On the origin of high-energy particles in the cusp diamagnetic cavity

Page 1

Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81

Contents lists available at SciVerse ScienceDirect

Journal of Atmospheric and Solar-Terrestrial Physics

1364-68

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/jastp

On the origin of high-energy particles in the cusp diamagnetic cavity

K. Nykyri a,n, A. Otto b, E. Adamson b, E. Kronberg c, P. Daly c

a Embry-Riddle Aeronautical University, Daytona-Beach, FL, USAb Geophysical Institute, University of Alaska Fairbanks, Fairbanks, AK, USAc Max Planck Institute for Solar System Research, Germany

a r t i c l e i n f o

Article history:

Received 14 March 2011

Received in revised form

25 June 2011

Accepted 23 August 2011Available online 1 October 2011

Keywords:

Cusp

Particle acceleration

MHD and Test Particle Modeling

Cluster Spacecraft Observations

26/$ - see front matter & 2011 Elsevier Ltd. A

016/j.jastp.2011.08.012

esponding author. Tel.: þ1 386 226 6714.

ail address: [email protected] (K. Nykyri).

a b s t r a c t

We have analyzed Cluster magnetic field and plasma data during a high-altitude cusp crossing in 2003.

The Cluster separation was � 5000 km and provided unique measurements of high energy particle

properties both inside the DiaMagnetic Cavity (DMC) and surrounding magnetosheath. Most of the high

energy electrons and protons had pitch angles of � 901 in the cavity and the high energy particle

intensities dropped as a function of distance from the cavity boundary. By assuming conservation of the

first adiabatic invariant for the electrons our analysis indicates that most of the high-energy electrons

in the diamagnetic cavity cannot directly originate from the magnetosheath or from the magneto-

sphere. Our test particle simulations in a local 3-D high-resolution MHD cusp model show that particles

can gain up to 40 keV and their pitch angles become nearly 901 in the local cusp geometry due to

gradients in reconnection ‘quasi-potential’ agreeing with the Cluster RAPID observations. These results

strongly support a local acceleration of particles in the cusp diamagnetic cavities.

& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The magnetosheath plasma has most direct access to the iono-sphere through the high-altitude cusps (Frank and Ackerson, 1971;Heikkila and Winningham, 1971). The cusps are a key structuralelement in the magnetosphere around which the geomagnetic fieldrotates by 3601. Therefore, there is always a region in the vicinity ofthe cusp and dayside magnetosphere where the geomagnetic fieldlines are anti-parallel to the Interplanetary Magnetic Field (IMF)making magnetic reconnection possible. When reconnection opensgeomagnetic field, denser solar wind plasma gets injected on thesefield lines reducing the magnetic pressure in this region (Lavraudet al., 2004). The large scale cusp structure is determined by theoccurrence of magnetic reconnection at the high-latitude magneto-pause for northward IMF (Burch et al., 1980; Lavraud et al., 2005a)and at the low-latitude magnetopause for southward IMF (Lavraudet al., 2005a; Reiff et al., 1977). This region of depressed magneticfield at the high-altitude cusp is called the Cusp Diamagnetic Cavityand it is frequently filled with high-energy particles (Chen and Fritz,1998, 2001; Fritz et al., 1999; Niehof et al., 2008; Nykyri et al., 2011a;Walsh et al., 2007, 2010; Whitaker et al., 2006, 2007; Zhang et al.,2005). Zhang et al. (2005) showed that ions with energies greater28 keV are present during 80% of the crossings while electrons above40 keV are present only during 22.5% of the exterior cusp crossings.

The origin of these particles has been a long-standing andcontroversial topic. Chen and Fritz (1998) and Chen (2008) have

ll rights reserved.

suggested that energetic ion populations up to MeV energies aregenerated by acceleration via ULF wave ‘turbulence’ present inDMC’s. In addition, Chen and Fritz (2001) and Fritz et al. (2003)showed that cusp particles consist of singly ionized ionospheric(Oþ) and higher charge state solar wind ions ðO4 þ3 and Heþ þ ).Vogiatzis et al. (2008) have shown examples of case studies of DMCswhere supra-thermal Oþ ions are observed simultaneously withbroadband low frequency waves concluding that Oþ ions are locallyaccelerated via wave–particle interactions. It also has been proposedthat particles are not accelerated in the cusp at all. Chang et al.(1998, 2000), Lin et al. (2007), and Trattner et al. (1999, 2001, 2010)have argued that the quasi-parallel bow shock is the source forenergetic ions in the cusp. Trattner et al. (2001) showed thatcharacteristic spectral breaks observed in the DMC are similar toions accelerated at the quasi-parallel bow-shock. Trattner et al.(2010) argued that there is a strong dependence of energetic ions inthe cusp and magnetic connection to the bow-shock. They showedan example of a DMC event that had fewer energetic ionsand demonstrated the lack of connection to the quasi-parallelbow-shock, which would prevent significant energetic particlefluxes from reaching the DMC at the northern cusp. Global hybridsimulations (Lin et al., 2007) have further confirmed that Fermi-typeacceleration in quasi-parallel shock and foreshock regions aregenerating energetic ions that can access the cusp region. However,the bow shock source would not explain high-energy electrons andsingly ionized Oxygen ions often observed in the DMC. It also hasbeen suggested that energetic cusp particles originate from themagnetosphere (Asikainen and Mursula, 2005, 2006; Blake, 1999;Delcourt and Sauvaud, 1998, 1999; Kremser et al., 1995; Lavraudet al., 2005b). Delcourt and Sauvaud (1999) showed that particles

Page 2

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81 71

can drift to the dayside magnetosphere from the near-Earth tail.Based on dependence of proton fluxes both at the cusp and high-latitude plasma sheet (HLPS) on geomagnetic activity, Asikainen andMursula (2005) suggested that HLPS is the source of cusp energeticprotons. They further showed that the both HLPS and cusp electronsdepend on the solar wind speed in the same way as energeticelectrons in the outer radiation belt, suggesting that the HLPS/cuspelectrons and outer radiation electrons are related.

In the present paper we discuss high-energy particle proper-ties during the high-altitude cusp crossing on 14 February 2003and utilize the results from Nykyri et al. (2011a), in order to putthe high-energy particle observations in context of diamagneticcavity structure: reconnection topology and orientations of theboundaries. Assuming adiabatic particle motion our resultsstrongly suggest that electrons are locally accelerated in theDMC. Although some higher energy ions may originate from thebow-shock or magnetosphere, our analysis of CIS and RAPID datapresented in Nykyri et al. (2011a) and herein demonstrate thatthe DMC is likely the main source also for high energy protonsand singly ionized oxygen ions at least for this event. We ask thereader to see Nykyri et al. (2011a) for the detailed description ofthis event—here we only give a brief overview of the event andfocus mainly on high-energy particle (HEP hereafter) properties.

