Response of the Earth’s magnetosphere to changes in the ... · PDF fileResponse of the Earth’s magnetosphere to changes in the solar wind ... Response of the Earth’s magnetosphere

ARTICLE IN PRESS

1364-6826/$ - se

doi:10.1016/j.ja

�CorrespondSciences, Unive

Tel.: +1310 82

E-mail addr

(R.L. McPherr

Please cite th

Atmospheric

Journal of Atmospheric and Solar-Terrestrial Physics ] (]]]]) ]]]–]]]

www.elsevier.com/locate/jastp

Response of the Earth’s magnetosphere to changesin the solar wind

Robert L. McPherrona,b,�, James M. Weyganda, Tung-Shin Hsua

aInstitute of Geophysics and Planetary Physics, University of California, Los Angeles, CA, USAbDepartment of Earth and Space Sciences, University of California, Los Angeles, CA, USA

Accepted 27 August 2007

Abstract

The solar wind couples to the magnetosphere via dynamic pressure and electric field. Pressure establishes the size and

shape of the system, while the electric field transfers energy, mass, and momentum to the magnetosphere. When the

interplanetary magnetic field (IMF) is antiparallel to the dayside magnetic field, magnetic reconnection connects the IMF

to the dipole field. Solar wind transport of the newly opened field lines to the nightside creates an internal convection

system. These open field lines must ultimately be closed by reconnection on the nightside. For many decades, it was

thought that a magnetospheric substorm was the process for accomplishing this and that all magnetic activity was a

consequence of substorms. It is now recognized that there are a variety of modes of response of the magnetosphere to the

solar wind. In this paper, we briefly describe these modes and the conditions under which they occur. They include

substorms, pseudo-breakups, poleward boundary intensifications (PBI), steady magnetospheric convection (SMC),

sawtooth injection events, magnetic storms, high-intensity long-duration continuous AE activities (HILDCAAs), and

storm-time activations. There are numerous explanations for these different phenomena, some of which do not involve

magnetic reconnection. However, we speculate that it is possible to interpret each mode in terms of differences in the way

magnetic reconnection occurs on the nightside.

r 2007 Published by Elsevier Ltd.

Keywords: Solar wind coupling; Geomagnetic activity; Response modes

1. Preliminary remarks

In September 2006, the International Symposiumon Recent Observations and Simulations of theSun–Earth System (ISROSES) was held in Varna,

e front matter r 2007 Published by Elsevier Ltd.

stp.2007.08.040

ing author. Department of Earth and Space

rsity of California, Los Angeles, CA, USA.

5 1882; fax: +1 310 206 8042.

ess: [email protected]

on).

is article as: McPherron, R.L., et al., Response of the E

and Solar-Terrestrial Physics (2007), doi:10.1016/j.jastp

Bulgaria. The principal author of this paper wasasked to present the keynote lecture to introduce thetopics of the meeting and to set its tone. In responseto this invitation, he chose to describe the variousways in which the magnetosphere responds tochanges in the solar wind. It is now recognized thatin addition to substorms and storms there are otherforms of geomagnetic activity, including pseudo-breakups, poleward boundary intensifications(PIB), steady magnetospheric convection (SMC),substorm sequences, and storm-time activations.

arth’s magnetosphere to changes in the solar wind. Journal of

.2007.08.040

www.elsevier.com/locate/jastp

dx.doi.org/10.1016/j.jastp.2007.08.040

mailto:[email protected]

mailto:[email protected]

dx.doi.org/10.1016/j.jastp.2007.08.040

ARTICLE IN PRESSR.L. McPherron et al. / Journal of Atmospheric and Solar-Terrestrial Physics ] (]]]]) ]]]–]]]2

In this paper, we briefly describe each of thedifferent types of geomagnetic activity using someillustrative data. As we present each phenomenon,we speculate how it might be produced by magneticreconnection at various locations in the magneto-sphere.

2. Introduction

The magnetosphere is created by the solar wind.Its size, shape, and behavior depend on the proper-ties of the solar wind plasma and its imbeddedmagnetic field. The solar wind applies normal stressto the Earth’s magnetic field through dynamic,thermal, and magnetic pressure. Together, thesedetermine the gross size and shape of the magneto-sphere. Tangential stress is applied by two mainprocesses: viscous interaction and magnetic recon-nection. The viscous interaction is always presentand is responsible for transporting closed magneticflux tubes from the dayside to the nightside (Axfordand Hines, 1961). Magnetic reconnection alsooccurs most of the time, but its effect on themagnetosphere is dramatically different when theinterplanetary magnetic field (IMF) is northward,parallel to the dayside dipole field, than it is when itis southward and antiparallel (Dungey, 1961). Fornorthward IMF magnetic reconnection connectsinterplanetary magnetic field lines to open field linesin the tail lobes. The consequence is a circulation offlux tubes in the lobes, but no net change in theamount of lobe flux (Crooker, 1992). The circula-tion drives field-aligned currents that close throughthe ionosphere around the magnetic poles (Iijimaand Shibaji, 1987). These currents are so far fromauroral zone and lower latitude magnetometers thatthey cause only weak perturbations. Times ofnorthward IMF are therefore considered geomag-netically quiet.

The situation is different for southward IMF(McPherron, 1991). Interplanetary magnetic fieldlines connect to closed, dipole field lines near thesub-solar point of the magnetopause. These newlyopened field lines are transported over the polarcaps and added to open field lines already present inthe tail lobes. The lobe flux increases with timesqueezing the plasma sheet. The tail current movesearthward and magnetic reconnection begins in thetail returning flux to the dayside. If the IMFremains continuously southward the magneto-sphere responds in a variety of different ways:pseudo-breakups; substorms; PIB; a succession of

Please cite this article as: McPherron, R.L., et al., Response of the E

Atmospheric and Solar-Terrestrial Physics (2007), doi:10.1016/j.jastp

substorms; SMC; sawtooth injection events; andhigh-intensity long-duration continuous AE activ-ities (HILDCAAs). Such times are considereddisturbed.

