What is a Global Auroral Substorm?

WHAT IS A GLOBAL AURORAL SUBSTORM?
R. D. Elphinstone, J. S. Murphree, and L. L. Cogger
Institute for Space Research, University of Calgary,
Calgary, Alberta, Canada
Abstract. The departure of the aurora from quiet lev- ing to explain the mechanism of the onset of this pheels
in a dynamic manner constitutes some type of auroral nomenon. There is, however, no single theory which
"breakup" event. Research into the auroral breakup stands out as clearly explaining the wide range of active
predates the International Geophysical Year (1957/ auroral phenomena. A synthesis which combines these
1958). This feature of the aurora, and the later, more theories and allows them to each explain individual
global concept of the auroral substorm, has become a aspects of the problem appears to be required. This has
focus for much of the auroral research that occurs today. led to a new way of understanding the active aurora as a
New instrumentation and global collaborations continue set of processes or modules which occur either coupled
to refine our knowledge of the substorm process and together or independent of one another to form a parhow
it proceeds in the ionosphere. In particular, global ticular event. This view represents a fundamental deparauroral
imaging has advanced our understanding of the ture from the view of the substorm as a single unchangdynamics
of the process and has given us the ability to ing entity. Auroral activity can rather be thought of as
put localized observations into a global perspective. Fun- the earthward end of a diverse set of ionospheric and
damentally new cycles of auroral activity are now under- magnetospheric processes which couple together to
stood to exist, and this has provided a means by which form different cyclical patterns. A symbolic representaauroral
activity can answer questions about magneto- tion of this modularization is presented to simplify fuspheric
substorm dynamics. Along with this wealth of
observations has come a wide range of theories purportture
schematics of large-scale auroral dynamics.
1. INTRODUCTION
Many people who live in the northern regions and are
outdoors at night are familiar with the beautiful movements
and patterns known as the "northern lights." The
flickering, changing curtains and arches of light, which
we can see with the naked eye, represent only the brightest
aspects of the phenomena named after Aurora, the
Roman goddess of dawn. These spectacular effects are a
dynamic perturbation to a more steady and quiet system
tion from excited states attributable to the collisions
between these (precipitating) electrons and protons
(with typical energies of a few keV) (an electron volt is
the energy unit associated with moving an electron
across a potential difference of 1 V) and atmospheric
particles, particularly atomic and molecular oxygen and
nitrogen. The bulk of these collisions occur in the ion-
osphere between about 80 and 300 km, and the resulting
of aurora and represent the visible manifestation of what
auroral intensity reflects at any given point the precipitation
rate (see work by Vallance Jones [1974, 1991] and
Rees and Lummerzheim [1991] for detailed descriptions
is commonly called an auroral substorm.
of the processes involved and the types of optical emis-
This light show is the earthward end of a complex sions).
interaction of fields and particles which begins at the This production of the aurora, with its variety of
Sun. Ions, electrons, and magnetic fields move outward colors (red, blue, and green), spatial distribution, and
from the Sun in what is called the solar wind and interact dynamics provides an essentially instantaneous record of
with the Earth's magnetic field to form the terrestrial plasma processes in the Earth environment. Thus the
magnetosphere. The distortions of the magnetic field aurora that one sees in the northern (and southern)
which result can be associated with large-scale current regions gives us a means to indirectly observe processes
systems which flow throughout the system, some of operating in and on the magnetosphere. To do this, one
which reach the ionosphere of the Earth along the needs to understand the "mappings" between the ionomagnetic
field lines. These Birkeland field-aligned cur- sphere and the magnetosphere and to know what the
rents can carry about 1 MA of current into the iono- relevant patterns in the auroral distribution are. At
sphere, and the associated power consumption is equiv- present, the auroral distribution is the only means at our
alent to that required by a large number of cities (10 s disposal by which the large-scale magnetospheric struc-
MW). A side effect of this is the aurora, which is radia- ture can be evaluated in a nearly instantaneous manner.
Copyright 1996 by the American Geophysical Union.
8755-1209/96/96 RG-00483515.00
ß 169e
Reviews of Geophysics, 34, 2 / May 1996
pages 169-232
Paper number 96RG00483

1 70 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
The magnetospheric cavity encloses a volume of the which cannot be simulated by magnetohydrodynamics.
order of 3 x 105 (Rr) 3 (Rr is an Earth radius, or about A prime example of this may be the onset mechanism for
6371 km), and we have throughout this volume only a the auroral substorm. For this, models which couple
scattering of satellites by which to establish the nature of across various scales and involve a hierarchy of models
its dynamics and form. One might compare this with a may need to be developed [e.g., Raeder et al., 1996].
similar problem of studying the Earth's oceans with a Unfortunately, predictions of substorms and their effects
few centimeter-long probes scattered at varying loca- will remain elusive until a detailed physical mechanism
tions and depths. Although the analogy is valid only to a which is viable is found. This is likely to occur mainly on
limited degree, it does help to illustrate the problems the basis of observational constraints.
with which the magnetospheric physicist is faced. Be- This paper will attempt to outline some of the history
cause of this the aurora, if used appropriately, can be of concerning auroral morphology' and what is called the
great aid to researchers in their studies of the magneto- auroral substorm and then describe how this informasphere
as we enter into a new era in space physics with tion might be of use to the magnetospheric physicist
the launch of new spacecraft and international ground studying an equivalent phenomenon known as the magcollaborations.
netospheric substorm (see excellent reviews by McPher-
The increasing use of space for remote sensing and ron [1979] and Fairfield [1992]). The rest of this first
communication has led to a need to understand and section will be devoted to the historical aspects of the
predict what has come to be known as "space weather." aurora and the concept of the substorm. The researcher
Lindqwister et al. [1995] believe that space weather dis- familiar with the history of the auroral substorm could
turbances may cost in excess of $100 million per year. pass over sections 1.1 and 1.2. Since an integral part of
Geosynchronous spacecraft can suffer a variety of prob- any review of the aurora should include information on
lems ranging from a loss of orientation (magnetically imaging from both the ground-based and satellite vieworiented
satellites) to ion penetration, surface charging, points, sections 1.3 and 1.4 cover some of the existing
and deep dielectric charging of satellite components imaging techniques and instrumentation. Obviously, it is
[Wilkinson, 1995]. To avoid such problems, engineers not possible to show image data from all of the imporrequire
a good knowledge of the "radiation climatology" tant instruments which have contributed to this field of
in the near-Earth environment, and satellite operators research. A brief description will also be given of how
need forecast tools which will enable them to take pre- the auroral substorm relates to the solar wind as well as
ventive measures against magnetic disturbances, sub- to a few of the well-known features of ionospheric curstorms,
and solar events. To accomplish this task, re- rents and convection (section 1.5).
searchers are increasingly looking to the aurora to Auroral substorm research abounds in terminology
understand and predict the dynamics of these events. which can overwhelm (and confuse) even veteran au-
Normally, confined to higher latitudes, the aurora's roral researchers. Following section 5.2 is a glossary
appearance over large portions of the Earth signals comprising a few of the necessary terms used in this
significant disturbances which can also impact human paper. The terms were selected mainly to aid the magactivities
on the ground. Electrical power utilities on the netospheric researcher in identifying the auroral forms
ground can sustain damage from substorm-related, described in this paper. Readers familiar with this maground-induced
currents which cause transformer prob- terial could go directly to section 2, which outlines an
lems resulting in power blackouts. These companies emerging modular concept of auroral and magnetoneed
forewarning of upcoming events to institute special spheric activity which allows auroral features to be more
operating procedures. With more than 100 communica- easily incorporated into magnetospheric physics. Section
tion satellites in geosynchronous orbit today and hun- 2 emphasizes the phenomenological auroral aspects of
dreds more planned for, space physics is beginning to the modular view since at this level it is evident that
take on a more applied approach with the hope of modularization is necessary. Such phenomenological
producing space-forecasting tools for satellite operators. studies provide nearly the only means to probe the
Some such tools already exist which attempt to pre- global dynamics of the system. It was decided that some
dict large-scale changes to the ring current based on details concerning each auroral module should be given
changes in the solar wind (e.g., Hudson et al. [1995] and since different modules will be of interest to different
the review by Feldstein [1992]). Unfortunately, our abil- researchers. Readers should select the particular auroral
ity so far to successfully predict new events is rather poor modular elements applicable to their work. Interdiscipli-
(see the review byJoselyn [1995]). Conditions in the solar nary readers might prefer to only read the introductory
wind can also be used to derive changes to magnetic remarks to this section and move on to section 3. In
indices more directly related to changes in the aurora order to make these substorm modules of use to the
[Vassiliadis et al., 1995, and references therein]. As well, space physics community a mapping framework for rethere
are large-scale magnetohydrodynamic (MHD) lating them to their magnetospheric counterparts is
simulations which can reproduce many observable pa- given in section 3.
rameters [e.g., Fedder et al., 1995]. Unfortunately, space Over the years, substorm models have developed to
plasma physics encompasses many scale sizes, some of explain various aspects of the magnetospheric (and oc-

