc5 pulsations observed on the ground areexpected to be stimulated by intense solar wind density fluctuations. On the otherhand, though the dynamics of the solar wind fluctuations and magnetosphericoscillations are very similar, the observed global Pc5 pulsations are not just aforced MHD response to solar wind forcing.

7. Discussion

Long-period magnetospheric oscillations can be excited either by large-scalesources, such as a boundary layer shear velocity KHI, solar wind pressurevariations or small-scale sources such as fluxes of energetic particles. Differencesin generation sources are expected to be revealed in the difference betweenthe transverse spatial structures, polarization and propagation velocities ofsuch waves. Large-scale sources predominantly excite MHD oscillations withlarge azimuthal scale, or small m, which can effectively penetrate into theinner magnetosphere and resonantly excite Alfven field line oscillations. In theazimuthal direction, these waves are expected to propagate in the same directionas the solar wind flow, that is anti-Sunward. Energetic particles with non-Maxwellian distributions over velocities or space generate small-scale, that ism � 1, field line oscillations. This theoretical notion has been supported bydirect measurements of the azimuthal scale of storm-time Pc5 using ionosphericradars (see references in the review by Pilipenko (1990)). These oscillationsare expected to propagate in the same azimuthal direction as the particledrift, i.e. sunward. Such small-scale oscillations are considerably screened by theionosphere from ground magnetometers. This screening may explain situationswhen periodic fluctuations of CNA were not accompanied by similar magneticpulsations (Krishnaswamy & Rosenberg 1987).

Consideration of the global Pc5 events presented here raises importantquestions about the physical nature and generation mechanisms of ULFpulsations and their impact on magnetospheric particles during strong magneticstorms. It has been assumed traditionally that the KHI of the interface between

Proc. R. Soc. A




the solar wind flow and the magnetosphere is the main generator of large-scalePc5 pulsations. Global MHD simulations of the steady solar wind–magnetosphereinteraction indeed showed the generation of surface waves driven by the KHI atthe flanks of the magnetopause or inner boundary of the low-latitude boundarylayer (Claudepierre et al. 2008). However, in the events considered here one cansee that favourable conditions for the KHI, though necessary, are not sufficient:Pc5 excitation has been evidently stimulated by solar wind density fluctuations.Indeed, during all storms, the solar wind V is very high, but Pc5 pulsations occuron 29 October and 31 October only during specific time intervals, triggered bysolar wind density fluctuations. In contrast, on 30 October, the solar wind Nis rather steady, without significant fluctuations. The lack of noticeable densityfluctuations, despite a very high solar wind velocity, evidently results in a lowlevel of ULF activity on this day.

The global nature of Pc5 oscillations may prompt a suggestion that thesepulsations are coherent eigenoscillations of the whole magnetospheric cavity.However, examination of the local spatial structure of Pc5 pulsations has revealedan azimuthal propagation pattern. This excludes the possibility of interpretingthese global Pc5 waves as cavity oscillations. We also note that the mechanismof kinetic instabilities, successfully applied for the interpretation of small-scale(m ∼ 30–100) Pc5 excitation by partial ring current protons in the evening sector,cannot be directly applied to the interpretation of global Pc5 wave generation.The Pc5 waves observed are well observed on the ground and cannot be poloidalsmall-scale waves; this has been confirmed by direct m measurements.

The possibility of triggered Pc5 wave excitation might be related to aqualitative distinction in regimes of the solar wind flow around the magnetosphereboundary under moderate and high velocities. Under conditions of moderate V ,unstable oscillations are localized at the magnetopause and decay exponentiallyinside the magnetosphere. Owing to resonant effects, the surface oscillationsinduce a localized field line response of a resonant magnetic shell (electronicsupplementary material). These oscillations do not grow to large amplitudesbecause they are convected rapidly by the solar wind into the magnetotail. Underhigh V conditions, the magnetosphere–magnetosheath interface becomes over-reflecting, that is, magnetospheric MHD modes are amplified upon reflectionfrom this moving boundary (Mann & Wright 1999; Mann et al. 1999). In thiscase, growing disturbances are not oscillations localized at the boundary, butoscillations of the entire MHD waveguide formed between the magnetopauseand a reflection point deep in the magnetosphere (Wright 1994). In a realisticmagnetosphere, probably, the rigid regime of waveguide excitation is realized,for which a finite-amplitude initial disturbance is necessary. The necessarydisturbances can be provided by solar wind plasma variations, as in the Halloweenstorm, or by the injection of an energetic electron cloud, as in the 21 November2003 storm.