A general conclusion from much of the recentwork on modes of magnetospheric response is thatthe response of the magnetosphere depends in anonlinear way on the level of solar wind driving(Pulkkinen et al., 2007). In addition, there isevidence that increased dynamic pressure can alterthe nature of the response (Lyons et al., 2005).Finally, there continues to be support for the ideathat sudden reductions in the convection electricfield (northward turnings of the IMF) and dynamicpressure pulses can trigger substorm onset (Hsu andMcPherron, 2002). In the following sections of thispaper, we briefly discuss the various distinct modesof magnetospheric response.

3. Magnetospheric response modes

3.1. Magnetospheric substorms

There have been numerous attempts to define amagnetospheric substorm. Akasofu (1964) definedit as:

The sequence of auroral events over the entirepolar region during the passage from auroralquiet through the various active phases tosubsequent calm is called an auroral substorm:it coincides with a magnetic (DP) substorm, withwhich it has some close relationships.

An early attempt to obtain a consensus definitionis reported in Rostoker et al. (1980). The authorsdescribe some of the signatures known to beassociated with substorms and argue that if thesesignatures are present then a substorm has occurred.Today an isolated substorm is known to consist ofthree phases: growth, expansion, and recovery.Features associated with the onset of the expansionphase are commonly used to establish the time lineof the substorm with the hope of determining thephysical processes responsible for the expansionphase. Some of the onset signatures include auroralbrightening, sudden onset of a negative bay nearmidnight, onset of a mid-latitude positive bay, aburst of Pi 2 pulsations, sudden dipolarization ofthe magnetic field, and particle injection at syn-chronous orbit.

The behavior of the AU and AL indices as afunction of time relative to auroral brightening is


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040

ARTICLE IN PRESS

0

100

200

300

AU

(n

T)

Substorm Drivers - AE Indices during 1997 - 1998

25%

50%

75%

90%

95%

# good / # total = 303 / 303

Percentiles = 25 50 75 90 95

-180 -120 -60 0 60 120 180

-500

-400

-300

-200

-100

Epoch Time (min)

AL (

nT

)

05%

10%

25%

50%75%

# good / # total = 303 / 303

Percentiles = 5 10 25 50 75

Fig. 1. Percentile lines of cumulative probability distributions (cdf) for AU and AL indices as a function of epoch time relative to an

auroral brightening identified in Polar satellite images (Liou et al., 2000). The shaded patches and the heavy lines define the quartiles of the

cdfs.

R.L. McPherron et al. / Journal of Atmospheric and Solar-Terrestrial Physics ] (]]]]) ]]]–]]] 3

illustrated in Fig. 1. The plot was constructed bysuperposing 303 sets of AU and AL tracescorresponding to brightening events identified inPolar spacecraft UV images by Liou et al. (2000). Itis obvious that the sudden development of awestward current is associated with the brightening.The majority of the events in this set are quiteweak as indicated by the shaded patches boundedby the upper and lower quartiles. Less than5% of the events in this list have a minimum ALless than �500 nT. The rapid change after thebrightening is 10–20min long. Recovery begins in40–60min.

If the IMF Bz is fluctuating with durations ofsouthward and northward field of the order of anhour, a sequence of substorms similar to an isolatedsubstorm develops. A typical substorm is about 3 hlong (McPherron, 1994), but substorm onsets canoccur quasi-periodically with an average timeseparation of 2.75 h, or randomly with an averageseparation about 5.76 h (Borovsky et al., 1993).During substorm sequences, the growth phase ofone substorm may begin even though signatures ofrecovery of the previous substorm are still inprogress.



A characteristic feature of clearly defined sub-storms is an ordered sequence of changes inconfiguration of the magnetosphere. These changescan be explained as a consequence of unbalancedreconnection. In the growth phase closed magneticflux from the dayside is opened and added to the taillobes. Closed magnetic flux in the plasma sheetmoves Sunward to replace the flux removed fromthe dayside. The amount of flux in the tail lobesincreases, causing the diameter of the auroral ovalto increase with time and the plasma sheet to thindue to compression. In the expansion phase closedfield lines in the plasma sheet begin to reconnectcreating shorter field lines on the earthward side ofthe x-line and a flux rope on the tailward side.Reconnection severs the last closed field lines at theedge of the plasma sheet releasing the flux rope thatis subsequently ejected down the tail. During theexpansion phase, lobe field lines reconnect addingclosed field lines to the plasma sheet that subse-quently move Sunward. In the recovery phase thex-line moves tailward expanding the plasma sheetbut still adding closed flux to the plasma sheet.Closed magnetic flux is returned to the dayside untilan equilibrium configuration is established.


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


3.2. Pseudo-breakups

In the original description of the auroral sub-storm Akasofu (1964) noted that occasionally anauroral arc would brighten and become active butwithin minutes would die down. He called suchevents ‘‘pseudo-breakups’’ after work originallydone by Elvey (1957). Elvey described these eventsin the following way:

If there are several arcs in the system, one ormore of the ones closer to the auroral zone maybreak up, with the pseudobreakup sometimesforming a rayed arc or breaking into bands andvarious rayed structures. These pseudo-breakupsappear to coincide with the minor bays that givethe irregularity to the main bay in the horizontalcomponent of the magnetic record.

Characteristics of these events reported byMcPherron et al. (1968) can be summarized in thefollowing way. A portion of an auroral arc brightensand field-aligned rays and undulations developoverhead. Frequently, the lower border of the arcturns red. Balloon measurements of X-rays show ashort-duration enhancement in electron precipita-tion. On the ground a burst of Pi 2 pulsations isrecorded in association with a weak (�100nT)negative bay. Within about 10min the arc becomesquiet and the various signatures disappear. Theseevents can occur anytime within the growth phasebut with increasing probability toward the expansiononset. Nakamura et al. (1994) made a detailed studyof several such events using synchronous particle andfield data, finding that such signatures as tail currentdiversion, dipolarization, and injection are alsoobserved. The authors concluded that the primarydifference between a pseudo-breakup and a substormexpansion is that pseudo-breakups have smaller scalethan expansions. Later, Nakamura et al. (2001a)examined the association of flow bursts in the tailduring pseudo-breakups, demonstrating that all flowbursts studied were associated with either pseudo-breakups, small poleward expansions, or PIBs.Nakamura et al. (2001b) show that flow burstsassociated with small expansions tend to occur closerto the Earth (X4�15Re) than those associated withPBIs. Kullen and Karlsson (2004) have done astatistical study and also conclude:

yThese results are in agreement with thescenario that pseudo breakups essentially arevery weak substorms.