We also show by using test particle simulations in a 3-D high-resolution cusp model that particles can gain energy (at least upto 40 keV) perpendicular to magnetic field if their drift pathscoincide with the reconnection ‘quasi’ potential. Evidence for theexistence of this potential is shown in the Cluster EFW data. Wethus suggest a mechanism different from resonant or stochasticwave–particle interactions responsible for local particle accelera-tion. Indeed, Nykyri et al. (2011b) have shown for the event thatmuch of the ULF turbulence during this DMC crossing was due tothe motion of transient reconnection signatures and cusp bound-aries suggesting that acceleration due to waves in DMCs may notbe as significant as initially thought.

Fig. 1. SC4 observations at 18:30–20:30 UT during diamagnetic cavity and magnetoshe

(in energy flux units, integrated over all angles) for protons (top panel) and for Oþ u

magnetic field, and lagged IMF by � 44 min.

2. Instrumentation

We use data from four instruments onboard Cluster spacecraft(Escoubet et al., 2001). The four spacecraft, multi-instrumentCluster mission has been ideal for studying the high-altitude cuspregions due to its highly elliptical polar orbit. The spacecraft werelaunched in 2000 and have been providing unique in situmeasurements of plasma and magnetic field properties aroundthe Earth for over 10 years. The tetrahedral formation and varyingspacecraft separations (between 100 and 10,000 km) make itpossible to study space plasma at different scales and computegradients of plasma and magnetic fields. This paper presents acase study of high-energy particle properties during Clusterpassage through high-altitude cusp at about 8RE altitude onFebruary 14 2003 when spacecraft separation was � 5000 km.From each spacecraft, we use magnetic field measurements fromthe Flux Gate Magnetometer (FGM) (Balogh et al., 2001), with asampling rate of 4 vectors/s; ion spectra, moments and fluxes vs.pitch angle from the ion COmposition and DIstribution Functionanalyzer (CODIF) from the Cluster Ion Spectrometer (CIS) (R�emeet al., 2001) from spacecraft 4 (SC4) and energetic particle data forall spacecraft from the RAPID (Research with Adaptive ParticleImaging Detectors) spectrometer (Wilken et al., 2001). Theelectric field data are obtained from Electric Field and WaveExperiment, EFW (Gustafsson et al., 2001).

3. Overview of the event

Fig. 1 shows an overview of SC4 proton and Oþ observationsfrom 18:30 to 20:30 UT during diamagnetic cavity and magne-tosheath crossings from CODIF instrument (see caption for moredetails on figure format).

The diamagnetic cavity (DMC) in this plot is characterized bydepressed magnetic field, magnetic field fluctuations and enhanced

ath crossings. The panels show from top to bottom omni-directional spectrogram

p to 40 keV (second panel), proton pitch angles, Oþ pitch angles, velocity vector,

Page 3

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–8172

omni-directional fluxes (intensities) of energetic (up to 40 keV)protons and oxygen ions (Oþ). As can be seen later in RAPID data,the energies of particles in the DMC extend to hundreds of keVduring this event but the cut-off here at 40 keV is due to energyrange of CIS-CODIF instrument. SC4 has the first encounter of a DMCbetween 18:45 and 19:17 UT and a second encounter between19:51 and 20:18 UT. The pitch angles for protons in the cavity aremostly isotropic showing more structure at the inner boundary(magnetosphere-cavity transition) and when IMF is changing moresouthward at � 20 : 03 UT. Pitch angles for oxygen ions in the DMCare less isotropic (in addition to prevailing perpendicular fluxes,during many intervals the Oþ fluxes in the DMC are either parallelor anti-parallel to magnetic field) which can be due their highernon-adiabaticity: Oþ ions are 16 times heavier than protons and forsame energy have a 16 times larger gyro-radius, therefore lowerfrequency waves or smaller gradients than for protons are requiredto violate their adiabaticity. Pitch angle scattering for oxygen withinone gyro-period can thus get Oþ either parallel or anti-parallel tomagnetic field and can be either swept into ionosphere or lost intomagnetosheath. SC4 traversed close to the magnetosheath-DMCboundary and observed high energy particles at � 19 : 24 UT and at� 19 : 42 UT when it briefly dropped back into cavity. Note that inaddition to Oþ observations during these brief re-encounters withthe cavity there are also significant fluxes of Oþ ions with 60–1201pitch angles in the magnetosheath at � 19 : 28 UT, � 19 : 33 UTand at � 19 : 44 UT which is consistent with trapped Oþ popula-tion of cavity origin appearing in the magnetosheath due to theirlarge gyro-radius.

By comparing the Cluster data and numerical simulations,Nykyri et al. (2011a) showed that the DMC forms more tailwardduring northward IMF and more sunward during southward IMF.

MSP CAVITY MSH

UT

Fig. 2. SC1 measurements of electron (top panel), proton (middle), helium (bottom) om

(MSP), cavity and magnetosheath (MSH). For SC4 the same plot is presented in Nykyri

The magnetic field fluctuations in the cavity were shown to be toa large extent spatially moving structures such as reconnectedflux tubes moving by the spacecraft but also some ‘real’ waveactivity such as ion cyclotron and magnetosonic modes waspresent (Nykyri et al., 2011b). Also the strong solar wind dynamicpressure variations move the cavity and embedded structure overthe spacecraft, resulting in ‘turbulent’ signature in the time series(Nykyri et al., 2011a, 2011b).

In Nykyri et al. (2011a) we also demonstrated that oxygen ionsand protons were leaking out of the cavity anti-parallel to themagnetic field during reconnection intervals and at higher ener-gies than the typical magnetosheath energies supporting the ideathat diamagnetic cavity can be the source for HEPs. In the nextsections we will discuss the high-energy particle signatures in thediamagnetic cavity and surrounding magnetosheath and utilizeTables 3 and 4 from Nykyri et al. (2011a) when discussing thespacecraft separations projected along boundary normals duringdifferent crossings in the cavity. Because the high-energy particlefluxes drop as a function of distance from the DMC, it is moreplausible that the DMC rather than bow-shock is the source forhigh-energy particles in the cavity.

4. High-energy particle observations

Fig. 2 shows energy–time spectrograms of omni-directionalelectron (top), hydrogen (middle) and helium (bottom) fluxesmeasured by RAPID instrument at SC1. We have highlighted theintervals corresponding to depressed magnetic field (DMC) and theregions surrounding the DMC with different colors. Comparison ofthis figure also with the total magnetic field measurements shown

MSHCAVITY

ni-directional particle fluxes (spectrograms) at 18:20–20:45 UT in magnetosphere

et al. (2011a).