Solar wind buffeting can produce a seminal MHD disturbance with a widespectrum of frequencies and wavenumbers. The group velocity in the azimuthaldirection (Y -axis) of an MHD mode trapped in the rectangular waveguide canbe estimated as Vg = vu(k)/vky = ky/u(k)V 2

A (Wright 1994). Further, small-scaledisturbances with m = kyLRE � 1 are convected rapidly with velocity Vg ≈ VAinto the magnetotail, whereas large-scale disturbances with m ∼1 and Vg � VAremain for a longer time in the MHD waveguide and grow to high amplitudes.

Proc. R. Soc. A




Thus, the waveguide dispersion results in the selection of modes with small m. Thefrequency of these modes is determined by the quantization condition kXLX ∼ pn.Estimates of the fundamental quarter-wavelength mode T ∼ 4LX/VA match theperiod of global Pc5 pulsations of approximately 6 min, assuming LX ∼ 5RE andVA ∼ 400 km s−1. The phase reversal near the local latitudinal minimum of globalPc5 pulsations (Kleimenova & Kozyreva 2005) can be related to the node in theradial structure of the waveguide mode. The low magnitude of the Alfven velocityis due to the dominance of O+ ions in the magnetosphere during the recoveryphase of strong magnetic storms.

An additional possibility for the stimulation of magnetospheric MHDoscillations is related to sporadic pressure enhancements in an unsteady solarwind flow. A pressure jump with amplitude DP and front thickness d causesthe displacement of the magnetopause to a new position, and during thisdisplacement, the boundary moves with acceleration g. The addition of theinertia force F = rg ∼ DP/d to the force balance at the boundary results inan increase of the KHI growth rate (Mishin 1993). This increase may bevisualized as a ‘plug-in’ of the Rayleigh–Taylor instability into the KHI. Accordingto this mechanism, the evolution of ULF waves should correspond better tothe derivative of pressure vP(t)/vt than to pressure itself P(t), if during anaccelerated motion of the magnetopause disturbances would grow to a noticeablemagnitude, that is g−1 > t. Therefore, the magnetospheric ULF spectra are to beshifted to higher frequencies when compared with the spectra of the solar windpressure fluctuations, e.g. H (f ) ∼ fP(f ). This shift has been noticed during thecomparison of the spectra of N (f ) fluctuations and ground pulsations H (f ) in the31 October event.

Examination of the spatial structure of geomagnetic ULF pulsations hassuggested that the physical mechanisms of intense Pc5 pulsations on 29 Octoberand 31 October are different. The 29 October event shows the presence ofspecific resonant distortions: latitudinal variation of the spectral power contentand apparent poleward phase propagation. Thus, the observed Pc5 pulsations onthis day are produced by the resonant excitation of Alfven field line oscillations.During the 31 October event, the resonant features are very weak, and this eventis better described as a result of magnetospheric waveguide excitation.

The events presented here show that traditional notions that periodicvariations of particle fluxes in the magnetosphere and CNA in the ionosphereare due entirely to passive ULF wave modulation are insufficient. Even thecomparison of magnetograms and riograms demonstrates that synchronizationbetween magnetic and CNA pulsations occurs only for a limited time period. Thissituation resembles the synchronization of two weakly coupled oscillatory systems.These systems may be an MHD waveguide/resonator and a system consisting ofcyclotron turbulence and electrons. The latter system is described by equationsfrom quasi-linear theory, which have the form of balance equations with their ownquasi-periodic solutions.