The very beautiful example of a pseudo-breakuppresented in Fig. 2 was studied by Partamies (2003).The top panel of the figure shows all sky cameraimages of a counter-clockwise auroral spiral. Thelower panel shows the spiral projected into geo-graphic coordinates over Scandinavia. Vectors tothe west and north of the spiral are the measuredionospheric flow velocity. Analysis of data from amagnetometer array indicates that the spiral is thelocation of the outward current from a narrowsubstorm current wedge. The ground magneticperturbation from this current was only �50 nT,but low-energy electrons were injected at a synchro-nous orbit nonetheless.

A possible cause of pseudo-breakups is transient,localized reconnection inside the plasma sheet.The short duration and limited extent of theevent might then be a consequence of the reconnec-tion process not proceeding to the lobes. Ifthis is true then one might expect that such eventsproduce a flux rope trapped in the plasmasheet. However, it is also possible that the reconnec-tion proceeds to the lobe releasing the flux rope, butthen for unknown reasons the reconnection isquenched.

3.3. Steady magnetospheric convection

SMC or convection bay are names given to adisturbed state of the auroral region characterizedby the absence of the usual signatures of a substormexpansion (Pytte et al., 1975, 1978; Sergeev, 1977).SMC events can be identified by finding intervals ofsteady auroral zone magnetic disturbance producedby the DP-2 current system (2-celled convection) inthe absence of the DP-1 system (substorm currentwedge). SMC should have a steady polar cappotential, a steady polar cap (PC) magnetic index,a fixed diameter of the auroral oval, stable locationsof ionospheric particle boundaries, and a constantlobe field magnitude and diameter. There should beno sudden intensifications of the westward electrojetnear midnight, no mid-latitude positive bays, andno dipolarizations at synchronous orbit. An SMCevent is characterized by a double auroral oval(Elphinstone et al., 1995; Pulkkinen et al., 1995) inwhich the diffuse equatorward border is separatedfrom the poleward boundary by a darkerregion. Auroral activations frequently occur atthe poleward edge of the double oval. These includebrightening and decay of arcs, formation offolds and loops, equatorward drifting arcs, and


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040

ARTICLE IN PRESS

Fig. 2. All sky images of an auroral spiral at top were taken during a pseudo-breakup. A geographic projection of this spiral is plotted

below on a map of Scandinavia. Arrows indicate the flow velocity of ionospheric plasma as measured by radar (from Partamies, 2003).

R.L. McPherron et al. / Journal of Atmospheric and Solar-Terrestrial Physics ] (]]]]) ]]]–]]] 5

north–south-oriented streamers. These activationsare now called PIBs (see below).

Many of the features seen during an SMC aresimilar to those seen in the recovery phase of asubstorm (Sergeev et al., 1996) so it is important todistinguish between recovery phases and SMC.Usually this is done by requiring that any eventidentified as an SMC has a duration longer than the90min of a typical substorm recovery.

The average behavior of the AE index duringmore than 2400 SMC events has been obtained byMcPherron et al. (2005) and O’Brien et al. (2002).For a sample of AE to be classified as part of an



SMC these authors required that AE4200 nT,d(AL)/dt4�25 nT/min, and that these two condi-tions are continuously true for at least 90min. Thefirst point of a sequence satisfying these threecriteria was taken as the start of an SMC.McPherron et al. (2005) show that an SMC typicallybegins with a partial recovery from a precedingsubstorm. Most SMCs last only a few hours butoccasionally they persist for as long as 10 h. Themedian AE during SMC was 200 nT with both AUand AL approximately equal. About half of allSMCs terminate with another substorm expansion,while the other half vanish as the IMF turns slowly


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


northward. The IMF Bz during these SMCs wasabout �2.5 nT, with fluctuations about 25% smallerthan those found during all activity with AE4200 nT. The solar wind velocity is also low duringSMCs (o400 km/s) with weaker than normalfluctuations. The longest duration SMCs occurduring the lowest solar wind velocity with thesmallest fluctuations (1/2 normal amplitude).

From the beginning, the lack of change inmagnetospheric configuration during an SMC eventhas suggested that SMCs are caused by balancedreconnection much as Dungey (1961) originallysuggested. Subsolar reconnection strips flux off thedayside and induces a flow of plasma and flux intothe dayside merging region. Flux is added to the taillobe, but is reconnected in the plasma sheet at thesame average rate as on the dayside. Forces at thenightside x-line push plasma and flux Sunward.These flow around the Earth replacing the plasmaand flux stripped from the dayside. This is morelikely to work if the reconnection region on thenightside is close to the Earth and can quicklyadjust to changes in the dayside reconnection rate.This also assures that the pressure inside convectingflux tubes does not build up to large values. It is also

Ensemble of Sym

Sym

H (

nT

)

Epoch Tim

-4 -3 -2 -1 0

-100

-80

-60

-40

-20

0

20

Fig. 3. Average behavior of the 1-min Sym-H index over a 20-year inter

a superposed epoch analysis. A storm consists of an initial phase of

recovery followed by a slow recovery. Sym-H is below �100nT in less



likely to work better when the solar wind conditionsare steady so no adjustments are needed.

3.4. Magnetic storms

A magnetic storm is an interval of several days’duration during which the magnetic field measuredat low latitudes undergoes a rapid depressionfollowed by a slow recovery. The median behaviorof the Sym-H index, a 1-min proxy for Dst, ispresented in Fig. 3. The three phases of a storm arewell displayed in this superposed epoch analysis ofstorms from two solar cycles. The intense mainphase is defined by the interval of rapid decreaseand in typical storms is less than 12 h long.For example, the median duration of all stormmain phases in the space age is 11 h. The averagebehavior plotted in the diagram hides many com-plex details in the variety of development ofmagnetic storms. Some storms have much longermain phases and these storms dominate the decilesof the cumulative distribution plotted in the figure.Some have no initial phase, some have long andothers have short main phases, some have multipleminima.