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 1 71
casionally auroral) substorm. Section 4 is an attempt to head). A connection was also made between magnetic
evaluate some of these based on the auroral patterns perturbations and optical aurora. This dates back at least
and dynamics. Section 5 puts the results from sections 2, to compass and naked eye work done by Hiorter and
3, and 4 together to illustrate how the modular view of Celsius in 1741 [Brekke and Egeland, 1983; Buchert et al.,
auroral and magnetospheric activity can be used to un- 1990].
derstand systematic cycles in the activity. In section 5, By the 1930s it was realized that there was a typical
processes are connected with their phenomenological progression of the aurora from a "quiet" diffuse glow to
counterparts, and a symbolic representation of the an dynamic phase of the aurora known as the auroral
modularization is introduced as a type of "shorthand" to
breakup [see Stagg, 1937]. As described by Kern and
describe more complex activity. The paper ends with
Vestine [1961], these early authors understood that a
questions for which future studies may yield more combreakup
event is preceded by some interval (less than an
plete answers.
hour) in which a glow is seen. The equatorward portion
Finally, it should be noted that a great amount of
recent work has been done in this field and that it would of this glow gradually brightens and forms an arc or arc
system in the E region of the ionosphere. After some
be impossible to include it all in this review paper. An
attempt has been made to include a reasonable sampling
time this system develops flutes, and the arc system
of material directly relevant to the thesis of the paper,
which subsequently develops exists in the F region of the
but undoubtedly some of the important work done by ionosphere. This then "breaks up" and eventually develthe
many researchers over the years will be accidently ops into auroral patches. This can occur more than once
passed over. It was decided to make this work of suffi- during a night. This description is one of the best that
cient length to enable a magnetospheric researcher to be can be found in auroral literature of the development of
able to find out at least some information concerning auroral features prior to breakup. The reader is referred
most of the auroral substorm patterns. The authors feel to work by Hewson [1937] for further information rethat
this will fill an important gap which exists in today's garding observations and theoretical models for the auliterature.
rora during the early part of this century.
More recently, the most significant advance took
1.1. Auroral History
place when it was realized that on a global scale the
The term aurora covers a rather broad range of aurora resembled an oval shape which lies at higher
spatial and temporal scales. Small-scale auroral arcs may magnetic latitudes on the dayside portion of the ionobe
only 50 rn wide and can flicker with periods much less sphere than on the nightside portion [Feldstein, 1960,
than 1 s. Other features such as the auroral oval may 1963]. Thus, when studying the auroral distribution, one
change size slowly but are rather constant aspects of the generally shows data in a Sun-fixed system based at
aurora and span thousands of kilometers in both the auroral heights beneath which the Earth's surface ronorthern
and southern polar regions. The auroral subtates.
This auroral asymmetry and the so-called auroral
storm has an intermediate-scale size and generally lasts oval are fundamentals of the auroral distribution which
of the order of 1-3 hours, with its most obvious effects
give clues to the configuration of the magnetosphere and
appearing in the nightside polar regions. Aspects of this
to the magnetospheric source regions of the aurora.
latter feature are what catches people's attention when
Today, specially designed coordinate systems remove
they notice the bright auroral forms visible to the naked
the asymmetries of the Earth's internal magnetic field
eye at night.
Although at first glance a seemingly random phenomand
convert the "auroral oval" into a circular form (see
enon, both in time and space, the aurora follows a
work by Heam et al. [1993, and references therein] for a
reasonably well defined spatial pattern, at least on a discussion of these coordinate transformations).
large scale. Perhaps the first progress in a systematic way At the same time the auroral oval was identified, the
came when Wargentin [1752] studied simultaneous au- concept of an auroral substorm was also introduced
roral observations at widely separated locations and [Akasofu, 1964]. This established that the auroral
concluded that the aurora made a ring about a location breakup discovered previously was not just a local prosomewhat
offset from the geographic pole, an observa- cess but extended over large distances. The studies mention
which we now know applies to both the northern tioned above were constrained to using sufficiently inand
southern hemispheres. Hansteen [1827] later decided
that this ring was centered about what would now
tense visible emissions from the aurora (generally
greater than 1 kR, equal to 103 photons cm -2 s -, where
be termed the magnetic dip pole and drew perhaps the a rayleigh (R) is a unit of photon surface brightness,
first realistic schematic of the large-scale auroral distri- described in more detail by Vallance Jones [1974]) and to
bution (see work by Brekke [1984] for a more detailed combining individual measurements into a statistical
history). Nordenskiold [1880, 1881] expanded on this pattern. It is only recently that satellite imagers have
concept and described two concentric rings centered enabled researchers to obtain large-scale nearly instanabout
the pole and named the phenomenon the "auroral taneous views which include the weaker emission feaglory"
(like the halo, or glory, surrounding a saint's tures (the so-called diffuse aurora).