In a collisionless near-Earth plasma, pitch–angle diffusion of electrons occursowing to their interaction with electron–cyclotron waves. In the approximationof weak pitch–angle diffusion, the quasi-linear theory equations are reducedto the equations of balance between the number of electrons N e in a fluxtube and turbulent noise energy W . Analysis of these equations showed thatrelaxation oscillations could take place in this combined cyclotron turbulence

Proc. R. Soc. A




and electrons system (Bespalov 1981). The eigenfrequency of these oscillationsU is determined by the particle source power F and the averaged turbulencegrowth rate g = goN/2, namely U2 = goF . In a steady state F = 2S , where S is theflux density of electrons precipitating into the ionosphere, hence U2 = 2goS . Thequality factor of relaxation oscillations is determined by the turbulence decay ratey: Q = U/2y. The occurrence of these eigenoscillations makes the electron fluxessensitive to external periodic disturbances with frequencies close to U. Thus, inthe region of energetic electron injection two coupled resonant contours occur: theMHD waveguide/resonator with characteristic frequency UA and the combinedsystem electrons + electron–cyclotron turbulence with eigenfrequency U. Duringthe moments when these frequencies match, U ∼ UA, the synchronization of thesetwo resonant systems can occur. Probably, this possibility has been realized inthe events presented here. The short duration of synchronization is due to thefrequency mismatch because of temporal variations of U(t).

Summarizing, the global Pc5 pulsations during the recovery phases of strongmagnetic storms caused by HSS are distinct from common Pc5 waves. Thisfact should be accounted for by any adequate models of relativistic electrondiffusion–acceleration by MHD waves/turbulence. In particular, the followingpeculiarities of ULF wave activity during strong magnetic storms should be takeninto consideration:

— The Pc5 excitation is produced by the combined effect of the high solarwind stream and elevated density fluctuations, triggering the KHI;

— The magnetospheric Pc5 waves can not be just localized Alfven fieldline oscillations, but more global MHD oscillations of the magnetosphericwaveguide with a significant compressional component;

— The region with intense Pc5 power can penetrate to much lower L-shells,similar to the equatorward shift of the auroral electrojet and relativisticelectron fluxes;

— The geomagnetic oscillations are coupled with periodic fluctuations ofenergetic magnetospheric particles.

8. Conclusion

We suggest that the high-speed solar wind flow around the morning flank ofthe magnetosphere results in the formation of a meta-stable system: a MHDwaveguide with a over-reflecting boundary. Spontaneous growth of thermalfluctuations (soft excitation) in such a system cannot be realized, because long-growing disturbances are convected into the magnetotail. For the rigid excitationof the system an initial disturbance of a finite magnitude is necessary, whichthen stimulates growth of magnetospheric waveguide modes. Such a disturbanceis produced by periodic fluctuations of the solar wind density.

In the region of electron injection, an interaction occurs between the MHDwaveguide modes and relaxation oscillations of electron fluxes. This interactioncannot be visualized as a mere modulation of electron fluxes by MHD waves,but is revealed as a short-term synchronization between geomagnetic waves andriometric pulsations.

Proc. R. Soc. A




Any adequate models of relativistic electron dynamics should incorporaterealistic information about the ULF wave characteristics during magnetic storms.Effective Pc5 excitation is produced by the combined effect of the high solar windstream and elevated density fluctuations. Magnetospheric Pc5 waves can be eitherlocalized Alfven field line oscillations or MHD oscillations of the magnetosphericwaveguide. The Pc5 dynamics during magnetic storm has the same tendency ofequatorward shift as has been observed for the auroral electrojet and relativisticelectron fluxes. In contrast to Alfven oscillations, the MHD waveguide modes havea considerable compressional component, as has been evidenced by GOES data.Therefore, global Pc5 pulsations could be a more efficient driver for relativisticelectrons than traditional Alfvenic Pc5 pulsations. This possibility should beconsidered by adequate models of electron dynamics.