Page 4

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81 73

in Fig. 2 in Nykyri et al. (2011a) reveals that the high energyparticles up to few hundred keV are present during the intervalswith depressed magnetic field (see also Figs. 3 and 4 in the presentpaper) and fluxes drop several orders of magnitude in the magne-tosheath (MSH) and magnetosphere (MSP). RAPID does not measurethe charge state, so unfortunately we are unable to address therelative abundances of Heþ or Heþþ. Some spacecraft have alsobrief re-encounters with the cavity boundary and high-energyparticles during the ‘sheath intervals’. An example of such aninterval can be seen between 19:40 and 19:45 UT at SC1. Note thatSC1 does not observe intensities of high-energy particles during thefirst ‘sheath interval’ as high as those observed by SC4 at 19:24:18–19:25:16 UT (see Figs. 3 and 4). This is because SC4 traversedcloser to the cavity boundary in the magnetosheath than SC1 and

RAPID proton fluxes oMSP CAVITY M

Fig. 3. Proton intensities at different energy channels for four spacecraft in magnetosph

channels 27.7–64.4 keV, 75.3–92.2 keV, 92.2–159.7 keV and 159.7–374 keV. The bottom

dropped back into cavity (unlike SC1 which remained in themagnetosheath) due to reduced solar wind dynamic pressure(Nykyri et al., 2011a).

4.1. Proton and electron intensities at different energy channels

The four top panels of Fig. 3 show 1 min averaged protonomni-directional fluxes (intensities) at energies of 27.7–64.4 keV,75.3–92.2 keV, 92.2–159.7 keV and 159.7–374 keV, respectivelyand Fig. 4 shows the same for electrons at energy channels of37.3–50.5 keV, 50.5–68.1 keV, 68.1–94.5 keV, and 94.5–127.5 keV,respectively. The bottom panel in both plots has the 1-min averagedmagnetic field measurements putting the high-energy particle mea-surements in context with the DMC. It is evident that all the

n 14th of February, 2003SH MSH

27.7-64.4 keV

75.3-92.2 keV

92.2-159.7 keV

159.7-374 keV

UT

CAVITY

ere (MSP), cavity and magnetosheath (MSH). The panels from top to bottom are for

panel shows the total magnetic field strength.

Page 5

RAPID electron fluxes on 14th of February, 2003MSP CAVITY MSH CAVITY MSH

37.3-50.5 keV

50.5-68.1 keV

68.1-94.5 keV

94.5-127.5 keV

UT

Fig. 4. Electron intensities at different energy channels for four spacecraft in magnetosphere (MSP), cavity and magnetosheath (MSH). Energy channels from top to bottom

are for channels 37.3–50.5 keV, 50.5–68.1 keV, 68.1–94.5 keV, and 94.5–127.5 keV. Total magnetic field is shown at the bottom panel.

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–8174

spacecraft observe enhanced intensities of energetic particles duringintervals of the depressed magnetic field and as the spacecraft areswept into the magnetosheath due to enhanced dynamic pressure at� 19 : 20 UT, the HEP intensities drop by 1–3 orders of magnitude atSC1, SC2 and SC4. For the first cavity interval � 18 : 50–19:18:30 UT,the intensities of 159.7–374 keV protons are about two orders ofmagnitude lower than the intensities of the 27.7–64.4 keV protons.This illustrates that fewer particles gain the highest energies in thecavity. This is in agreement with the suggested acceleration mechan-ism discussed in Sections 5 and 6 showing that only those particleswhose drift paths coincide with the gradients of reconnection quasi-potential are energized. Particles also need to remain in the DMCsufficiently long time without scattering into loss cone so that theycan be re-cycled though this potential to gain the highest energies(see Section 5 for more details).

Large gradients in particle intensities are observed at the tail-lobemagnetosphere-cavity boundary and at the cavity-magnetosheathboundary. For example, at 18:43 UT SC4 measures about two ordersof magnitude (� 2980 flux units) higher proton intensities than SC1at 27.7–64.4 keV energies. Table 3 of Nykyri et al. (2011a) indicatesthat SC4 and SC1 are separated 3760 km along boundary normalmeasured by SC4 at 18:36–18:53 UT. The gradient in protonintensities of the lowest energy channel along this boundary normaldirection can thus be estimated to be � 0:79 flux units/km.Magnetic field strength is � 15 nT smaller at SC4 than at SC1. WhenSC1 crosses into the weak field region at 18:51 UT it measuresroughly the same intensities as SC4, and intensities drop again whenit moves back to the magnetosphere at � 18 : 55 UT. These transi-tions for SC1 are produced by back and forth motion of themagnetosphere and the cavity boundary over SC1 (Nykyri et al.,

Page 6

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81 75

2011b). Magnetic field magnitude has a large gradient at the tail-lobe magnetosphere cavity boundary, so assuming adiabatic particlemotion, particles that have pitch angles of 901 in the cavity cannotmove into the region of higher magnetic field, which would explainthe gradient in the intensities at this boundary. However, pitch anglescattering into the loss-cone via some non-adiabatic process wouldallow some particles to escape the DMC. The same applies also forthe cavity-magnetosheath boundary since the magnetic field in themagnetosheath is larger than in the cavity. Indeed, at 19:23 UTdifference between fluxes (of first proton channel) at SC3 (whichremains in the cavity) and SC1 (which has moved into magne-tosheath) is roughly 7990 flux units and the distance between SC3and SC1 along ‘best’ boundary normal measured by SC1 at 19:17–19:20 UT is 4810 km (see Table 3 in Nykyri et al., 2011a) yielding agradient of � 1:66 flux units/km.

It is now useful to estimate the gyro radii of the particles in thecavity. Depending on its drift path, a proton trapped in the cavitywill encounter spatially and temporally varying magnetic fields.Assuming that a particle in the cavity only has perpendicularvelocity, the gyro radii of the 27.6, 64.4, 75.3, 92.2, 159.7,374.0 keV protons are 600, 920, 990, 1100, 1440 and 2200 kmrespectively, for a magnetic field of 40 nT (which is a typicalmaximum field observed in the cavity between 18:50 and19:18:30 UT by SC1 and SC4). Assuming the conservation of thefirst adiabatic invariant, the perpendicular energy of the protonwill decrease by a factor of 4, when the particle drifts into amagnetic field of 10 nT (a typical minimum field observed in thecavity between 18:50 and 19:18:30 UT by SC1 and SC4). Thecorresponding gyro radii in 10 nT field region are 1200, 1830,1980, 2200, 2890, 4420 km, respectively.

As SC4 drops briefly back into the diamagnetic cavity from themagnetosheath at � 19 : 24 UT it observes similar plasma proper-ties as SC3 and also re-encounters the enhanced HEP intensities.During this interval, both SC2 and SC1 remain in the magnetosheathmeasuring enhanced (from background) intensities but about anorder of magnitude lower than SC4. SC1 which is 1200–2470 kmoutward of SC4 along the boundary normal direction (Nykyri et al.,2011a) measures proton intensities that are an order of magnitudelower than those observed by SC2 and two orders of magnitudelower than intensities measured by SC4. Note, that in the cavity, SC1observed only slightly higher intensities than SC2, but at � 19 :24 UT the order is reversed and SC2 measures much higherintensities than SC1 probably because SC2 gets closer to the high-energy particle source (detected by SC4) than SC1. However, theintensities of the highest energy channel for the 159.7–374 keVprotons are more similar at SC1 and SC2.