25%

50%

75%

H for 1981-2001

e (days)

1 2 3 4

val. The time of minimum Sym-H was used as the reference time in

elevated Sym-H followed by a rapid decrease, and then a rapid

than 20% of the storms.


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040

ARTICLE IN PRESSR.L. McPherron et al. / Journal of Atmospheric and Solar-Terrestrial Physics ] (]]]]) ]]]–]]] 7

There are two main types of magnetic storms.Gradual commencement magnetic storms withoutobvious sudden commencements or initial phasesare created by corotating interaction regions(CIRs). These storms are generally weak but areoften of long duration. Constant magnetic activityin the recovery phase of these storms caused byAlfven waves (see discussion of HILDCAAs below),and equinoctial projection effects of the spiralIMF (Russell–McPherron effect) are responsiblefor the acceleration of relativistic electrons.These storms are most common in the decliningphase of the solar cycle when recurrent high-speedstreams are produced by the Sun. Suddencommencement storms are caused by the impactof an interplanetary coronal mass ejection (ICME)on the Earth. ICMEs are huge bubbles of plasmaejected from the Sun. Some of these containmagnetic clouds or flux ropes within which themagnetic field slowly and smoothly rotates from oneorientation to another. ICMEs are most frequentnear solar maximum. Many ICMEs travel atsupersonic speeds relative to the solar wind andare preceded by interplanetary shocks. A largevariety of storm signatures results from whetherthere is a shock, whether the compressed IMFbehind the shock is southward, whether the leadingedge of the flux rope is southward or northward,and so on. Storms produced by ICME are muchlarger that those produced by CIRs. The resultsplotted in Fig. 3 are averages over both types ofstorms.

3.5. Sawtooth injection events

Sawtooth injection events are identified by thecharacteristic wave form of energetic proton fluxesat synchronous orbit (Belian et al., 1995; Borovsky,2004; Reeves et al., 2003). These events occur whenthe solar wind driver is moderately strong andsteady (Pulkkinen et al., 2007) with Bz��7 nT andEy�3mV/m. Typical events consist of 3–8 teethwith a quasi-periodic spacing of �2.7 h. Thesynchronous injections are associated with strongstretching and dipolarization of the synchronousmagnetic field, with correlated southward turningsof the plasma sheet magnetic field in the middle tail,with auroral brightenings, with intensifications ofthe westward electrojet near midnight (Chao-Songet al., 2003) with mid-latitude positive bays and Pi 2pulsations. Henderson (2004) has shown that apreviously well-studied substorm interval, the



CDAW-9C study, was actually a sawtooth eventby today’s definition. The phenomena associatedwith a single tooth are identical to those seen duringisolated substorms except that a double auroral ovalis present as it is during SMC events (Hendersonet al., 2006). Also, the morning sector is filled witheastward-drifting omega bands, and at the end ofthe expansion north–south streamers develop in thepre-midnight sector. During sawtooth events, thereis an unusually strong current flowing around theevening–afternoon side of the Earth. Because thefield is strongly stretched, protons drift very rapidlygiving the appearance that they arrive simulta-neously over the entire nightside (Pulkkinen et al.,2006). Nonetheless, mid-latitude magnetic perturba-tions suggest that the substorms span larger regionsin local time than normal substorms (Clauer et al.,2006).

The result of a superposed epoch analysis of solarwind conditions associated with the onset ofsawtooth injections is presented in Fig. 4. Medianvalues of solar wind velocity, density, and dynamicpressure are typical of the normal solar wind.However, the strength of the magnetic field is nearlytwice as high as normal with median Bz reachingalmost 9 nT southward at onset. Because of the highmagnetic field strength, the solar wind Alfven Machnumber and plasma beta are somewhat lower thannormal. The results show a maximum in IMF fieldstrength at onset and a weak tendency to turn to thenorth at this time. The only parameter in the plotthat shows a dramatic change at sawtooth onset isthe polar cap index, which exhibits a suddenincrease, presumably due to substorm field-alignedcurrents.

Sawtooth injection events are caused by a steadysolar wind driver as are SMC events. However, thedriver is much stronger for sawtooth events. It isgenerally believed that these events are largesubstorms that occur quasi-periodically with aperiod characteristic of the state of the magneto-sphere. As the driver remains constant, energy isadded to the tail lobes faster than it can be removed.The configuration of the inner magnetospherebecomes distorted and a substorm expansionoccurs. Since there is a steady input the distortionrepeats and there is another expansion. This systemis also likely to be susceptible to external triggeringas conditions in the tail approach the criticalthreshold for a substorm expansion. This mayexplain the reported association between onset anddynamic pressure pulses.


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040

ARTICLE IN PRESS

-600 -300 0 300 600

300

400

500

600Velocity

-600 -300 0 300 600

0

5

10

15

20Mag(B)

-600 -300 0 300 600

-20

-10

0

10

20GSM-Bz

-600 -300 0 300 600

-6

-3

0

3

6GSM-Ey

-600 -300 0 300 600

0

5

10

15Density

-600 -300 0 300 600

0

2

4

6

Epcoh (min)

Dyn Pres

-600 -300 0 300 600

0

25

50

75

Epcoh (min)

Proton Vthermal

-600 -300 0 300 600

0

2.5

5

7.5

10

Epoch (min)

Alfven Mach #

-600 -300 0 300 600

0

0.5

1

1.5

Epoch (min)

Plasma beta

-600 -300 0 300 600

0

2

4

6

Epoch (min)

Polar Cap Index

Fig. 4. A superposed epoch analysis of solar wind conditions associated with �130 sawtooth injections. Epoch zero is the time at which a

sudden increase occurs in the flux of protons at synchronous orbit.