We acknowledge the data provision of Finnish riometer array (Sodankylä Observatory, T. Turunen),IMAGE (Finnish Meteorological Institute, A. Viljanen), CANOPUS (University of Alberta,I. Mann), GIMA (University of Alaska), Greenland Array (Danish Meteorological Institute,J. Watermann), 210 MM (Kyushu University, K. Yumoto), LANL satellites (Los Alamos NationalLaboratory, G. Reeves), OMNI and ACE (NSSDC, N. Papitashvili) and GOES (Space EnvironmentCenter, H. Singer). This study is supported by grants from the U.S. National Science Foundation(ATM-0827903) to Augsburg College (VAP, MJE) and from RFBR 07-05-00185a (VBB). Weappreciate the thorough reading and constructive suggestions of both reviewers.

References

Baker, D. N. et al. 1998 Coronal mass ejections, magnetic clouds, and relativistic magnetosphericelectron events: ISTP, J. Geophys. Res. 103, 17 279–17 291. (doi:10.1029/97JA03329)

Baker, D. N., Kanekal, S. G., Li, X., Monk, S. P., Goldstein, J. & Burch, J. L. 2004 An extremedistortion of the Van Allen belt arising from the Hallowen solar storm in 2003. Nature 432,878–880. (doi:10.1038/nature03116)

Baransky, L. N., Green, A. W., Fedorov, E. N., Kurneva, N. A., Pilipenko, V. A. & Worthington, W.1995 Gradient and polarization methods of ground-based monitoring of magnetospheric plasma.J. Geomag. Geoelectr. 47, 1293–1309.

Bespalov, P.A. 1981 Precipitation waves in the auroral ionosphere Geomagn. Aeron. 21, 1018–1022.Claudepierre, S. G., Elkington, S. R. & Wiltberger, M. 2008 Solar wind driving of magnetospheric

ULF waves: pulsations driven by velocity shear at the magnetopause. J. Geophys. Res. 113,A05218. (doi:10.1029/2007JA012890)

Coroniti, F. V. & Kennel, C. F. 1970 Electron precipitation pulsations. J. Geophys. Res. 75,1279–1289. (doi:10.1029/JA075i007p01279)

Elkington, S. R., Hudson, M. K. & Chan, A. A. 1999 Acceleration of relativistic electrons viadrift-resonant interaction with toroidal-mode Pc5 ULF oscillations. Geophys. Res. Lett. 26,3273–3276. (doi:10.1029/1999GL003659)

Engebretson, M. J., Cahill L. J., Arnoldy R. L., Anderson, B. J., Rosenberg, T. J., Carpenter,D. L., Inan, U. S. & Eather, R. H. 1991 The role of the ionosphere in coupling upstream ULFwave power into the dayside magnetosphere. J. Geophys. Res. 86, 1527–1542. (doi:10.1029/90JA01767)

Engebretson, M. J., Glassmeier, K.-H., Stellmacher, M. & Hughes, W. J. 1998 The dependence ofhigh-latitude Pc5 wave power on solar wind velocity and on the phase of high-speed solar windstreams. J. Geophys. Res. 103, 26 271–26 283.

Fedorov, E., Pilipenko, V., Engebretson, M. J. & Rosenberg, T. J. 2004 Alfven wave modulationof the auroral acceleration region. Earth Planets Space 56, 649–661.

Proc. R. Soc. A


http://dx.doi.org/doi:10.1029/97JA03329

http://dx.doi.org/doi:10.1038/nature03116


http://dx.doi.org/doi:10.1029/JA075i007p01279

http://dx.doi.org/doi:10.1029/1999GL003659





Keiling, A., Wygant, J. R., Cattell, C., Peria, W., Parks, G., Temerin, M., Mozer, F. S., Russell, C.T. & Kletzing, C. A. 2002 Correlation of Alfven wave Poynting flux in the plasma sheet at 4-7 REwith ionospheric electron energy flux. J. Geophys. Res. 107, 1132. (doi:10.1029/2001JA900140)

Kessel, R. L., Mann, I. R., Fung, S. F., Milling, D. K. & O’Connell, N. 2004 Correlation of Pc5wave power inside the magnetosphere during high speed streams. Ann. Geophys. 22, 629–641.