Table 3 in Nykyri et al. (2011a) shows that separation betweenSC1 and SC2 projected along the SC4 boundary normals calculatedbetween 19:24:18 and 19:25:12 UT are ranging from 70 km to1690 km with an average distance of 830 km. This is of the orderof gyro radii of the 27.7–75.3 keV protons assuming a fieldstrength of 20 nT. If there is a high energy particle source at thevicinity of SC4 or SC3, it is probable that the gyro-orbit of some ofthese lower energy particles would reach SC2 in the magne-tosheath but not quite SC1, as it is about one gyro-radius outwardof SC2 in the magnetosheath at � 19 : 24 UT. At the highestenergy channel, the intensities measured by SC1 and SC2 aresimilar because the gyro orbits of the protons become larger thanthe separation between SC1 and SC2.

SC1 and SC4 have a second encounter while in the magne-tosheath with cavity-like plasma at � 19 : 40219 : 43 UT. Theproton intensities measured by SC3 and SC4 are about the same atthe lowest energy channel as during the first sheath interval but forthe higher energy channels the intensities are lower. For SC1 theproton intensities of all the energy channels are higher than duringthe first sheath interval and for SC2 the intensities of the lowest two

energy channels are higher than those of the first interval and are ofthe same order for the highest two energy channels.

For SC1 the observation of these enhanced proton intensitiescompared to the previous interval is no surprise, because unlikeduring the first sheath interval, it now moves from magnetosheath-like plasma to similar plasma environment as encountered by SC3and SC4, characterized by enhanced temperature and stagnantplasma. Because there are no ion (CIS) measurements for SC2, wecannot say whether it transitions to a similar plasma environment asthe other three spacecraft but it does measure a depression in themagnetic field strength (although smaller than SC1 and SC4), so thisis plausible. Table 4 in Nykyri et al. (2011a) shows that compared tothe previous magnetosheath interval, now SC1 and SC4 are closer toeach other along the boundary normal direction (separation of SC1and SC4 projected along boundary normals calculated by SC1 at19:40:28–19:42:28 UT and by SC4 at 19:40:12–19:41:42 UT rangebetween 60 and 660 km calculated by using two different methods)than SC2 and SC4 (separation along normals range between 640 and1090 km), which may explain why intensities measured by SC1 andSC4 are higher than those measured by SC2.

Compared to proton intensities, the electron intensities are lowerand show more variability in the cavity. For example, the electron(unlike proton) intensities of all energy channels and at all space-craft are significantly reduced � at 19:05 UT. This fine structuremay be explained by differences in drift paths for electrons andprotons: The particle drift path is a combination of gradient,curvature and E� B-drift. The gradient and curvature drift velocitiesdepend on the particle’s energy and thus they control the motion ofhigher energy particles, whereas E� B-drift tends to control themotion of the lower energy particles. Gradient and curvature driftsalso depend on charge, making electrons and protons drift toopposite directions. For these reasons, and also due to largelydifferent size of gyro-radii for electrons and protons, their driftpaths are completely different. The diamagnetic cavity is filled withmagnetic field fluctuations, so during one gyro-orbit an ion can ‘see’a completely different magnetic field structure compared to anelectron which will result in different drift paths. Because theelectron drift paths are more constrained than those of protons,the region of maximum electron energy gain is more confined whichcan lead to reduced intensities observed at � 19 : 05 UT. Thereforethere is an increased likelihood for protons to encounter the regionsof reconnection ‘quasi’-potential (see Section 5) and get accelerateddue to their less confined drift-paths and subsequently to beobserved by the spacecraft. This may explain why proton intensitiesin the cavity are higher than electron intensities at correspondingenergies. This also leads to apparently larger gradients in intensitiesbetween cavity and magnetosheath for protons than for electronswhich at first may seem puzzling based on the gyro-radius argu-ment. Although the majority of the electrons in the cavity aretrapped (as illustrated in Section 4.2), there are some electrons(same applies for protons) that are in the loss cone and can escape tothe magnetosheath from the cavity, which can explain some of thefinite fluxes of electrons and protons in the magnetosheath inaddition to gyro-radius leakage. Comparing for example the thirdenergy channel for protons and fourth energy channel for electrons(approximately the same energy range for both species) for SC4 inthe magnetosheath at 19:30–19:40 UT shows that proton fluxes areabout twice higher in the magnetosheath which can be due (i) gyro-radius leakage, (ii) bow-shock ions, (iii) pitch angle scattering fromcavity due to higher non-adiabaticity of protons.

4.2. Electron pitch angles at the magnetosheath and at the

diamagnetic cavity

Fig. 5 represents energy-time spectrogram (top panel) andpitch angle (PA) distribution (bottom panel) for electrons. The

Page 7

Fig. 5. SC4 measurements of electron intensities (top) and pitch angle distribution of 37.3–50.50 keV electron differential fluxes (bottom). The pitch angle distribution is

measured by IES instrument in three polar directions (each within 201). One is in the direction perpendicular to the magnetic field; another 901 to the first one and third is

either parallel or anti-parallel to the spin axis. Therefore, the pitch angles of 01 and 1801 are always measured. Particles outside the envelope bounded by the black curve

are in the loss cone. Particles inside the black envelope cannot directly originate from the magnetosheath or magnetosphere.

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–8176

differential flux of electrons is spread for a wide range of pitchangles showing most flux at PA of 70–1101. There are very fewelectrons traveling parallel or anti-parallel to the magnetic field.The pitch angle distribution is measured by IES instrument inthree polar directions (each within 201). One is in the directionperpendicular to the magnetic field; another at 901 to the first oneand the third is either parallel or anti-parallel to the spin axis.Therefore, the pitch angles of 01 and 1801 are always measured.

Also during the ‘sheath’ interval at � 19 : 24 UT and at 19:40–19:42 UT, there are not many electrons with PA’s of 1801 or 01.The black lines represent the boundary of the loss cone for theparticles inside the cavity: particles that have pitch anglesbetween the black lines are trapped and cannot originate fromthe magnetosheath or magnetosphere directly, but need to beprocessed by some non-adiabatic processes inside the cavity.

This loss cone pitch angle

a¼v?cavity

vJcavity

¼ a tan1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Bsheath=Bcavity�1p !

ð1Þ

is calculated assuming the conservation of the particle energy andassuming that particles in the magnetosheath or magnetosphereonly have perpendicular kinetic energy as they get reflected whentheir parallel velocity goes to zero

1=2mv2Jcavityþ1=2mv2

?cavity¼ 1=2mv2

?sheathð2Þ

and assuming the conservation of the first adiabatic invariant

mcavity ¼1=2mv2

?cavity

Bcavity¼ msheath ¼

1=2mv2?sheath

Bsheathð3Þ

From Eq. (3) it follows that

v2?sheath

¼Bsheath

Bcavityv2?cavity

ð4Þ

and substituting Eq. (4) into Eq. (2) yields Eq. (1) which can befurther simplified as

a¼v?cavity

vJcavity

¼ a sinðffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiBcavity=Bsheath

qÞ ð5Þ

The loss cone pitch angle calculation uses a constant observedsheath field value of 55 nT up to 20:20 UT, and value of 44 nTafter 20:20 UT, the cavity field uses the 4 s averaged magneticfield observed by SC4.