Fig. 5. A PBI event observed by meridian scanning photometers on December 23, 1996 at Rankin Inlet, Nunavut Territory, Canada, and

Gillam, Manitoba, Canada. Two red bars are the moon. Black arrow denotes magnetic midnight. A substorm expansion occurred in the

interval 0230–0320UT and produced a double oval. Equatorward extending PBIs occur between 0400 and 0600UT (adapted from (Zesta

et al., 2002).

R.L. McPherron et al. / Journal of Atmospheric and Solar-Terrestrial Physics ] (]]]]) ]]]–]]]8

3.6. Poleward boundary intensifications

PBIs often occur during and after a substormrecovery as well as in SMC and sawtooth events.A representative PBI event taken from the work ofZesta et al. (2002) is presented in Fig. 5. The PBIsare auroral intensifications that begin at the pole-ward edge of the aurora and then move equator-ward with time. In this event, a substorm growthphase occurred in the interval 0130–0230UT asevident in the equatorward motion of auroral arcs.The expansion phase defined by poleward expan-sion of the aurora occurred in two stages: the firstafter 0235UT and the second after 0304UT.This 101 expansion ended at 0320UT with theformation of the high-latitude portion of a double



oval. This bright form slowly moved equatorwardfor �25min and thereafter every 5–20min producedequatorward-moving bright auroral forms thatgenerally reached the bright region at the equator-ward edge of the oval. A second substorm expan-sion occurred at 0815UT. There were no auroralemissions poleward of the form producing the PBIs,indicating that they are formed at or close to theboundary between open and closed field lines.

This PBI event occurred during a long interval ofconstantly southward IMF that began around1140UT on December 22 and lasted until nearly0900UT on December 23. The minimum Bz duringthis interval was about �5 nT and the solar windspeed was �380 km/s. These are conditions that onemight expect, either SMC or sawtooth events, and


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


associated with them, PBI. Both GOES-8 and 9magnetometers in synchronous orbit saw verystrong dipolarization during the �0230UT expan-sion and a weaker one after 0800UT, but there is noindication of periodic substorms. Neither do avail-able proton detectors at synchronous orbit suggestsawtooth events. Because the IMF is southward, butthere are no substorm expansions, the interval from0320 to 0810UT might be classified as an SMCevent although the slow changes in the location ofthe polar cap boundary make this identificationproblematic.

Previous observations of PBI reviewed by Zestaet al. (2002) suggested that PBI are caused bytransient and localized reconnection at the distantx-line. They are associated with earthward-flowingbursty bulk flows (BBFs) in the tail. They primarilyoccur in the recovery phase of substorms, SMC, andsawtooth events. They can occur during quiet timesand in the expansion phase as well, but lessfrequently. Using all sky camera images as well asmeridian scanning photometers Zesta et al. (2002)identify at least three classes of PBI events, whichmay represent separate processes in the tail. Theseinclude auroral spirals on the poleward boundaryof the oval, east–west arcs moving equatorward,and narrow north–south auroral forms thatgrow equatorward. The arcs occur closer to duskwhile the north–south forms occur closer to mid-night. PBI rarely occur post midnight. Spirals andarcs may be created by processes other thanreconnection.

3.7. Storm-time activations

A long-standing argument concerns the possibledifference between isolated substorms and sub-storms during magnetic storms. Baumjohann(1996) used AMPTE IRM data to examine the lobefield during the two types of substorms. He foundthat only the storm-time substorms exhibit increaseand decrease of the lobe field. He speculated thatonly storm-time substorms are produced by a mid-tail x-line and that normal substorms are caused bysome other instability closer to the Earth. However,McPherron and Hsu (2002) used the same substormselection criteria and found that both types ofsubstorms exhibited an increase in lobe fieldstrength before an expansion and a decrease after.Storm-time substorms were simply larger thanisolated substorms. It should be noted that it isdifficult to identify and time substorm onset during



the main phase of a storm and some activations ofthe westward electrojet may not have been identifiedas substorms.

A study of storm-time activations by Pulkkinenet al. (2007) tried to determine whether all activa-tions are substorms. For a set of 150 suddenactivations of the westward electrojet (AL), abouthalf had an increase in field inclination at synchro-nous orbit although two-thirds showed a reductionafter onset (dipolarization). About half of the eventshad no particle injection at synchronous orbit.These observations suggest that there are manystorm-time activations that do not exhibit all theexpected signatures of a substorm.

A separation of the data into high and low Machnumber solar wind, keeping the median level ofdriving a constant, found no differences in themagnetospheric response. However, when theevents are separated by interplanetary electric fieldor dynamic pressure, differences are seen. Highelectric field causes stronger activity in all indices.However, when the index is normalized by thestrength of the driver Ey the magnetosphericresponse is proportionally weaker with a strongerdriver. Dynamic pressure also has a significanteffect on activity with a stronger response whenpressure is high. But when the data are normalizedby Ey there is no effect on the field inclination atsynchronous orbit, but high-pressure responses areproportionally weaker than low-pressure responses.

3.8. HILDCAAs

Many magnetic storms have a prolonged slowrecovery phase. During these recoveries theauroral electrojet index AE is often found to becontinuously disturbed for long intervals of time.Tsurutani and Gonzalez (1987) named such inter-vals HILDCAA. They found that these events wereassociated with long trains of Alfven waves in thesolar wind. Often the wave trains follow aninterplanetary shock, but sometimes they areassociated with high-speed streams without shocks.They postulated that these wave trains are respon-sible for magnetic reconnection each time they turnthe IMF southward and so produce a long sequenceof substorm activity. This led Tsurutani andGonzalez (1987) to conclude

It is thus surmised that the majority of allaurora1 activity (is associated with Alfvenicfluctuations of this type.


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


Fig. 6 demonstrates the validity of this premisefor the high-speed streams following thestream interface of a CIR. The map shows thedynamic cumulative probability distribution(cdf) of total wave power calculated accordingto the definition used by Roberts and Goldstein(1990):

E ¼ 12hdv2 þ db2

i.