Kivelson, M. G. & Pu, Z. Y. 1984 The Kelvin–Helmholtz instability on the magnetopause. Planet.Space Sci. 32, 1335. (doi:10.1016/0032-0633(84)90077-1)

Kleimenova, N. G. & Kozyreva, O. V. 2005 Spatial-temporal dynamics of Pi3 and Pc5 geomagneticpulsations during the extreme magnetic storms in October 2003. Geomagn. Aeron. [Englishtransl.] 45, 71–79.

Kleimenova, N. G. & Kozyreva, O. V. 2009 Penetration of Pc5 geomagnetic pulsations to unusuallylow latitudes during the recovery phase of a superstrong magnetic storm of October 31, 2003.Geomagn. Aeron. [English transl.] 49, 1–5.

Kleimenova, N. G., Kozyreva, O. V. & Ranta, H. 1997 Pc5 pulsations in geomagnetic Meld andriometers absorption in the morning sector of auroral latitudes. Geomagn. Aeron. [Englishtransl.] 37, 552–578.

Kleimenova, N. G., Kozyreva, O. V., Manninen, J. & Ranta, A. 2005 Unusual strong quasi-monochromatic ground Pc5 geomagnetic pulsations in the recovery phase of November 2003superstorm. Ann. Geophys. 23, 2621–2634.

Kozyreva, O. V., Pilipenko, V. A., Engebretson, M. J., Yumoto, K., Watermann, J. & Romanova, N.2007 In search of new ULF wave index: comparison of Pc5 power with dynamics of geostationaryrelativistic electrons. Planet. Space Sci. 55, 755–769. (doi:10.1016/j.pss.2006.03.013)

Krishnaswamy, S. & Rosenberg, T. J. 1987 Modulated ionospheric absorption unaccompanied bysignificant geomagnetic pulsations. Mem. Natl Inst. Polar Res. 48, 171–182.

Liu, W. W., Rostoker, G. & Baker, D. N. 1999 Internal acceleration of relativistic electrons by large-amplitude ULF pulsations. J. Geophys. Res. 104, 17 391–17 407. (doi:10.1029/1999JA900168)

Loto’aniu, T. M., Mann, I. R., Ozeke, L. G., Chan, A. A., Dent, Z. C. & Milling, D. K. 2006 Radialdiffusion of relativistic electrons into the radiation belt slot region during the 2003 Halloweengeomagnetic storms. J. Geophys. Res. 111, A04218. (doi:10.1029/2005JA011355)

Mager, P. N. & Klimushkin, D. Yu. 2007 Generation of Alfv’en waves by a plasma inhomogeneitymoving in the Earths magnetosphere. Plasma Phys. Rep. 33, 391–398.

Mann, I. R. & Wright, A. N. 1999 Diagnosing Pc5 magnetospheric waveguide mode excitationmechanisms. Geophys. Res. Lett. 26, 2609. (doi:10.1029/1999GL900573)

Mann, I. R., Wright, A. N., Mills, K. & Nakariakov, V. M. 1999 Excitation of magnetosphericwaveguide modes by magnetosheath flows. J. Geophys. Res. 104, 333. (doi:10.1029/1998JA900026)

Mathie, R. A. & Mann, I. R. 2001 On the solar wind control of Pc5 ULF pulsation power at mid-latitudes: implications for MeV electron acceleration in the outer radiation belt. J. Geophys.Res. 106, 29 783–29 796. (doi:10.1029/2001JA000002)

Mishin, V. V. 1993 Accelerated motions of the magnetopause as a trigger of the Kelvin–Helmholtzinstability. J. Geophys. Res. 98, 21 365–21 371. (doi:10.1029/93JA00417)