Interestingly, the loss cone pitch angle curve is able to high-light many of the features of the electron differential fluxes atdifferent pitch angles: between 18:40 and 18:50 UT there is asymmetric distribution of particles at 10–601 and at 120–1701 inthe loss-cone. These loss cone particles can possibly originatefrom the magnetosheath due to reconnection tailward of the cuspand the anti-parallel counterpart can be the reflected population.Note that particles with finite parallel velocity in the magne-tosheath or magnetosphere are deeper in the loss cone. Thetrapped population in the cavity cannot directly originate frommagnetosphere or magnetosheath without non-adiabatic pitchangle scattering.

Between 18:55 and 19:17 UT SC4 is in the cavity and observesmost particles with very oblique pitch angles that could not havedirectly originated from the magnetosheath or magnetosphere.When studying the fluxes in the loss cone, there are higherparticle fluxes around 18:57 UT with pitch angles of 165–1701than with 25–101, indicating that these electrons are mostly

Page 8

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81 77

streaming out of the diamagnetic cavity. Between 19:09 and19:17 UT the parallel and anti-parallel population in the losscone is more symmetric. SC4 observes these counter-streamingparticles a few minutes before moving into the magnetosheath, soit may be quite close to the sheath–cavity boundary. This losscone part of the fluxes may also directly originate from themagnetosheath and is getting reflected at the lower altitudes.However, the majority of these fluxes are trapped and cannotoriginate from magnetosheath or magnetosphere directly withoutsome non-adiabatic process.

Waves with frequencies equal or higher than electron (proton)gyro-frequency, or strong gradients within electron (proton) gyro-radius in magnetic field or plasma parameters are required inorder to violate adiabaticity for electrons (protons) which could

Fig. 6. Test particle locations in simulation coordinates (1 length unit corresponds to 1R

times in a 3-D high-resolution MHD cusp model for adiabatic (a) and non-adiabatic (b)

(c) at 10 different times in the simulation box (a and b): particles are launched into t

velocity distributions (dark gray triangles). The cusp funnel is visible in x–z plane (b)

referred to the web version of this article.)

result either in pitch angle scattering of trapped cavity particlesinto loss-cone or similarly leakage of magnetosheath/magneto-sphere particles into the cavity. Unfortunately, for this event wedo not have simultaneous observations of the dayside magneto-sphere so thus we cannot confirm whether there are any suchwaves present that could allow leakage of magnetospheric parti-cles into DMC. The magnetic field power spectra from the STAFFinstrument (not shown) indicates that there is no significant wavepower at the vicinity of electron cyclotron frequencies from� 300 Hz to 2 kHz in the DMC or magnetosheath, so it is expectedthat wave acceleration for electrons during this event is minimal.However, the role that fluctuations in the DMC play in scatteringneeds to be further investigated and quantified. They can con-tribute to pitch angle scattering to the loss cone which might

E, magnetic dipole is aligned along z-direction) and particle energies at 10 separate

particles. The color code for both species shows particle energies and pitch angles

he simulation box with energies between 10 and few hundred eV with isotropic

. (For interpretation of the references to color in this figure legend, the reader is

Page 9

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–8178

explain how particles could leak into magnetosheath or iono-sphere from the DMC.

The fluxes during two sheath intervals at � 19 : 24 UT and at19:41–19:43 UT are mostly outside of the loss cone, althoughthere are also some fluxes in the parallel/anti-parallel range.These trapped particle fluxes are measured as SC4 dips back intothe cavity. At about 19:41 UT the solar wind dynamic pressurestarts do decrease from 4 nPa to 3.3 nPa for about 2 min. Thispressure change results in an expansion of the magnetospherewhich can explain why SC1, SC4 and possibly SC2 are sweptbriefly back to a more cavity-like plasma environment.

The above trapping condition calculation also applies for thepossible magnetospheric source where magnetospheric fieldstrength is the same or higher than the field measured in themagnetosheath which is usually the case.

Cluster Sc4 EFW, FGM & R2003, 18:30:00-

Ey<0

Cavity

Fig. 7. SC4 measurements of electric field components (top three panels), magnetic fie

(bottom panel) between 18:30 and 20:30 UT. The full electric field vector is derived as

5. Comparison with test particle simulations

The full description of the test particle simulations will begiven in another paper, here we will briefly illustrate the cap-ability of a local acceleration mechanism to energize particles inthe DMC. Fig. 6 shows test particle trajectories in a high resolu-tion 3-D cusp model in x�z and x�y planes in simulation units (1unit corresponds to 1RE), see model description in Adamson et al.(2011). Particle locations are shown for 10 separate times for twoparticle species: (1) particles with one hundredth of the protonmass to emulate more adiabatic behavior typical for electrons (a),and (2) particles with 16 times the proton mass to mimic non-adiabatic behavior typical for protons and oxygen ions (b). Thecolor code for both species shows particle energies and pitchangles (c) at 10 different times in the simulation box: e.g dark

APID 14th of February,20:30:00 UT

Ey>0

Cavity

ld strength (fourth panel), proton intensities (fifth panel) and electron intensities

suming E� B¼ 0.

Page 10

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81 79

gray triangles are 10 eV–3 keV particles with all the possible pitchangles; light grey squares correspond to 3–6 keV particles andwith pitch angles mostly between 501 and 1501; green asteriskscorrespond to � 10 keV particles with pitch angles mostlybetween 601 and 1401; yellow asterisks corresponds to 30–40 keV particles with � 901 pitch angles.

Particles are released into the simulation box with energiesbetween 10 and few hundred eV and with isotropic velocitydistributions (dark gray triangles). One can see that both typesof particles gain energy up to 40 keV in the cavity perpendicularto the magnetic field resulting in pitch angles of closely 901(yellow asterisks), which is in excellent agreement with theobserved pitch angles in the RAPID data shown in Fig. 5. Thisacceleration is due to gradients in reconnection ‘quasi’-potential.Because electrons are more adiabatic, they have more narrowlyconfined drift paths. Therefore the likelihood for spacecraft toencounter electrons that are energized in this potential field issmaller than to encounter protons whose less adiabatic motioncovers a larger region in the cavity. We think that this is thereason for the high-energy electron fluxes in the RAPID data beingsmaller and showing more variability than proton fluxes inthe DMC.

6. Electric field observations in the DMC

Fig. 7 shows 4 s averaged electric field measured by SC4between 18:30 and 20:30 UT together with total magnetic fieldand electron and proton intensities. Unfortunately, some data aremissing during magnetosheath intervals and at the end.