In this formula, the magnetic field is normalizedby the ratio of 3-h averages of Alfven velocityto average magnetic field so as to have thesame units as velocity. Before the interface(zero epoch time) the median trace (labeled 50%)is about 50 nT. In the high-speed stream afterthe interface median wave power rises rapidly inless than a day, but takes many days to decay.We have also constructed cdfs of the AE indexand the normalized cross helicity defined by Robertsand Goldstein (1990). The pattern for AE isvirtually the same as the pattern for wave power.However, the normalized cross helicity fallsto less than 0.4 before the stream interface but isabout 0.9 after the interface. These results implythat strong Alfvenic fluctuations are correlatedwith elevated AE at least in the high-speed streamsthat create CIRs.

Epoch T

E (

(km

/s)2

)

CDFof E En

-5 -4 -3 -2 -10

200

400

600

800

1000

1200

1400

1600

1800

2000

Fig. 6. Total energy in solar wind fluctuations associated with high-



4. Discussion

In the preceding section, we have enumerated avariety of modes of response of the magnetosphereto the solar wind. These include PBI, pseudo-breakups, storm-time activations, isolated sub-storms, substorm sequences; SMC; magneticstorms, sawtooth injection events, and HILDCAAs.Undoubtedly these modes are frequently misidenti-fied when only a few data types are used in a study.For example, it is difficult to determine whether apseudo-breakup is different from a PBI. Similarly, itis difficult to rule out the possibility of smallsubstorms during an SMC, or to distinguish anSMC from a magnetic storm when solar winddriving is strong.

Nearly all geomagnetic activity is associated withintervals of southward IMF so magnetic reconnec-tion must be the underlying process that suppliessolar wind energy to the magnetosphere. Magneticfield lines opened on the dayside and transportedto the tail must eventually be returned to thedayside to maintain flux balance. Thus, reconnec-tion must also occur in the tail rather frequently.The question is whether all of these response modesdirectly involve magnetic reconnection, or whetherother processes use energy derived from reconnec-tion-driven convection to cause a phenomenon.

ime ((Days))

semble for 1995

0 1 2 3 4 5

speed solar wind following the stream interface within a CIR.


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


This seems to be the position of advocates of thecurrent disruption model for substorm onset. It isalso the predominant opinion regarding the cause ofmagnetic storms. It is continuous global convectionthat delivers particles to the inner magnetosphere,not discrete substorm expansions.

It seems very likely that both the waveform of thesolar wind driver and its strength determine whichresponse mode occurs. Isolated substorms aregenerally produced by a slowly fluctuating IMF,which first causes dayside reconnection and laternightside reconnection. SMC is produced by mod-erate and steady driving with balanced reconnectionon the day- and nightside. Sawtooth substormsequences are created by strong, steady driving.HILDCAAs occur when the IMF is continuouslydriven by interplanetary Alfven waves present inhigh-speed streams. Pseudo-breakups seem to beaborted or quenched substorms. PIBs are probablytransient and localized bursts of reconnection on amore distant x-line. Storm-time activations prob-ably include pseudo-breakups, PIBs, and weaksubstorms.

It is not known what effect the state of themagnetosphere has in producing a particularresponse mode. Mass loading of the plasma sheetby ionospheric oxygen may have a dramatic effect inthe tail, and eventually on the dayside whenconvection transports plasma to the dayside recon-nection region. Ionospheric conductivity is alsoextremely important. Additionally, it is not knownto what extent sudden changes in the strength ofconvection (rotations of the IMF) or dynamicpressure pulses may trigger a change in the stateof the magnetosphere. There are advocates of theposition that all substorms are triggered (Lee et al.,2004; Lyons, 1995, 1996) or that no substorms aretriggered (Freeman and Morley, 2005; Morley andFreeman, 2006).

Another important question concerning solarwind–magnetosphere coupling is what is the quan-titative relation between the strength of the driverand the amplitude of the response. Although notreviewed here, there have been numerous sugges-tions that the strength of a particular interactiondepends on prior activity. Thus, single-dip anddouble-dip storms might exhibit different degrees ofcoupling. Weak drivers are more efficiently coupledto the magnetosphere than strong drivers. Thedevelopment of a storm depends on whether therewas a long interval of northward IMF prior to thestorm. The strength of coupling depends on factors



other than the solar wind electric field such as thestrength of turbulence or the level of dynamicpressure.

It should not be forgotten that the solar cycle andgeometry of the IMF are also important and usuallyneglected factors in determining the strength ofsolar wind coupling and possibly the mode ofresponse. The solar cycle controls the typeof structures in the solar wind, the strength of theIMF, the polarity of the IMF, and the amount ofultraviolet impinging on the ionosphere (henceconductivity). Geometry of the Sun and Earthrotation and dipole axis are also very important.Magnetic activity tends to be stronger at equinoxthan at other times because of the Russell–McPher-ron effect, the Rosenberg–Coleman effect, the axialeffect, and the equinoctial effect. For example, thegeoeffectiveness of a CIR depends on the polarity ofthe IMF during the high-speed streams. If theseason is near equinox and the IMF obeys the rule‘‘spring to fall away’’ a HILDCAA is almost certainto occur because the IMF fluctuations due to Alfvenwaves are continuously biased southward and thesolar wind velocity is high.

5. Conclusion

It was once thought that substorms were thedominant mode of response of the magnetosphereand that storms were simply a superposition ofsubstorms. Now we have identified a variety ofresponse modes besides these two. It is apparentthat the type of response depends on the waveform of the solar wind electric field and its strength.There is considerable evidence that changes in theresponse mode can be triggered by sudden changesin either the electric field or dynamic pressure. Theparticular response mode probably also depends onthe internal state of the magnetosphere, which inturn depends on the recent history of the solar windinteraction as well as on many other factors.Considerable work needs to be done to furtherdescribe the characteristics of these response modesand to determine the physical processes thatproduce them. Quantitative relations between thesolar wind variables and various measures of theinternal state of the magnetosphere need to bedetermined as a function of the magnetosphericstate. Eventually it should become possible to pre-dict the probability of a particular type of activityand the probability that it will create a disturbanceof a given size. As our physical understanding


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


increases we should be able to transition fromempirical models to physics-based simulations ofthe system (O’Brien et al., 2002).