Nose, M., Iyemori, T., Sugiura, M., Slavin, J. A., Hoffman, R. A., Winningham, J. D. & Sato, N.1998 Electron precipitation accompanying Pc5 pulsations observed by the DE satellites and ata ground. J. Geophys. Res. 103, 17 587–17 604. (doi:10.1029/98JA01187)

O’Brien, T. P. & McPherron, R. L. 2000 An empirical phase space analysis of ring currentdynamics: solar wind control of injection and decay. J. Geophys. Res. 105, 7707–7719.(doi:10.1029/1998JA000437)

O’Brien, T. P., McPherron, R. L., Sornette, D., Reeves, G. D., Friedel, R. & Singer, H. J. 2001Which magnetic storms produce relativistic electrons at geosynchronous orbit? J. Geophys. Res.106, 15 533–15 544. (doi:10.1029/2001JA000052)

Olson, J. V. & Rostoker, G. 1978 Longitudinal phase variations of Pc4-5 micropulsations.J. Geophys. Res. 83, 2481. (doi:10.1029/JA083iA06p02481)

Olson, J. V., Rostoker, G. & Olchowy, G. 1980 A study of concurrent riometer andmagnetometer variations in the Pc4-5 pulsation band. J. Geophys. Res. 85, 1695. (doi:10.1029/JA085iA04p01695)

Proc. R. Soc. A



http://dx.doi.org/doi:10.1016/0032-0633(84)90077-1

http://dx.doi.org/doi:10.1016/j.pss.2006.03.013











http://dx.doi.org/doi:10.1029/JA083iA06p02481





Panasjuk, M. I. et al. 2004 Magnetic storms in October 2003. Space Res. [English Transl.] 42, 509.Paquette, J. A., Matthews, D. L., Rosenberg, T. J., Lanzerotti, L. J. & Inan, U. S. 1994 Source

regions of long-period pulsation events in electron precipitation and magnetic fields at SouthPole Station. J. Geophys. Res. 99, 3869–3877. (doi:10.1029/93JA03439)

Perry, K. L., Hudson, M. K. & Elkington, S. R. 2005 Incorporating spectral characteristics of Pc5waves into three-dimensional radiation belt modeling and the diffusion of relativistic electrons.J. Geophys. Res. 110, A03215. (doi:10.1029/2004JA010760)

Pilipenko, V. A. 1990 ULF waves on the ground and in space. J. Atmos. Terr. Phys. 52, 1193–1209.(doi:10.1016/0021-9169(90)90087-4)

Pilipenko, V., Yagova, N., Romanova, N. & Allen, J. 2006 Statistical relationships between thesatellite anomalies at geostationary orbits and high-energy particles. Adv. Space Res. 37,1192–1205. (doi:10.1016/j.asr.2005.03.152)

Pokhotelov, O. A., Pilipenko, V. A., Nezlina, Yu. M., Woch, J., Kremser G., Korth, A. &Amata, E. 1986 Excitation of high-beta plasma instabilities at the geostationary orbit: theoryand observations. Planet. Space Sci. 34, 695–712. (doi:10.1016/0032-0633(86)90124-8)

Pokhotelov, O. A., Pilipenko V.A. & Parrot M. 1999 Strong atmospheric disturbances as apossible origin of inner zone particle diffusion. Ann. Geophys. 17, 526–532. (doi:10.1007/s00585-999-0526-2)

Posch, J. L., Engebretson, M. J., Weatherwax, A. T., Detrick, D., Hughes, W. J. & Maclennan,C. G. 1999 Characteristics of broadband ULF magnetic pulsations at conjugate cusp latitudestations. J. Geophys. Res. 104, 311. (doi:10.1029/98JA02722)

Potapov, A., Guglielmi, A., Tsegmed, B. & Kultima, J. 2006 Global Pc5 event during 29–31October 2003 magnetic storm. Adv. Space Res. 38, 1582–1586. (doi:10.1016/j.asr.2006.05.010)

Reeves, G. D. 1998 Relativistic electrons and magnetic storms: 1992–1995. Geophys. Res. Lett. 25,1817–1820. (doi:10.1029/98GL01398)

Romanova, N. & Pilipenko, V. 2008 ULF wave indices to characterize the solar wind—magnetosophere interaction and relativistic electron dynamics. Acta Geophys. 57, 158–170.