−50

5

Ey G

SM[m

V/m

]

0

50

|B| [

nT]

−50

5

Ey G

SM[m

V/m

]

0

50

|B| [

nT]

−5

05

Ey G

SM[m

V/m

]

18:30 19:00 10

50

100

|B| [

nT]

Cluster EFW, FGM 14th of Feb

sc1

sc2

sc3

Ey<0

Ey<0

Ey<0

Cavity

Cavity

Fig. 8. SC1, SC2 and SC3 measurements of the y-component of the electric field, Ey and

IMF is northward resulting on average negative Ey. During the second cavity interval,

The electric field components are highly variable in the cavitywith typical amplitudes of 4 mV/m and maximum amplitudes of8 mV/m. At the transition from magnetosphere into the cavity at� 18 : 30–18:45 UT where there are large changes in the high-energy particle intensities, there are also large changes in thex-component of the electric field.

Fig. 8 shows Ey and total magnetic field for SC1, SC2 and SC3 forthe same time interval as for SC4. At all spacecraft the y-componentof the electric field, Ey, is on average negative in the cavity duringnorthward IMF orientation, is zero when IMF Bz is zero and, becomespositive when IMF becomes negative. This indicates that thedominant contribution to Ey comes from the �v� B electric field.This electric field is strong evidence for a gradient in the reconnec-tion ‘quasi’-potential and particles may gain or lose energy depend-ing on their drift path with respect to this potential. Assuming anaverage Ey magnitude of 1 mV/m particles would need to drift 1RE inthis field in order to gain 6.4 keV in energy. Considering that DMCsizes of 6RE have been reported (Fritz et al., 2003) and assuming thatIMF orientation remains reasonably steady it would be possible forprotons (electrons) to gain � 40 keV if their average drift pathsremain parallel (anti-parallel) to this electric field.

In reality, there are always some fluctuations in the IMF whichcauses the reconnection site and thus also the gradients ofreconnection potential to change location. Therefore, it may bepossible to re-cycle particles several times through these ‘quasi’-potential fields which would likely result in greater energy gain.For the present event IMF remains northward (with slightfluctuations) for 1 h 40 min. Assuming roughly a drift velocity of100 km/s it would take a particle about 1 min to drift 1RE and gain6.4 keV in energy assuming that the particle remains trapped andis not lost into loss cone in this time-scale.

9:30 20:00 20:30

ruary, 2003, 18:30:00-20:30:00 UT

Ey>0

Ey>0

Ey>0

Cavity

Cavity

Cavity

magnetic field strength between 18:30 and 20:30 UT. During first cavity interval

IMF rotates southward resulting on average positive Ey.

Page 11

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–8180

Our test particle simulations suggest that the energy is gainedin the perpendicular direction and our observations have shownthat the loss-cone is very narrow for the electrons in weakmagnetic field inside the DMC which suggest that particles canbe trapped for a sufficiently long time for energization even above40 keV energies.

7. Discussion and conclusions

In this case study it has been demonstrated that DMC is filledwith high-energy electrons, protons and helium ions. Here wewill summarize the main findings:

1.

High-energy particle intensities are several orders of magni-tude lower in the magnetosheath and magnetosphere than inthe DMC.

2.

Intensities drop as a function of distance from the DMC:spacecraft deep in the cavity measure higher intensities thanthose close to DMC boundary in the magnetosheath, consistentwith the finite gyro-radius effect.

3.

Assuming adiabatic particle motion, energetic electron pitchangles in the cavity are not consistent with a magnetosphericor magnetosheath source.

4.

Observed electric fields in the cavity are consistent withelectric fields due to the reconnection ‘quasi’-potential in thehigh-resolution MHD cusp model.

5.

Test particle motions for adiabatic and non-adiabatic particlesare consistent with RAPID electron and proton data: electrondrift paths are more confined as they are nearly adiabatic andtheir intensities show more structure in the RAPID datacompared to proton intensities.

6.

Observed electron pitch angles in the cavity are consistentwith test-particle simulations: energy is gained in the perpen-dicular direction up to 40 keV generating large fluxes ofelectrons with 901 pitch angles.

7.

In order to violate the first adiabatic invariant for electrons,waves with frequencies comparable to electron cyclotronfrequency or gradients of the order of electron inertial lengthor gyro-radius would be required. The magnetic field powerspectra from the STAFF instrument (not shown) indicate thatthere is no significant wave power at the vicinity of electroncyclotron frequencies from � 300 Hz to 2 kHz in the DMC ormagnetosheath.

8.

Pitch angle scattering for energetic ions could be large,suggesting that energetic ions in the cavity could potentiallyoriginate from the quasi-parallel shock or magnetosphere.During this event the quasi-parallel bow shock maps tothe observed DMC (Trattner et al. 2011). However, the fact thatthe high energy proton fluxes drop orders of magnitude withinjust a few thousand km outward along the cavity-magnetosheathboundary normal appears difficult to reconcile with bow-shocksource: the width of the region in the magnetosheath that mapsto the quasi-parallel bow-shock has not yet been quantitativelydemonstrated.

9.

Throughout the magnetosheath, the high-energy proton fluxesremained low except during brief intervals when SC4 and SC1dropped back into cavity due to changes in solar wind dynamicpressure. However, the high-energy Oþ fluxes did not drop asmuch in the magnetosheath as the proton fluxes and weremostly at 60–1201 pitch angles, consistent with leakage fromthe DMC due to the large gyro radius of the oxygen ions.

The presented results for the case study strongly support theidea that electrons are locally accelerated in DMC: assuming theconservation for first adiabatic invariant, the observed pitch

angles in the cavity show that the loss-cone for electrons isnarrow and the vast majority of the high-energy electrons in thecavity cannot adiabatically originate from the magnetosheath ormagnetosphere. The small portion of the particles that willencounter the electron diffusion region are not obliged to con-serve their adiabatic invariance. In situ observations of electronpitch angles close to electron diffusion region in magnetotail havebeen reported by He et al. (2008). They show that at the vicinity ofthe null-point the electrons become bi-directional (pitch anglesclose to 01 and 1801). So even assuming that all of the electrons inmagnetospheric flux tube that gets reconnected would gothrough the electron diffusion region, would result in fluxes in01 and 1801—not in 901.

Note also that in the tail-lobe magnetosphere RAPID did notobserve any high-energy particles and when Cluster entered thefirst cavity interval (during northward IMF), the high-energyparticles were present. Our analysis of the reconnection intervalsin Nykyri et al. (2011a) showed that for the northward IMF the deHoffman–Teller frame velocities of the reconnected flux tubeswere consistent with the reconnection occurring tailward of thecusp, so it is not clear how the particles on closed field lines at thedayside magnetosphere would make it to the cusp when recon-nection opens up the field lines tailward of the cusp -yet the high-energy particles (electrons, protons, oxygen, helium) wereobserved both during northward and southward IMF.