Acknowledgments

The authors would like to thank the NationalScience Foundation for its support through a GEMGrant ATM 02-1798 and a Space Weather GrantATM 02-08501. Support for this work was alsoprovided by a NASA Research at 1 AU Grant NG-04GA93G. R.L.M. also thanks the NSF Center forIntegrated Space Model (CISM) for support thougha subcontract on Grant ATM-0120950 to LASP inBoulder, CO. We also thank K. Liou for providinghis list of auroral brightenings. Data used in thiswork were obtained from a variety of sources,including the NASA NSSDC and WDC-B forGeomagnetism in Kyoto, Japan.

References

Akasofu, S.-I., 1964. The development of the auroral substorm.

Planetary and Space Science 12 (4), 273–282.

Axford, W.I., Hines, C.O., 1961. A unifying theory of high-

latitude geophysical phenomena and geomagnetic storms.

Canadian Journal of Physics 39, 1433–1464.

Baumjohann, W., 1996. Near-Earth plasma sheet dynamics.

Paper presented at Three-Dimensional Magnetosphere, D3.1

Symposium of COSPAR Scientific Commission D which was

held during the 30th COSPAR Scientific Assembly, Hamburg,

Germany, 11–21 November 1994.

Belian, R.D., Cayton, T. E., Reeves, G. D., 1995. Quasi-periodic,

substorm associated, global flux variations observed at

geosynchronus orbit. In: Ashour-Abdalla, M., et al. (Eds.),

Space Plasmas: Coupling between Small and Medium Scale

Processes, Geophysical Monograph Series, #86. American

Geophysical Union, Washington, DC, pp. 143.

Borovsky, J., 2004. Global Sawtooth Oscillations of the

Magnetosphere. EOS Transactions of the AGU 85 (49), 525.

Borovsky, J.E., Nemzek, R.J., Belian, R.D., 1993. The occur-

rence rate of magnetospheric-substorm onsets: random and

periodic substorms. Journal of Geophysical Research 98 (A3),

3807–3813.

Chao-Song, H., Reeves, G.D., Borovsky, J.E., Skoug, R.M., Pu,

Z.Y., Le, G., 2003. Periodic magnetospheric substorms and

their relationship with solar wind variations. Journal of

Geophysical Research 108 (A6) SMP19-11-12.

Clauer, C.R., Xia, C., Welling, D., DeJong, A., Henderson,

M.G., 2006. Characterizing the 18 April 2002 storm-

time sawtooth events using ground magnetic data.

Journal of Geophysical Research—Part A—Space Physics

111 (A4), 13.

Crooker, N.U., 1992. Reverse convection. Journal of Geophysi-

cal Research 97 (A12), 19363–19372.

Dungey, J.W., 1961. Interplanetary magnetic field and the

auroral zones. Physical Review Letters 6, 47–48.



Elphinstone, R.D., Murphree, J.S., Hearn, D.J., Cogger, L.L.,

Sandahl, I., Newell, P.T., Klumpar, D.M., Ohtani, S.,

Sanvaud, J.A., Potemra, T.A., Mursula, K., Wright, A.,

Shapshak, M., 1995. The double oval UV auroral distribu-

tion. I. Implications for the mapping of auroral arcs. Journal

of Geophysical Research 100 (A7), 12075–12092.

Elvey, C.T., 1957. Problems of auroral morphology. Proceedings

of the National Academy of Sciences 43 (1), 63–75.

Freeman, J.W., Morley, S.K., 2005. On the need for a solar wind

trigger. Eos Transactions of the AGU 86 (52) Fall Meeting

Supplement, Abstract SM31A-0381.

Henderson, M.G., 2004. The May 2–3, 1986 CDAW-9C interval:

a sawtooth event. Geophysical Research Letters 31 (11), 5.

Henderson, M.G., Skoug, R., Donovan, E., Thomsen, M.F.,

Reeves, G.D., Denton, M.H., Singer, H.J., McPherron, R.L.,

Mende, S.B., Immel, T.J., Sigwarth, J.B., Frank, L.A., 2006.

Substorms during the 10–11 August 2000 sawtooth event.

Journal of Geophysical Research 111 (A6), 1–32.

Hsu, T.S., McPherron, R.L., 2002. An evaluation of the

statistical significance of the association between northward

turnings of the interplanetary magnetic field and substorm

expansion onsets. Journal of Geophysical Research 107

(A11), 15pp.

Iijima, T., Shibaji, T., 1987. Global characteristics of northward

IMF-associated (NBZ) field-aligned currents. Journal of

Geophysical Research 92 (A3), 2408–2424.

Kullen, A., Karlsson, T., 2004. On the relation between solar

wind, pseudobreakups, and substorms. Journal of Geophysi-

cal Research 109 (A12), 20.

Lee, D.Y., Lyons, L.R., Yumoto, K., 2004. Sawtooth oscillations

directly driven by solar wind dynamic pressure enhancements.


Liou, E., Meng, C.I., Newell, P.T., Takahashi, K., Ohtani, S.I.,

Lui, A.T.Y., Brittnacher, M., Parks, G., 2000. Evaluation of

low-latitude Pi2 pulsations as indicators of substorm onset

using Polar ultraviolet imagery. Journal of Geophysical

Research 105 (A2), 2495–2505.

Lyons, L.R., 1995. A new theory for magnetospheric substorms.


Lyons, L.R., 1996. Evidence suggests external triggering of

substorms. EOS Transactions of the AGU 77 (9), 88–89.

Lyons, L.R., Lee, D.Y., Wang, C.R., Mende, S.B., 2005. Global

auroral responses to abrupt solar wind changes: dynamic

pressure, substorm, and null events. Journal of Geophysical

Research—Part A—Space Physics 110 (A8), 15.

McPherron, R.L., 1991. Physical processes producing magneto-

spheric substorms and magnetic storms. In: Jacobs, J. (Ed.),

Geomagnetism. Academic Press, London, pp. 593–739.