Rostoker, G., Skopke, S. & Baker, D. N. 1998 Relativistic electrons in the magnetosphere. Geophys.Res. Lett. 25, 3701–3704. (doi:10.1029/98GL02801)

Spanswick, E., Donovan, E. & Baker, G. 2005 Pc5 modulation of high energy electron precipitation:particle interaction regions and scattering efficiency. Ann. Geophys. 23, 1533–1542.

Samson, J. C. & Rostoker, G. 1981 Response of dayside Pc5 pulsations to substorm activity in thenighttime magnetosphere. J. Geophys. Res. 86, 733. (doi:10.1029/JA086iA02p00733)

Shprits, Y. Y., Thorne, R. M., Horne, R. B., Cartwright, M., Russell, C. T., Baker, D. & Kanekal, S.2006 Acceleration mechanism responsible for the formation of the new radiation belt during the2003 Halloween solar storm. Geophys. Res. Lett. 33, L05104. (doi:10.1029/2005GL024256)

Skoug, R. M., Gosling, J. T., Steinberg, J. T., McComas, D. J., Smith, C. W., Ness, N. F., Hu, Q. &Burlaga, L. F. 2004 Extremely high speed solar wind: 29–30 October 2003. J. Geophys. Res.109, A09 102. (doi:10.1029 2004JA010494)

Southwood, D. J. & Kivelson, M. G. 1981 Charged particle behavior in low-frequencygeomagnetic pulsations. I. Transverse waves. J. Geophys. Res. 86, 5643–5655. (doi:10.1029/JA086iA07p05643)

Stasiewicz, K. et al. 2000 Small scale Alfvenic structure in the aurora. Space Sci. Rev. 92, 423–453.(doi:10.1023/A:1005207202143)

Tverskaya, L. 2000 Diagnosing the magnetospheric plasma structures using relativistic electrondata. Phys. Chem. Earth (C) 25, 39–42.

Viall, N. M., Kepko, L. & Spence, H. E. 2009 Relative occurrence rates and connection of discretefrequency oscillations in the solar wind density and dayside magnetosphere. J. Geophys. Res.114, A01201. (doi:10.1029/2008JA013334)

Weatherwax, A. T., Rosenberg, T. J., Maclennan, C. G. & Doolittle, J. H. 1997 Substormprecipitation in the polar cap and associated Pc5 modulation. Geophys. Res. Lett. 24, 579–582.

Wright, A. N. 1994 Dispersion and wave coupling in inhomogeneous MHD waveguides. J. Geophys.Res. 99, 159–167. (doi:10.1029/93JA02206)

Yumoto, K. & Saito, T. 1980 Hydromagnetic wave driven by velocity shear instability in themagnetospheric boundary layer. Planet. Space Sci. 28, 789. (doi:10.1016/0032-0633(80)90076-8)

Proc. R. Soc. A





http://dx.doi.org/doi:10.1016/j.asr.2005.03.152


http://dx.doi.org/doi:10.1007/s00585-999-0526-2

http://dx.doi.org/doi:10.1007/s00585-999-0526-2


http://dx.doi.org/doi:10.1016/j.asr.2006.05.010





http://dx.doi.org/doi:10.1029 2004JA010494



http://dx.doi.org/doi:10.1023/A:1005207202143





Generation of magnetic and particle Pc5 pulsations during ...€¦ · Table 1. Ground stations....

Documents

Transcript of Generation of magnetic and particle Pc5 pulsations during ...€¦ · Table 1. Ground stations....