For this event we did not have any simultaneous observationsof the closed field lines on dayside magnetosphere, so we cannotconfirm the magnitude of high-energy particle fluxes on thesefield lines and whether any plasma waves were present in thisregion to allow diffusion of high-energy magnetospheric particlesinto DMC.

The presented results do not exclude a bow-shock ion sourcebut set some strict constraints for this source that need to bebetter understood, e.g the width of the magnetosheath regionsurrounding the cavity that maps to the bow-shock, observationsof Oþ ions with 60–1201 pitch angles in the magnetosheath andthe presence of accelerated electrons outside of the DMCloss cone.

Although the present test-particle simulations demonstratethe acceleration only up to 40 keV energies is achieved viareconnection quasi-potential, it may be possible that particlesare re-cycled through this potential making energy gain up tohundreds of keV possible. Note that this mechanism works bothfor electrons and ions. It also has not yet been clearly demon-strated that ion or electron cyclotron wave activity has largeenough amplitudes in DMCs to efficiently accelerate ions andelectrons in the cavity. However, this still may be the case forsome events, and it could be that the net acceleration is acombination of both of these mechanisms.

The presented results strongly suggest that cusp diamagneticcavity may be an efficient particle-accelerator and heater ingeospace. Although, the presented results are only for a singleevent and thus more statistics on high-energy particle and waveproperties in the DMC and surrounding regions would berequired, a different case study by Walsh et al. (2010) show thatDMC is the main source for high-energy electrons for their event.Any magnetospheric or magnetosheath source must now demon-strate how the 901 electron pitch angles in the DMC are possible.In this paper we have shown using test particle simulations thatenergetic particles with 901 pitch angles can be created in theDMC due to acceleration in reconnection ‘quasi-potential’ inagreement with Cluster observations.

To consider broader impacts of our results we note that cusp-like structures are universal showing up in the coronal magneticfields in the Sun and other stars. Scaling of this heating mechan-ism in the cusp to typical coronal magnetic field properties

Page 12

K. Nykyri et al. / Journal of Atmospheric and Solar-Terrestrial Physics 87–88 (2012) 70–81 81

indicates that the heating mechanism presented in this papermay also contribute to the heating of the solar corona (Nykyriet al., 2009).

References

Adamson, E., Otto, A., Nykyri, K., 2011. 3-D mesoscale MHD simulations of a cusp-like magnetic configuration: method and first results. Annales Geophysicae 29(May), 759–770.

Asikainen, T., Mursula, K., 2005. Energetic particle fluxes in the exterior cusp andthe high-latitude dayside magnetosphere: statistical results from the Cluster/RAPID instrument. Annales Geophysicae 23 (September), 2217–2230.

Asikainen, T., Mursula, K., 2006. Reconnection and energetic particles at the edgeof the exterior cusp. Annales Geophysicae, 1949–1956.

Balogh, A., Carr, C.M., Acuna, M.H., Dunlop, M.W., Beek, T.J., Brown, P., Fornangon,K.-H., Georgescu, E., Glassmeier, K.-H., Harris, J., Musmann, G., Oddy, T.,Schwingenschuh, K., 2001. The Cluster magnetic field investigation: overviewof in-flight performance and initial results. Annales Geophysicae 19,1207–1217.

Blake, J.B., 1999. Comment on ‘‘cusp: a new acceleration region of the magneto-sphere’’ by J. Chen et al. Czechoslovak Journal of Physics 49, 675.

Burch, J.L., Reiff, P.H., Spiro, R.W., Heelis, R.A., Fields, S.A., 1980. Cusp regionparticle precipitation and ion convection for northward interplanetary mag-netic field. Geophysical Research Letters 7 (May), 393–396.

Chang, S., Scudder, J.D., Fennell, J.F., Friedl, R., Lepping, R.P., Russell, C.T., Trattner,K.J., Fuselier, S.A., Peterson, W.K., Spence, H.E., 2000. Energetic magnetosheathions connected to the Earth’s bow shock: possible source of cusp energeticions. Journal of Geophysical Research 105, 5471–5488.

Chang, S., Scudder, J.D., Fuselier, S.A., Fennell, J.F., Trattner, K.J., Pickett, J.S., Spence,H.E., Menietti, J.D., Peterson, W.K., Lepping, R.P., Friedel, R., 1998. Cuspenergetic ions: a bow shock source. Journal of Geophysical Research 25,3729–3732.

Chen, J., 2008. Evidence for particle acceleration in the magnetospheric cusp.Annales Geophysicae 26, 1993–1997.

Chen, J., Fritz, T.A., 2001. Energetic oxygen ions of ionospheric origin observed inthe cusp. Geophysical Research Letters 28 (April), 1459–1462.

Chen, J.S., Fritz, T.A., 1998. Correlation of cusp MeV helium with turbulent ULFpower spectra and its implications. Geophysical Research Letters 25,4113–4116.

Delcourt, D.C., Sauvaud, J., 1998. Recirculation of plasma sheet particles into thehigh-latitude boundary layer. Journal of Geophysical Research 103,26521–26532.

Delcourt, D.C., Sauvaud, J., 1999. Populating of the cusp and boundary layers byenergetic (hundreds of keV) equatorial particles. Geophysical Research Letters104, 22635–22648.

Escoubet, C.P., Fehringer, M., Goldstein, M., 2001. The cluster mission. AnnalesGeophysicae 19, 1197.

Frank, L.A., Ackerson, K.L., 1971. Observations of charged particle precipitation intothe auroral oval. Journal of Geophysical Research 76, 3612.

Fritz, T., Chen, J., Sheldon, R., Spence, H., Fennell, J., Livi, S., Russell, C., Pickett, J.,1999. Cusp energetic particle events measured by POLAR spacecraft. Physicsand Chemistry of the Earth C 24, 135–140.

Fritz, T.A., Chen, J., Siscoe, G.L., 2003. Energetic ions, large diamagnetic cavities,and Chapman–Ferraro cusp. Journal of Geophysical Research (Space Physics)108, 1028. doi:10.1029/2002JA009476.

Gustafsson, G., Andre, M., Carozzi, T., Eriksson, A.I., Falthammar, C.-G., Grard, R.,Holmgren, G., Holtet, J.A., Ivchenko, N., Karlsson, T., Khotyaintsev, Y., Klimov,S., Laakso, H., Lindqvist, P.-A., Lybekk, B., Marklund, G., Mozer, F., Mursula, K.,Pedersen, A., Popielawska, B., Savin, S., Stasiewicz, K., Tanskanen, P., Vaivads,A., Wahlund, J.-E., 2001. First results of electric field and density observationsby Cluster EFW based on initial months of operation. Annales Geophysicae 19(October), 1219–1240.