McPherron, R.L., 1994. The substorm growth phase. In: Kan,

J.R., et al. (Eds.), Substorms 2. Geophysical Institute,

University of Alaska, Fairbanks, AK, pp. 99775–100800.

McPherron, R.L., Hsu, T.S., 2002. A comparison of sub-

storms occurring during magnetic storms with those occurring

during quiet times. Journal of Geophysical Research 107

(A9), 16pp.

McPherron, R.L., Parks, G.K., Coroniti, F.V., Ward, S.H., 1968.

Studies of the magnetospheric substorm, II. Correlated

magnetic micropulsations and electron precipitation occur-

ring during auroral substorms. Journal of Geophysical

Research 73 (5), 1697–1713.

McPherron, R.L., O’Brien, T.P., Thompson, S.M., 2005. Solar

wind drivers for steady magnetospheric convection. In: Lui,


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040


A.T.Y., et al. (Eds.), Multiscale Coupling of Sun–Earth

Processes. Elsevier, Amsterdam, pp. 113–124.

Morley, S. K., Freeman, M. P., 2006. On the association between

northward turnings in the IMF and substorm onset.

Presented at International Symposium on Recent Observa-

tions and Simulation of the Sun–Earth System (ISROSES),

Varna, Bulgaria, September 17–22.

Nakamura, R., Baker, D.N., Yamamoto, T., Belian, R.D.,

Bering III, E.A., Benbrook, J.R., Theall, J.R., 1994. Particle

and field signatures during pseudobreakup and major

expansion onset. Journal of Geophysical Research 99 (A1),

207–221.

Nakamura, R., Baumjohann, W., Brittnacher, M., Sergeev, V.A.,

Kubyshkina, M., Mukai, T., Liou, K., 2001a. Flow bursts

and auroral activations: onset timing and foot point location.


Nakamura, R., Baumjohann, W., Schodel, R., Brittnacher, M.,

Sergeev, V.A., Kubyshkina, M., Mukai, T., Liou, K.,

2001b. Earthward flow bursts, auroral streamers, and small

expansions. Journal of Geophysical Research 106 (A6),

10791–10802.

O’Brien, T.P., Thompson, S.M., McPherron, R.L., 2002. Steady

magnetospheric convection: statistical signatures in the solar

wind and AE. Geophysical Research Letters 29 (7).

Partamies, N., 2003. A pseudo-breakup observation: localized

current wedge across the postmidnight auroral oval. Journal

of Geophysical Research 108 (A1) SIA4-1-10.

Pulkkinen, T.I., Baker, D.N., Pellinen, R.J., Murphree, J.S.,

Frank, L.A., 1995. Mapping of the auroral oval and

individual arcs during substorms. Journal of Geophysical

Research 100 (A11), 21987–21994.

Pulkkinen, T.I., Ganushkina, N.Y., Tanskanen, E.I., Kubyshki-

na, M., Reeves, G.D., Thomsen, M.F., Russell, C.T., Singer,

H.J., Slavin, J.A., Gjerloev, J., 2006. Magnetospheric current

systems during stormtime sawtooth events. Journal of

Geophysical Research 111 (A11), 1–13.

Pulkkinen, T.I., Partamies, N., McPherron, R.L., Henderson,

M.G., Reeves, G.D., Thomsen, M.F., Singer, H.J., 2007.

Statistical analysis of stormtime activations and sawtooth

events. Journal of Geophysical Research 112 (A01205).



Pytte, T., McPherron, R.L., Hones Jr., E.W.W.H.I., 1975. The

distinction between substorms and convection-driven auroral-

zone negative bays (abstract SM114). EOS Transactions of

the AGU 56 (6), 435.

Pytte, T., McPherron, R.L., Hones Jr., E.W., West Jr., H.I., 1978.

Multiple-satellite studies of magnetospheric substorms: dis-

tinction between polar magnetic substorms and convection-

driven negative bays. Journal of Geophysical Research 83

(A2), 663–679.

Reeves, G.D., Henderson, M.G., Skoug, R., Thomsen, M.,

Borovsky, J.E., Funsten, H.O., Brandt, P.C.S., Mitchell, D.J.,

Jahn, J.-M., Pollock, C.J., McComas, D., Mende, S.,

2003. IMAGE, POLAR, and geosynchronous obser-

vations of substorm and ring current ion injection. In:

Sharma, A.S., et al. (Eds.), Disturbances in Geospace:

The Storm–Substorm Relationship, Geophysical Monograph

142. American Geophysical Union, Washington, DC,

pp. 91–101.

Roberts, D.A., Goldstein, M.L., 1990. Do interplanetary Alfven

waves cause auroral activity? Journal of Geophysical Re-

search 95 (A4), 4327–4331.

Rostoker, G., Kawasaki, K., Akasofu, S.I., Foster, J., Green-

wald, R.A., Kamide, Y., Lui, A.T.Y., McPherron, R.L.,

Russell, C.T., 1980. Magnetospheric substorms—definition

and signatures. Journal of Geophysical Research 85 (A4),

1663–1668.

Sergeev, V.A., 1977. On the state of the magnetosphere during

prolonged periods of southward oriented IMF. Physical and

Solariterrestrial Potsdam 5, 39.

Sergeev, V.A., Pellinen, R.J., Pulkkinen, T.I., 1996. Steady

magnetospheric convection: a review of recent results. Space

Science Reviews 75 (3–4), 551–604.

Tsurutani, B.T., Gonzalez, W.D., 1987. The cause of high-

intensity long-duration continuous oi4AEo/i4 activity

(HILDCAAs): interplanetary Alfven wave trains. Planetary

and Space Science 35 (4), 405–412.

Zesta, E., Donovan, E., Lyons, L., Enno, G., Murphree, J.S.,

Cogger, L., 2002. Two-dimensional structure of auroral

poleward boundary intensifications. Journal of Geophysical

Research 107 (A11) SIA61-20.


.2007.08.040

dx.doi.org/10.1016/j.jastp.2007.08.040

Documents

Response of the Earth’s magnetosphere to changes in the ... · PDF fileResponse of the Earth’s magnetosphere to changes in the solar wind ... Response of the Earth’s magnetosphere