1 72 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
1.2. Substorm Research
scale spirals or surge forms). It was also known that
One of the problems one runs across in substorm auroral forms drifted in the opposite sense to the current
research is that large data sets from around the world flow in the ionosphere and that this relation breaks down
are required to assemble a coherent picture of the over- near the time of auroral breakup. North-south structures
all process. A given auroral event will likely have only several hundred kilometers long linked two latitudinally
certain data available, while a separate event may well separated auroral systems near the location (at about
have supporting data of an entirely different nature. As midnight) where the drifts changed from east to west
a result, research in this field has a tendency to be [Davis, 1962]. These features could have been a variety
founded on case studies which may be biased toward the of forms, some of which are outlined in section 2. Arcs
viewpoint the researcher is attempting to prove and near the magnetic pole were known to be oriented
constrained by the measurement limitations inherent in differently than those in the auroral zone. These features
the data set(s) used. A consequence of this is that are now an important part of auroral research and are
researchers, when confronted with examples of an au- called by a variety of names, including Sun-aligned arcs,
roral distribution which rapidly changes its intensity high-latitude polar arcs, theta auroras, teardrops, and
(loosely termed "auroral activations") and pattern, can horse collar auroras.
disagree as to its significance with respect to the sub- Probably, the first description of how auroral dynamstorm
process. Hence a variety of terminologies has ics at one location were related to those at another,
arisen to describe auroral activations, some of which are widely separated point was made byAkasofu [1963], who
synonymous with an auroral substorm, but others of developed Figure la, based on a single night's observawhich
recognize the complexity of auroral dynamics as tion. In this diagram (and in general throughout this
evidenced in observations of the auroral distribution. paper) the direction of the Sun (noon) is toward the top
The terms and descriptions of auroral breakup and of the panels and midnight is at the bottom (the view is
pseudobreakup originated in the years prior to the In- looking down on the northern magnetic pole). In this
ternational Geophysical Year (IGY), 1957/1958 [Stagg, Sun-fixed reference frame the Earth's rotation makes an
1937; Elvey, 1957]. Before that time it was known that observer on the ground appear to rotate anticlockwise
the most equatorward arc system was the one which (eastward) beneath the auroral distribution, as is illusactivated
[Elvey, 1957, p. 71]:
trated in Figure la by the small circle representing the
area which a ground-based observer is able to see. Thus
The final breakup of the display appears to be at the over a 24-hour time period an observer beginning at
time that the arc farthest from the auroral zone
midnight (magnetic local time) would traverse through
disrupts. This time is a very critical phase in the 0600, 1200, 1800, and 2400 magnetic local time (MLT) in
auroral display. The arc brightens and quivers just that order. In this event, two auroral breakups (with
before disrupting into rays, bands, and draperies that
single surge forms visible) were seen within 25 min of
spread rapidly over a large part of the sky. Very
shortly thereafter the display may change to flaming
each other (second and fourth schematics). In the fifth
aurora and pulsating surfaces. The diffuse surfaces
panel the "surge" has shifted westward (clockwise direcmay
recede toward the auroral zone becoming more tion) and is now located near 2100 MLT. This surge
diffuse and fading, thus terminating the display. motion has led to the term westward traveling surge
(WTS) to describe the dynamics of the breakup process.
This recounting remains a reasonable ground-based de- After each breakup event the aurora over a large local
scription of breakup even today: an arc intensifies and time extent is shown to have a "broken-up" or chaotic
moves poleward, forming a distortion in the auroral appearance.
distribution which is now called a "bulge." The term Using additional 1-min data from all-sky cameras, this
"pseudobreakup" was also introduced at that time. picture was revised to become the view in Figure lb
What was not known was how this small-scale auroral [Akasofu, 1964]. In this scheme, only a single breakup is
display was related to the larger-scale configuration. At presented, and the substorm concept is introduced as
about this time the concept of using all-sky camera data being the result of a single auroral breakup event. In this
was developed as a means to produce synoptic maps of view the global auroral distribution is altered from a
the larger-scale auroral pattern using multiple cameras "quiet time" distribution in a single unified process
[Elvey, 1957]. This technological advance still only al- termed the auroral substorm. The point where the actilowed
variations with timescales greater than 1 rain to be vation was first observed has been termed the onset
monitored. Detailed investigations into rapidly evolving location, and it was believed at that time that the
auroral forms had to wait until the advent of high- breakup always began around the midnight meridian. (It
sensitivity cameras in the mid-1960s.
is now known that the onset location can vary over many
After the IGY a large data set was available to inves- hours of local time and is on average located near 2300
tigate the large-scale features of the auroral distribution. MLT [Craven and Frank, 1991; Elphinstone et al.,
By 1962, loop structures [Davis, 1962] were known to 1995a].) According to this scheme, auroral activity could
drift slowly westward within which clockwise motions be divided into two basic stages: expansion phase (10-30
could be discerned (today these would be called large- min) and recovery phase (30 min to 3 hours). Each of

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 1 73
(2)
(3) (4)
(5) (6)
Figure la. Early schematics of the auroral substorm showing
the development based on one event from Akasofu [1963].
Magnetic latitude is shown, and magnetic noon is to the top,
midnight is at the bottom, and dusk is to the left. Westward is
in a clockwise direction. The small circle (in the night sector),
which rotates anticlockwise over the time interval, represents
the field of view of a ground observer rotating with the Earth.
Substorm activity is such a complex phenomenon that
a single schematic is unlikely to represent all possible
variations. For example, in the original Akasofu [1964]
substorm schematic (Figure lb) an arc system is seen
poleward of an expanding substorm bulge. In later schematics
this arc system disappeared (see Figure 2). It is
also significant to note that the Akasofu [1964] schematic
includes the fading of arcs in the polar cap, while in later
versions this feature was eliminated from the diagrams.
In actuality these and other variations are all observed to
occur. In the classical substorm there is only one WTS
which occurs at the west edge of an expanding bulge. We
know today that these features exist in large vortex
streets (multiple auroral vortices occurring in a long
string or street), spanning thousands of kilometers (for
example, the interested reader might question which
particular surge is the WTS in the fourth example given
in Figure 2 of Murphree et al. [1989]). Also, these features
do not necessarily travel, and when they do, it is in
an episodic manner.
More recently, it has been determined that auroral
expansionlike features last only 5-10 min and can repeat
every 5-30 min [Sergeev, 1974]. These have been referred
to as microsubstorms by Sergeev and later as auroral
intensifications by Rostoker et al. [1980]. The auroral
expansion bulge itself is composed of multiple elements
whereby the poleward motion of the bulge proceeds by
the formation of new arc systems at its poleward boundary
[Sergeev and Yahnin, 1979; Yahnin et al., 1983; Sergeev
et al., 1987]. In recent years the concept of a double
oval which is reminiscent of the double auroral glories
noted by Nordenskiold [1880-1881] has begun to play an
essential role in the understanding of mapping auroral
forms to the magnetosphere [Elphinstone and Hearn,
1992]. This feature also played no role in the initial
conceptualization of the auroral substorm.
In these earliest versions of the substorm the aurora
was still believed to occur in circular auroral zones
centered on the magnetic poles. However, the frequency
these was further subdivided into three parts. Expansion of occurrence of aurora in this circular zone was much
covers the time of poleward motion of the auroras after lower in the dayside zone than in the nightside zone
breakup, while recovery is the retreat equatorward back (because the dayside aurora was actually occurring furto
the quiet time distribution. In the latter stages of ther poleward). As a result of this belief the dayside
expansion, omega bands (another major auroral pattern auroral changes could not be characterized, since they
associated with substorms) apparently developed in the occurred at a higher latitude than was assumed. By the
early morning sector (fourth schematic of Figure lb). In time of the second schematic, shown in Figure lb, the
general, when the substorm bulge is ill-defined in local auroral oval concept of Feldstein [1960] was incorpotime,
the division between expansion and recovery be- rated on the nightside, but still very little was known
comes vague. When one region of the aurora is moving concerning the dayside morphology. The dayside auroral
poleward, another is moving equatorward, and in some distribution was even more difficult to characterize becases
the midnight sector aurora may return to a quiet cause the aurora there could be seen only during interstate
while auroras at other local times are still progress- vals of darkness centered around the winter solstice.
ing poleward. In principle, the two phases are easily Further, its location at high magnetic latitude made it
distinguished on a local scale at least. In practice, the difficult to find ground observatories capable of recorddivision
can be difficult to time accurately, which can ing the dayside aurora. These reasons help explain why
lead to problems in detailed comparisons with other it was not until the work of Feldstein and Starkov [1967]
measurements and definitions of expansion/recovery that the dayside aurora was introduced into the subphases.
storm schematics (see far left column of Figure 2). In

174 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
A. T=O . T=O"-5 MIN
80"
70'
60'
C. T=5-10 MIN D, T=10-30 MIN
E. T=30 MIN-I HR F T=1-2 HR T=2-3 HR
F.-.- A
Figure lb. The view of auroral activity generalized from multiple observations. Reprinted from Akasofu
[1964] (with kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington OX5
this later version, fan arcs appear in both the afternoon bulge begins in the morning sector with the WTS initially
and morning sectors shortly after breakup. These basic located near midnight. In this schematic the WTS travchanges
were later incorporated into the schematic given eled past the 1800 MLT meridian over the course of the
by Akasofu [1968], which comprises the center left col- 1-3 hour substorm. It is interesting to note that once the
umn of Figure 2. In this new schematic, other modifica- timescales given in Figures la and lb were selected,
tions to the nightside were introduced. The substorm these were adopted in later schematics. It is now known
70'
70 ø