He, J., Zong, Q., Deng, X., Tu, C., Xiao, C., Wang, X., Ma, Z., Pu, Z., Lucek, E., Pedersen,A., Fazakerley, A., Cornilleau-Wehrlin, N., Dunlop, M.W., Tian, H., Yao, S., Tan,B., Fu, S., Glassmeier, K., Reme, H., Dandouras, I., Escoubet, C.P., 2008. Electrontrapping around a magnetic null. Geophysical Research Letters 351, L14104.doi:10.1029/2008GL034085.

Heikkila, W.J., Winningham, J.D., 1971. Penetration of magnetosheath plasma tolow altitudes through the dayside magnetospheric cusps. Journal of Geophy-sical Research 76, 883.

Kremser, G., Woch, J., Mursula, K., Tanskanen, P., Wilken, B., Lundin, R., 1995.Origin of energetic ions in the polar cusp inferred from ion compositionmeasurements by the Viking satellite. Annales Geophysicae 13, 595–607.

Lavraud, B., Fedorov, A., Budnik, E., Grigoriev, A., Cargill, P., Dunlop, M., R�eme, H.,Dandouras, I., Balogh, A., 2004. Cluster survey of the high-altitude cuspproperties: a three-year statistical study. Annales Geophysicae 22, 3009–3019.

Lavraud, B., Fedorov, A., Budnik, E., Thomsen, M.F., Grikoriev, A., Cargill, P.J.,Dunlop, M.W., Reme, H., Dandouras, I., Balogh, A., 2005a. High-altitude cuspflow dependence on IMF orientation: a 3-year cluster statistical study. Journalof Geophysical Research 110, A02209. doi:10.1029/2004JA010804.

Lavraud, B., R�eme, H., Dunlop, M.W., Bosqued, J., Dandouras, I., Sauvaud, J., Keiling,A., Phan, T.D., Lundin, R., Cargill, P.J., Escoubet, C.P., Carlson, C.W., McFadden,J.P., Parks, G.K., Moebius, E., Kistler, L.M., Amata, E., Bavassano-Cattaneo, M.,Korth, A., Klecker, B., Balogh, A., 2005b. Cluster observes the high-altitudeCUSP region. Surveys in Geophysics 26, 135–175.

Lin, Y., Wang, X.Y., Chang, S., 2007. Connection between bow shock and cuspenergetic ions. Geophysical Research Letters 34, 11107-þ .

Niehof, J.T., Fritz, T.A., Friedel, R.H.W., Chen, J., 2008. Interdependence of magneticfield and plasma pressures in cusp diamagnetic cavities. Geophysical ResearchLetters 35 (June), 11101-þ .

Nykyri, K., Otto, A., Adamson, E.T., Kronberg, E., Daly, P.W., Lavraud, B., 2009. Onthe origin of high-energy electrons in cusp diamagnetic cavities, AmericanGeophysical Union, Fall Meeting, abstract #SM31C-03.

Nykyri, K., Otto, A., Adamson, E., Dougal, E., Mumme, J., 2011a. Cluster observa-tions of a cusp diamagnetic cavity: structure, size and dynamics. Journal ofGeophysical Research 116, A03228.

Nykyri, K., Otto, A., Adamson, E., Tjulin, A., 2011b. On the origin of fluctuations inthe cusp diamagnetic cavity. Journal of Geophysical Research 116, A06208.

Reiff, P.H., Burch, J.L., Hill, T.W., 1977. Solar wind plasma injection at the daysidemagnetospheric cusp. Journal of Geophysical Research 82 (February),479–491.

R�eme, H., Aoustin, C., Bosqued, J.M., Dandouras, I., Lavraud, B., et al., 2001. Firstmultispacecraft ion measurements in and near the Earth’s magnetospherewith the identical Cluster ion spectrometry (CIS) experiment. Annales Geo-physicae 19, 1303–1354.

Trattner, K.J., Fuselier, S.A., Peterson, W.K., Chang, S., 1999. Comment on:‘‘correlation of cusp MeV helium with turbulent ULF power spectra and itsimplications’’. Geophysical Research Letters 26, 1361–1362.

Trattner, K.J., Fuselier, S.A., Peterson, W.K., Chang, S.-W., Friedel, R., Aellig, M.R.,2001. Origins of energetic ions in the cusp. Journal of Geophysical Research106, 5967–5976.

Trattner, K.J., Petrinec, S.M., Fuselier, S.A., Peterson, W.K., Friedel, R., 2010. Cuspenergetic ions as tracers for particle transport into the magnetosphere. Journalof Geophysical Research 115, 4219.

Trattner, K.J., Petrinec, S.M., Fuselier, S.A., Nykyri, K., Kronberg, E., 2011. Journal ofGeophysical Research 116 (A9), A09207. doi:10.1029/2011JA016617.

Vogiatzis, I.I., Sarris, T.E., Sarris, E.T., Santolık, O., Dandouras, I., Robert, P., Fritz,T.A., Zong, Q.-G., Zhang, H., 2008. Cluster observations of particle accelerationup to supra-thermal energies in the cusp region related to low-frequencywave activity—possible implications for the substorm initiation process.Annales Geophysicae 26 (March), 653–669.

Walsh, B.M., Fritz, T.A., Klida, M.M., Chen, J., 2010. Energetic electrons in theexterior cusp: identifying the source. Annales Geophysicae 28 (April),983–992.

Walsh, B.M., Fritz, T.A., Lender, N.M., Chen, J., Whitaker, K.E., 2007. Energeticparticles observed by ISEE-1 and ISEE-2 in a cusp diamagnetic cavity on 29September 1978. Annales Geophysicae 25 (January), 2633–2640.

Whitaker, K.E., Chen, J., Fritz, T.A., 2006. CEP populations observed by ISEE 1.Geophysical Research Letters 33 (December), 23105.

Whitaker, K.E., Fritz, T.A., Chen, J., Klida, M., 2007. Energetic particle sounding ofthe magnetospheric cusp with ISEE-1. Annales Geophysicae 25 (June),1175–1182.

Wilken, B., Daly, P.W., Mall, U., Aarsnes, K., Baker, D.N., Belian, R.D., Blake, J.B., Borg,H., Buchner, J., Carter, M., Fennell, J.F., Friedel, R., Fritz, T.A., Gliem, F., Grande,M., Kecskemety, K., Kettmann, G., Korth, A., Livi, S., McKenna-Lawlor, S.,Mursula, K., Nikutowski, B., Perry, C.H., Pu, Z.Y., Roeder, J., Reeves, G.D., Sarris,E.T., Sandahl, I., Søraas, F., Woch, J., Zong, Q.-G., 2001. First results from theRAPID imaging energetic particle spectrometer on board Cluster. AnnalesGeophysicae 19, 1355–1366.

Zhang, H., Fritz, T.A., Zong, Q., Daly, P.W., 2005. Stagnant exterior cusp region asviewed by energetic electrons and ions: a statistical study using clusterresearch with adaptive particle imaging detectors (RAPID) data. Journal ofGeophysical Research (Space Physics) 110, 5211.