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 1 75
Feldstein & Starkov (1967) Akasofu (1968)
a b
c
Montbriand (1971)
T=0 T=0 T=0
Fukunishi (1975)
d
T =-2--1 h
T=0-5 m T=0-5 m T=0-5 m T =-1 -0 h
T-5- 10m T--5- 10m T=5- 10m
\
T= 10-30 m T = 10-30 m
T=15-30m
T=0- 10m
T=10-30m
T 30 60m T 30 60m T 30 60m T 30 60m
T-- I -2h T-- I -2h T= I -2h T-- I -2h
Figure 2. Some of the various auroral schematics which have developed over the years (revised from
Montbriand [1992]). (a) Dayside shown for the first time. Reprinted from Feldstein and Starkov [1967] with
kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington OX5 1GB, UK. (b)
High-latitude arcs are eliminated in the view byAkasofu [1968]. Reprinted by permission of Kluwer Academic
Publishers. (c) Montbriand [1971] shows the equivalent proton aurora substorm. Reprinted by permission of
Kluwer Academic Publishers. (d) A view of how both the electron and proton auroras act during a substorm
[Fukunishi, 1975].

1 76 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
that an entire optical substorm (with the appropriate came quite stretched. Baker et al. [1978] found that
phases) can last for as little as 30 min [Murphree et al., preferentially field-aligned electron distributions oc-
1991].
curred at geosynchronous orbit for 30 min to 1.5 hours
The schematics discussed above were based on opti- prior to virtually every substorm. Thus growth phase
cal emissions associated with electron precipitation. became an integral part of substorm studies based on
Equivalent schemes have been developed based on the magnetospheric observations.
486.1-nm optical aurora (associated with proton precip- Support for the growth phase has been less evident in
itation). Similar general development (see center right the aurora, and some researchers even today will claim
column of Figure 2, from Montbriand [1971]) was ob- there is no distinctive auroral signature associated with
served, although the position of the substorm bulge was growth phase. Some of the early ionospheric auroral
put in the evening sector and the WTS did not appear to researchers who supported the concept of a growth
travel as far westward. Later schematics developed by phase were Starkov et al. [1971], Kaneda [1973], and
Fukunishi [1975] (Figure 2d) incorporated both electron Vorobjev et al. [1976]. These people found that the dayand
proton auroral developments. In this version it was side auroral activity often intensified and that the aurora
claimed that spatial and temporal developments for the would drift equatorward at 100-250 m s - about 1 hour
two forms of aurora are quite different. The proton before auroral breakup. This view is reflected in the
aurora was found to lay equatorward of the electron substorm schematic by Fukunishi [1975] (Figure 2d,
aurora premidnight with a crossover in the midnight and top).
postmidnight sectors. In the latter region, diffuse elec- It is probably obvious today that enhanced convection
tron and proton auroras overlapped. During expansion (see glossary) implies enhanced field-aligned currents
phase the electron auroras produced a surge form in the into the ionosphere (via the divergence of the associated
premidnight region which had little proton-associated electric field in an ionosphere of constant conductivity)
aurora, whereas in the postmidnight sector the situation which in turn are associated with precipitating particles.
was reversed, with the proton aurora forming the dom- Thus a growth phase in the magnetosphere is very likely
inant portion of the substorm bulge.
to have auroral signatures in the ionosphere. As we have
All of these schematics (with the exception of the seen, however, historically the substorm concept was
later schematic by Fukunishi [1975]) shared the common developed without good coverage of the regions where
view that the auroral substorm began with the auroral growth effects might be seen. Further, the lack of ground
breakup and that only two phases of the substorm were coverage could not ensure that an auroral breakup was
involved. In the early 1970s, McPherron [1970] and oth- not the cause of the activity. As a result the existence of
ers [see McPherron, 1972] introduced the idea of a the growth phase itself and definitely any associated
growth phase in the magnetosphere which precedes the auroral phenomena were controversial topics for some
substorm auroral breakup. This is supposedly a period of time. Today, a growth phase in the magnetosphere is not
enhanced convection and a time when magnetic flux is generally questioned, but the associated ionospheric sigadded
to the magnetotail. This new concept was based natures are still quite controversial. Global images of the
mainly on the use of case studies and so was criticized on auroral distribution under fully sunlit conditions have
the basis that the events were chosen to support the helped to identify unambiguously growth phase signaconcept
[McPherron, 1974; Akasofu and Snyder, 1972]. tures both on the dayside and on the nightside [Elphin-
Other critics, such as Rostoker [1974, p. 325], held the stone et al., 1991a, 1995a].
view that since equatorward arc motion was seen during Another feature of the auroral activity not recognized
both "growth" and recovery phases, "growth phase pe- earlier was periods of magnetic disturbances for which
riods were defined by virtue of subsequent expansion no apparent expansion phase signatures were seen
phase activity." It should be noted here that in a similar [Kokubun, 1977]. This has been associated with steady
manner, this is how pseudobreakups today are distin- enhanced magnetospheric convection and has been exguished
from "true" breakups. (Koskinen et al. [1993] tensively studied by Yahnin et al. [1994]. Another signifhave
found that pseudobreakups and breakups are es- icant auroral pattern which went unrecognized for a long
sentially distinguished by their strength and their conse- time was the existence of and indeed formation of polar
quences.) Critics believed that only rarely could a growth arcs during substorms [e.g., Ismail and Meng, 1982; Murphase
be manifested without substorm expansion signa- phree et al., 1987].
tures existing simultaneously somewhere else in the As optical observations were compared with magnetic
nightside ionosphere (see the discussion by Walt [1974]). records, it was found that a broad correspondence could
The growth phase idea was initially supported by be made between an auroral breakup and a magnetic
observations of the dayside magnetopause moving in- pulsation called a Pi 2. This event is in fact frequently
ward (assumed to be attributable to a reconnection used as a working definition of an onset [Rostoker et al.,
process) when the interplanetary magnetic field (IMF) 1980], although each of the multiple elements which
B z turned southward and subsequently a substorm took comprise a bulge is also apparently associated with Pi 2
place [Aubry et al., 1970]. Walker et al. [1976] found that activity [Sergeev, 1981]. Unfortunately, threshold probprior
to substorm onset, the magnetotail field lines be- lems also exist for this pulsation, which makes the exact

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 1 77
timing of an onset (defined this way) ambiguous [Elphin- This representation shows that previous to the breakup
stone et al., 1995a]. For a discussion of some of the event, the arcs at the poleward boundary were moving
characteristics of Pi 2 pulsations (a period of 40-150 s), poleward at a speed similar to the expanding bulge
see work by Yumoto et al. [1990, and references therein]. (about 0.3-0.5 km s-). For short intervals during the
Those authors have attributed these pulsations to global expansion, more rapid motion is seen. Meanwhile the
forced field line oscillations in a resonant magneto- arcs inside the bulge are moving equatorward (seen at
spheric cavity or to normal modes of the tail [see also about 55 min after the beginning of Plate lb).
Hopcraft and Smith, 1986; Takahashi et al., 1992]. Lester The 486.1-nm emission is a hydrogen line (H) due to
et al. [1983] have characterized Pi 2 pulsations in relation impact excitation of neutral hydrogen. Charge exchange
to the substorm current wedge. Another magnetic sig- alters the initial H + precipitation to a mixture of charged
nature associated with breakup is the Pi(b) pulsation. and neutral hydrogen (see work by Edgar et al. [1973] for
This has a broad frequency range (from the Pi 2 fre- details). This emission allows the proton aurora to be
quency up to 10 Hz) and has been attributed to a compared with its electron counterpart. Plate lc shows
precipitation-driven ionospheric source [Grant et al., keograms based on 486.1-nm scanning photometer data
1988; Grant and Bums, 1995].
at two locations for the same times as in Plate lb. The
1.3. Ground-Based Observations
emissions in the top right of Plate lc were observed at
large zenith angles (i.e., close to the observer's horizon)
We have described the gradual evolution of substorm and so differ somewhat from those seen in the center of
schematics and related some of the ambiguities in de- the observation on the left-hand side because of some
fining onset, the duration, and the phases of an auroral geometrical distortion. In order to get the emissions at
substorm. In this subsection we shall show some ground- the two locations to agree in latitude, it was necessary to
based auroral data to illustrate how a breakup can ap- put the hydrogen emission at 130 km altitude. A compear
from the ground.
parison of the 557.7-nm (electron) and 486.1-nm (pro-
Plate la shows a quite "clean" auroral breakup event ton) data shows that the electron breakup arcs lie within
on January 4, 1995, recorded in 557.7-nm emission by a the proton precipitation but poleward of its peak (see
charge-coupled device (CCD) all-sky imager (ASI) lo- also section 3.3). Note that when the breakup electron
cated at Gillam, Manitoba, Canada. These observations arc brightens and moves poleward, the proton emission
were taken between 2200 and 2300 MLT, and the images shows both a poleward and equatorward expansion. An
cover a 70 ø zenith angle field of view (i.e., the outer circle unusual feature that can be seen in the observation on
represents the locus of points 20 ø above the horizon). the right-hand side of Plate lc is the appearance of a
The view is presented as though one were looking down proton arc approximately coincident (in time) with auon
the imaging area in order to be consistent with global roral breakup but existing nearly 6 ø equatorward of the
satellite images presented later. North is at the top, and main hydrogen arc system.
west to the left. Some of the patterns discerned are Vallance Jones et al. [1982] have compared the proton
dependent on the sensitivity of both the imager and the and electron signatures as a function both of local time
scheme used to display the data. For example, in the first and time from onset. They found that the local time
two observations in Plate la, faint arcs and diffuse emis- relative to midnight was important in determining the
sion can be seen at intensities less than 1 kR. At 0459:06 location of the proton aurora relative to the electron
UT and for other images near this time, two faint forms aurora. Toward midnight the electron arcs exist closer to
separated in longitude can be seen located within the this proton precipitation. This was substantiated in work
broader diffuse emission. These have faded by 0503:06
UT, and by 0506:07 UT the most equatorward arc has
by Deehr [1994] and Samson et al. [1992].
brightened. This coincided with a Pi 2 signature at a 1.4. Global-Scale Satellite Imaging
nearby ground magnetometer. The activity rapidly pro- In contrast with the local view just presented, global
gressed poleward, and by 0515:07 UT, multiple distor- auroral imaging from satellites provides a larger-scale
tions known as folds or spirals can be seen at the pole- context in which to put the previous information. Imagward
boundary of the expanding substorm bulge. By ing of this nature was first reported [Anger et al., 1973]
0527:06 UT most of the field of view of the camera was using the scanning photometer on board the ISIS 2
filled with bright arcs. These gradually faded, leaving satellite. This imager was capable of acquiring one image
only an intense arc system at the poleward boundary by of both 557.7- and 391.4-nm emission every orbit (--100
0546:07 UT.
min) with a spatial resolution of about 10 km. This was
Images supply a necessary two-dimensional view of the first time that the entire auroral distribution could be
the developing aurora, but in order to describe the obtained in just a few minutes; this enabled determinadynamics
of the system it is useful to look at keograms of tion of the nearly instantaneous equivalent to the statisthe
same data. These are latitude-versus-time plots tical auroral oval. Another very successful set of imagers
(along the Gillam magnetic meridian) of the evolving were those associated with the Defense Meteorological
intensity. Plate lb shows the keogram developed from Satellite Program (DMSP) set of satellites [Rogers et al.,
the ASI data, assuming an emission altitude of 120 km. 1974]. These broadband imagers had similar time reso-

1 78 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
04:50:0 05:00:Oe 5:05:07
05:10:07 05:1 5:07 0 5:27'.0B
05:31:06 05'36:08 05:46:07
Sltm= e (557. nm 0:59-05: 6/ 50104UT)
Plate 1. (a) All-sky camera images (ASI) of an auroral breakup event. North is to the top, west is to the left.
(b) A keogram constructed from the ASI data to illustrate auroral motions. Note the poleward moving arcs
prior to onset and the sporadic poleward motions during the expansion. (c) Keograms from a scanning
photometer showing the proton aurora development. Note that the peak in this aurora lies equatorward of the
electron breakup arc. The observation at right shows additional proton precipitation (beginning at onset) far
equatorward of the main band of precipitation.
t

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 179
5
66
66
63-
AS! [Gilloft at 'I JO kin]
043,-0600 UT! 95004
&7,7 nrn
' I i
0 20 40 60
TIME, MINUTE]
Plate 1. (continued)
MPA [Gillam at fi33 km]
0,13'1-0600 U T/950 'I 0,1
486,, nrn
I I
28.9 k.N
0 5
C.
_ .
!
lution, but the spatial resolution of about 3 km allowed
detailed auroral patterns to be described. The Japanese
began a new type of auroral imaging with the ultraviolet
(UV) imager on board the Kyokko satellite [Kaneda et
al., 1977]. Choosing UV wavelengths made it possible to
view the aurora under full sunlit conditions and there-
fore opened up a new realm of auroral observing. As
well, this new imager was a snapshot camera, meaning
that all portions of the image were acquired simultaneously
and rapidly (every 2 min). Imagers of the scanning
variety have the drawback that different points on
the image are acquired at different times, which has
serious limitations for studying the dynamics of the substorm
process. The Dynamics Explorer (DE) program
also used UV imaging in multiple wavelengths using the
spin-scan technology. These imaging photometers had
full frame temporal resolution of 12 min [Frank et al.,
1981]. CCD snapshot UV type imagers were introduced
onto the Viking satellite by Anger et al. [1987], achieving
an exposure time of 1 s and a repetition rate limited only
by telemetry and satellite spin period (20 s for the Viking
imager). The quest for better time resolution to study
the rapid auroral dynamics has led to imagers with
repetition rates of 8 s (the Akebono satellite) [Oguti et
al., 1990] and 6 s with the Freja satellite [Murphree et al.,
1994a].
Plate 2 illustrates how the northern auroral distribu-
tion looks as viewed from a satellite (Viking in this case).
The left schematic shows the Earth with continental
outlines along with the statistical auroral oval as viewed
IJ_l
62
6O
56
MPA [Pinawa ot fi30 km]
0,131-0600 U It 90 ' 04
486, nrn
I ! I
4C oU
IIMF [MiN!,.,I F) il ,v [ MN[ IE. I
Plate 1. (continued)
tR

180 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 ! REVIEWS OF GEOPHYSICS
Vkng Satellite
055753 ut
ARGE-SCALE -URO. ALIg AGING
AE(11 $TAT1OI', S
Plate 2. A global satellite view (Viking) of the auroral distribution. The left image shows a simulated view
from the satellite. The white box shows the region to be acquired as an image. Also shown are the continental
outlines, the day-night terminator line, and the statistical auroral oval. The right panel shows the actual
observations of the double auroral oval after being transformed into a CGM coordinate system and having the
dayglow subtracted (most of the image was taken in full sunlight). Magnetic local times are shown every 6
hours, and 60 and 80 magnetic latitude (MLAT) circles are also given. The small circles represent 45 ø zenith
angle field of views (at 120 km) of the 11 ground stations used to determine the AE index in 1970. Also shown
for one of these stations are the field of views at 20 ø , 45 ø , 70 ø , and 85 ø . Note that most of the useful all-sky
camera data is for zenith angles less than 70 ø .
from the perspective of the satellite. The white line [Elphinstone and Hearn, 1992; Elphinstone et al.,
traversing the Earth represents the dividing mark be- 1995b].
tween regions of the Earth which are sunlit and those The small circles shown in the right image of Plate 2
which are not (the Sun is in the direction indicated by correspond to the horizon field of views (45 ø zenith
the arrow in the bottom left corner). The white box angle) of ground observers located at various magneindicates
the region of space to be imaged. The right tometer sites around the world. These were the sites (in
image shows some of the acquired auroral data after 1970) from which the auroral electrojet (AE) index was
transforming the data into a corrected geomagnetic determined. Note that the dayside aurora lies quite far
coordinate system (noon at the center top and dusk to poleward of most of these. For the station located near
the center left) and after the dayglow contributions midnight, additional horizon field of views (at 120 km
have been removed. Note that most of the aurora altitude) are indicated (20 ø, 45 ø, 70 ø, and 85ø). These
presented here could not be recorded at all by ground horizons show that as one looks more obliquely from the
observers since it occurred in full daylight. The main ground, a larger spatial distance in the ionosphere is
UV oval is the more equatorward of the two arc covered. This effect results in large perspective distorsystems
seen throughout the morning sector. We have tions of auroral features as the angle from zenith inused
the term main UV oval to distinguish this feature creases. Zenith angles more than about 45 ø result in
from the broader region usually identified with the significant spatial distortions. This conveys some of the
auroral oval. It also has an equivalent form at visible problems involved in observing the global dynamics of
wavelengths. We shall see in section 3 that this "dou- the aurora based on ground data. Stations which are
ble oval" structure is an important feature in under- useful on the nightside are not useful on the dayside, and
standing mapping from the ionosphere to the magne- so a large international infrastructure must exist to protosphere
in the later stages of an auroral substorm vide even a rudimentary coverage on a global scale.

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 181
1.5. Solar Wind, Currents, Convection, and the magnetic storm was simply a combination of a compres-
Aurora
sion of the magnetosphere and a series of injections
The main objective of this paper is to discuss active associated with the substorm process. Others believed
auroral phenomena, but a necessary element of this is to that the substorm was a more incidental event with
understand the relationship between the aurora and convection providing the ring current energization reother
geophysical data sets. For historical and other sponsible for a storm. Discussion of some of this conreasons
the most significant of these geophysical data troversy regarding the substorm/storm relationship has
sets is that provided by magnetic field measurements been given by Feldstein [1992]. Today, simulations by
(primarily on the ground). These deal predominantly Hudson et al. [1995] lead one to conclude that ring
with three phenomena important to auroral activity: current enhancements (and thus a magnetic storm) can
magnetic pulsations, ionospheric currents, and field- occur without the absolute need for an auroral substorm.
aligned currents. Magnetic indices (such asA U, AL, and This can come about from induced electric fields accel-
AE) have been used extensively to characterize the erating solar wind particles attributable to an interaction
ionospheric electrojet current flow.
of an interplanetary shock with the Earth's magneto-
These electrojets are closely tied to field-aligned cur- sphere. On the other hand, there has been considerable
rents, which in turn can be linked to electron particle success in re-creating the Dst index (storm activity) from
precipitation and hence to the aurora. Theories concern- the AE index (more directly related to substorm activiing
the substorm process sometimes make predictions ty). These calculations assume that energetic particle
regarding these indices (e.g., the near-Earth neutral line injection events associated with auroral substorms enmodel
described later), and considerable effort has been hance the ring current and give rise to the equatorial
put into predicting these indices based on the solar wind magnetic field changes. Thus magnetic storms are likely
conditions [e.g., Goertz et al., 1993]. It is therefore an to be due to a superposition of magnetospheric comimportant
aspect of auroral substorms to understand a pression, substorm injections, and convection. There
few of the basic relationships with these indices. are, however, details of this process which are not yet
The A U magnetic index is a measure of the eastward well understood. For example, why do not all substorms
electrojet and is associated with stations located in the give rise to a magnetic storm, and what governs this
dusk sector [Allen and Kroehl, 1975]. The AL index relation? What signature might one expect in the aurora
measures the westward electrojet with contributions from an interplanetary shock? How do these signatures
arising primarily from the postmidnight and dawn sec- relate to the substorm process? Gonzalez et al. [1994]
tors. Thus A U is naturally related to afternoon sector have given an excellent overview of some of these topics
auroral variations, while AL is more closely connected and of the storm/substorm relationship in general.
with the auroral substorm bulge and morning sector The ionospheric currents from which some of these
activity. These indices are related to the ground mag- indices are derived form only a part of the three-dimennetic
perturbations attributable to the currents flowing sional current circuit linking the ionosphere to the magat
ionospheric heights.
netosphere. During active times there are other specific
Another index, called Dst, is closely linked to the current elements which play a crucial role as illustrated
concept of a magnetospheric storm and to variations at in Figure 3. On the left (from Sergeev et al., 1996] is a
the magnetopause and energy stored in the ring current. three-dimensional view of the Earth (not to scale) show-
This index is related to the equatorial region magnetic ing the substorm current wedge (SCW), the partial ring
perturbations resulting from these magnetospheric cur- current (DRP), and the azimuthally symmetric ring currents.
During a magnetic storm, currents of several rent (DR). The SCW is intimately associated with the
megaamperes can flow in the ring current. A magnetic disruption (at onset) of the magnetospheric current
storm is considered to have occurred when this index is which flows in a dawn-to-dusk sense (cross-tail current).
less than about -30 nT. As discussed by Gonzalez et al. This disruption is linked to the ionosphere via field-
[1994], Chapman [1962, p. 9] made a link between the aligned currents (upward to the west and downward to
concept of the magnetospheric storm and the newly the east) which close in a westward electrojet (associated
emerging idea of a polar substorm:
with the magnetic perturbation which produces changes
A magnetic storm consists of sporadic and intermittent
polar disturbances, the lifetime being usually one
or more hours. These I call polar substorms. Although
polar substorms occur most often during magnetic
storms, they also appear during rather quiet
periods when no significant storm is in progress.
in the AL index). The DR is primarily responsible for
the storm time Dst variations (a westward current causes
a southward or negative magnetic perturbation at the
Earth's surface). The partial ring current may be partly
responsible for the eastward auroral electrojet driven by
closure field-aligned currents. In the ionosphere, westward
and eastward electrojet Hall currents probably flow
Thus, while there appears to be a general agreement in a manner similar to that shown in the illustration from
that storms cannot exist without substorms, the latter Grafe [1994], on the top right of Figure 3. These currents
can frequently occur without the development of a mag- produce magnetic perturbations on the ground. The
netic storm. In the past, some researchers believed a above field-aligned current elements form a part of the

182 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
Recovery
12
explosive etectrojets
l Current into ionosphere
½..:..'.; Current away from ionosphere
Expansion (severe)
12
667,0 9
Figure 3. The main current systems flowing in the ionosphere during a substorm. (Left) Schematic of three
of the current systems and their closure in the magnetosphere [from Sergeev et al., 1996]. (Top right) Dominant
electrojets existing during a substorm [from Grafe, 1994]. (Bottom right) Multiple field-aligned currents seen
during an auroral substorm [from Iijima and Potemra, 1978].
large-scale pattern which has been characterized in
terms of its direction of flow with respect to the ionosphere.
The pattern of observed field-aligned currents
[from Iijima and Potemra, 1978] is given on the bottom
right of Figure 3.
As noted earlier it was recognized by Davis [1962]
-1000" 0 4130 AU ' AL 70 65 GO 55
0200 , ,,'"-.__ ...... :
dusk/| ,x'" i [ - -===- --< --- dawn
000 \ ".'.:-. .---'%_.-'""):' /
ß ,. o .' I 'ø'' oø'
OgO0 / /''
0700 !
13800
0900
1000
1000[m/s] midnight
that auroral drifts were in general opposite to the direction
of current flow in the ionosphere. Figure 4, from
Nakamura and Oguti [1987], confirms this view at least
for pulsating auroral patches. This observation gives
information which can then be used to infer that valid
explanations for auroral features (i.e., omega bands and
Figure 4. Auroral drifts on February 16, 1980
[from Nakamura and Oguti, 1987]. The struc-
tures were mostly pulsating auroral patches. The
A U and AL indices are plotted in the inner
circle.

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 183
potentially the azimuthally spaced auroral forms linked
with onset) associated with pulsating aurora must also be a
related to E x B (where E is the electric field and B is
the magnetic field) plasma drifts.
Directly driven versus unloading signatures. The 600
flow of ionospheric currents is intimately related to the
electric field pattern imposed by the magnetosphere.
Under the simplifying assumption of equipotential mag-
netic field lines, any perpendicular (to the ionospheric x
m 400
magnetic field line) electric field results in an E x B drift _z
of plasma, termed convection. Although normally this
drift has no net current associated with it, in the ionosphere,
collisions reduce the mobility of the ions resulting
in Hall currents, which are the electrojet currents 200
described above. While some auroral forms follow this
plasma drift in the ionosphere, other features such as the
poleward edge of the substorm bulge do not. Thus two
fundamentally different types of auroral motion are
seen; one is directly related to convection, while the
other is related to magnetospheric topology changes or
to the growth of some instability which eventually be-
comes independent of the external driver (the solar b
wind). The large-scale convection associated with the
dawn and dusk sector electrojets (see Figure 3) is often
related directly to variations in parameters in the solar 21
wind. The component of magnetic indices for which this
is true is called the directly driven component, while the
other is called the unloading aspect of auroral activity F-
("unloading" since built-up energy is rapidly dissipated).
Akasofu [1981] considered the entire substorm process
to be a process driven directly from the solar wind, while
others described a period of energy build up in the u
ß 0
magnetosphere which is subsequently explosively re-
0
leased [e.g., Baker et al., 1981]. Both directly driven and 15 -
unloading systems are likely to actually operate [Nishida
and Kamide, 1983].
I SOLID SQUARE-LANL PRECURSOR
For the purposes of this paper, unloading activity will
OPEN CIRCLES-AURORAL PRECURSORS
be associated with the auroral breakup and the explosive
SOLID CIRCLES-MAGNETOMETER PRECURSORS
poleward motion, while growth phase phenomena and
auroral dynamics occurring without an obvious connec-
I
112300
I
112600
I
112900
I
113200
tion to the breakup will be considered as directly driven.
TIME (hhmns UT) OCT 19
The view adopted in this paper is that the energy buildup Figure 5. (a) AE index for an event on October 19, 1986,
process has associated with it enhanced convection, which had a very strong growth phase. The portion of the graph
which is effectively directly driven by the solar wind. The labeled "directly driven" occurred before substorm onset (1132
enhanced convection can be associated with character- UT), based on Viking auroral data. Note that for this event the
istic auroral changes which are growth phase signatures. growth phase electrojet index is nearly as large as that found after
Superimposed on this condition of enhanced convection substorm onset. (b) Various precursors in the afternoon/dusk/
evening sector seen in the 12 min prior to substorm onset. The
is an explosive unloading of energy due to some form(s)
open circles show the auroral precursors, and the solid circles
of instability in the magnetosphere. The combination of
show precursors in ground magnetometer data. The solid square
these two processes gives the complex activity evident in
shows a precursor signature seen at geosynchronous altitude. This
the auroral distribution. Elphinstone et al. [1991a] have strongly supports a causal relationship between some dayside
shown an example of how these two different processes auroral features and a subsequent substorm onset.
can appear in the auroral distribution. In that example,
afternoon and early morning sector growth phase auroral
activity could be associated with large AL/AU cation, 1995) and geosynchronous injections (G. Reeves,
magnetic perturbations. Figure 5a shows the AE (A U - personal communication, 1995) that substorm onset oc-
AL) index for this event. We know from the global curred between 1131 and 1132 UT. Thus the activity
images, Pi 2 information (G. Burns, personal communi- prior to this (marked as "directly driven" in Figure 5a)
18 -
I _...d I
I I
DIRICTLY DRIVEN
I
I
I
+UNLOADIN(
1100U 113UT-ONSET
,,// 1120JT
i
I I .I
103000 110000 113000
TIMF. (hhmm.q.q liT} ON Of TT'. 19. 1986
FIT: MLT= a UT(min before onset) + b
a= .96 +- .08 MLT/min
b= 22.9+- .5MLT
r=- .95 p = 7e-10
SPEED IS 26 L km/s
L=I 0 ==> 260 km/s O
T0(12MLT)= 1120UT
o
ONSET

184 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
was not related to the unloading of energy due to the characteristics of the solar wind may play a role in the
onset of some instability. Note that most of this directly occurrence of substorms.
driven activity occurred in the 12 min prior to substorm Tsurutani et al. [1990] showed that the continuous
onset and that it was of a magnitude comparable to that magnetic activity which follows a storm period is actually
seen during the unloading portion of the substorm. associated with Alfvenic wave intervals in the solar wind
This activity showed a pronounced temporal develop- and that the activity is driven by the southward compoment
from the dayside to the nightside (Figure 5b). This nent of the Alfvenic intervals. The concept that northantisunward
motion in the ionosphere could be seen in ward IMF plays a role in magnetic activity was extended
the magnetometer data (solid circles in Figure 5b), the by Rossberg [1989], who demonstrated that high solar
auroral image data (open circles), and even in geosyn- wind speeds increased magnetic activity for IMF Bz
chronous satellite data (solid square). This is to our northward. One must then ask if this activity is being
knowledge the first example of very clear dayside-to- driven directly by the solar wind or whether the solar
nightside progression of activity such that there ap- wind speed can play a significant role in substorm onset
peared to be a causal relationship with substorm onset. itself. Sudden impulses in solar wind pressure are also
This event also showed that without auroral imaging, significant factors, as seen in the work by Iyemori and
directly driven magnetic activity cannot easily be distin- Tsunomura [1983]. There are some indications that aziguished
from the activity associated with an auroral
muthally spaced auroral form (AAF) onsets (see section
onset of explosive poleward motion (most researchers
2.1 below) are associated with solar wind density enwould
choose 1120 UT to be the time of substorm onset hancements [Elphinstone et al., 1995a]. It is apparently
also true that some auroral breakup events are triggered
based only on the AL/A U indices). The fact that directly
by internal events [Horowitz, 1985].
driven (growth phase) auroral activity can reach compa-
It is evident that the relationship between what occurs
rable levels of AL and A U to that of expansion phase
in the solar wind and what results in the ionosphere is an
activity may indicate why the directly driven reconnecimportant
part of understanding the causes of auroral
tion model of Goertz et al. [1993] was apparently so
activity. In order to do this accurately, however, one
successful in reproducing AL and A U indices. Even
needs to know how well substorm activity is reflected in
though this model neglected certain fundamental asthe
magnetic indices listed above. These indices have
pects of the substorm process, it could nonetheless rebeen
created from a limited set of ground station data so
produce much of the index activity, perhaps because a
that if activity occurs in regions not covered by these
large component of the activity was directly driven for
stations, the activity will be missed. Are there time
the events chosen.
periods when no substorm activity exists? Are these
Causal relationships. One goal of the newly emergtimes
related to specific features in the solar wind, and if
ing field of space weather is to be able to predict sub- so what can this teach us concerning the substorm prostorm
activity based upon conditions in the solar wind. cess? These questions can be addressed using magnetic
Some of the initial work in support of this goal was done indices only if one can trust them to give accurate
by Fairfield and Cahill [1966], who showed that a change information concerning auroral activity.
of the interplanetary magnetic field (IMF) direction Figure 6 shows the AL/A U indices, the solar wind
from north to south was necessary but not sufficient for pressure, IMF B z, and Dst index for a 2-day period
an auroral breakup to occur. More recently, Kamide et which was very quiet. The magnetic indices strongly
al. [1977a] showed that it was virtually certain that a suggest no substantial substorm activity occurred. The
substorm would occur if the 1-hour-averaged IMF Bz sum of Kp (a planetary magnetic index) for the first day
was less than -3 nT (i.e., southward). The rationale for was only 8 +, while for the second it was only 6+ (both
this behavior is that the magnetic flux in the magnetotail of these values represent very quiet intervals). The llml
lobes is a function of the rate of magnetic reconnection and A U indices did not exceed 200 nT and only exon
the dayside magnetopause and in the magnetotail. ceeded 100 nT for two very short intervals. These also
This rate is governed by the Bz component of the IMF: indicate no major substorm activity. (For a substorm one
as Bz decreases, the magnetotail lobe flux increases, expects these indices to increase during expansion phase
which in turn sets the stage for a magnetospheric sub- and decrease during recovery phase.) As seen in Figure
storm and the associated auroral breakup.
6, the Dst index was almost always greater than 0 (i.e.,
It is apparent, however, that the correlation between not a storm period but the magnetosphere was either
substorm occurrence and the Bz component of the IMF compressed or had enhanced magnetopause currents). It
is complicated by the interaction between the driven and is clear that two of the positive excursions in the Dst
unloadihg components of the substorm process. Nishida were directly related to solar wind pressure enhance-
[1972] showed that substorms occurred when Bz was ments (both near 1200 UT on each day). Even though
northward, and Caan et al. [1977] have shown that a IMF Bz was negative for the first interval, the Dst index
turning of the IMF from southward to northward can indicates that no storm resulted and no obvious AL or
also play a role, presumably triggering a substorm by a A U signature was seen (there were two small AL excurdecrease
in convection. Further, it is clear that other sions between 0500 and 0800 UT). Viking auroral data

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 185
450
< 150
UNIVERSAL TIME (hhmmss)
120000 240000 120000
-150
-300 DNPO ]77DAYSIDE Io Ni7 I P.A.O ;NPO
-450
lO
z6 . .i,. v
2
m 3
-- ' " '
- ATA FOR ULY 19 and 20,186
Figure 6. Solar wind, Dst, and AL/A U data for a very quiet 2-day period. The AL index remained greater
than -200 nT and was generally greater than -50 nT. The sum of Kp for each day was 8 + and 6 +. Auroral
activity was observed several times during this interval at the lines marked on the plots. Terms used are as
follows: DNPO, a dayside precursor event followed by a substorm; no dayside, dayside activity was virtually
absent for this period; O, onset observed; and P.A.O., polar arc intensification followed by pseudobreakup.
showed distinct dayside precursors followed by a well- Magnetic indices are apparently not always sufficient
defined auroral substorm, which was clearly associated to identify auroral substorm events. A large infrastrucwith
the first pressure enhancement. This was followed ture currently exists to support the generation of magby
a time period during which the aurora was very quiet netic activity indices. What is somewhat surprising is that
and no obvious dayside (or nightside) auroral activity a similar effort does not currently exist and is not
was taking place even though IMF B z was southward at planned for the production of indices based on global
times and the solar wind pressure was high. No dayside auroral data. In principle, global UV auroral images
traveling convection vortices (TCVs) were reported dur- could be used to generate auroral indices. Tentatively,
ing this interval (nor were they for the second interval, one could envision three such indices. The first would
from 0800 to 2100 UT on July 20) [McHenry et al., 1990]. describe an approximate energy input into the iono-
This suggests that high solar wind pressure is not always sphere via the global integrated auroral intensity above
necessary for auroral activity. This may be related to some threshold [e.g., Cogger and Murphree, 1990]. The
whether the pressure is increasing or decreasing. second would be a measure of the location of the peak in
During the rest of the period there was a total of six the main UV oval distribution and could be defined
observed nightside auroral activations, three of which using several local time locations. The third could be the
were multiple onsets separated by about 20 min each area within which the integrated auroral intensities were
and three of which had dayside (or polar arc activations) found. Indices of this nature would allow magnetoauroral
precursors to the nightside onsets. At least five spheric researchers quick and direct access to the state
of these events occurred when IMF Bz was northward of the auroral ionosphere in a much more direct way
(see Figure 6). The event labeled "P.A.O." corresponds than can currently be done based on magnetic indices. In
to the intensification of a polar arc followed by an practice, these indices are difficult to construct algorithauroral
activation which brightened, propagated west- mically, and imagers to date cannot continuously obward
for 2-3 min, and then faded. This event was ap- serve the auroral distribution. Nevertheless, we can
parently triggered by the solar wind pressure increase, look forward to the time when such indices become
but ionospheric or magnetospheric conditions were in- available and can be compared to existing magnetic
sufficient to trigger a full auroral substorm. This 2-day indices.
period serves to illustrate that the mechanisms for driv- Substorm current wedge. As mentioned above, the
ing auroral activity are still not clear and that activity can SCW forms in association with a disruption of the crossoften
occur without obvious signatures in the standard tail current at the time of auroral breakup. Figure 7
magnetic indices.
illustrates how the SCW relates to the poleward expand-

186 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
:-,r :*'';.< ................. ............. ;:'-:;ii!*:*::-:- ''-**"*'" .....
::.. ..:..: : .........
0 ?::-.: ....... ._-...DMSP- ' :.'
Viking.' '-4.. ........ --:.:.;c::Jf, .......... .; .............................
1812 to 1:, UT
. .
:..::;.:::. .......... .:. ...... ..:-' '-'.
?:..
::: : . : : . ,
.'"':'-:-:. up:-: D '-:" ' .'. '
:) ...... :t.'/ ..... .. - ......... :::-:. < .......... '::
. .-:.. - ß ...... :::....;:.....e- -:. '-'j:
: .... -. ..... ... :. :.f-":;: ....
..... ...
.. J ,
. -";::-.:
: ,%:: ........
-
1: .: : UT
APRIL 1,1,.: SUBSTORM CURRENT WEDGE
Figure 7. An example of how the substorm current wedge (SCW) relates to the expanding auroral bulge
(revised from Sergeev et al. [1996]). The upward and downward field-aligned currents occur at the edges of the
bulge. The thicker white meridian lines show the previous location of the currents in order to illustrate the
motion of the SCW. Also shown in the top left are the DMSP and Viking satellite trajectories with particle
boundaries marked on them. Apparently, closed field lines extend poleward of the main UV oval. Numbers
within the plate correspond to points on the Viking trajectory as follows: 0, equatorward edge of trapped ions;
1, equatorward edge of electrons and isotropic ions; 2, rapid energy change in ions (peak in electron flux); 3,
gap in both ions and electrons; 4, poleward edge of ions; and 5, poleward edge of electrons.
ing bulge for a simple "classical" substorm event. The
event was part of a coordinated data analysis workshop
(CDAW-9); midlatitude magnetometer'data were analyzed
by Sergeev et al. [1996] to determine the locations
of the associated upward and downward field-aligned
In a classical substorm one considers convection dur-
ing substorms to be influenced dominantly by the substorm
bulge (or SCW) and directly driven activity. We
have seen that in some schematics of the substorm
currents. These have been superimposed in Figure 7 on
process, polar arcs disappear prior to onset. Polar arcs
are commonly thought to occur during IMF B north-
Viking image data during the expansion phase. For this ward conditions. Thus, as B turns southward prior to
event the upward field-aligned current moves westward onset, the polar arcs should disappear. One interpretaalong
with the spiral form (i.e., WTS) at the west tip of tion of this is that the southward turning creates open
the bulge. The downward field-aligned current is field lines in the polar cap, and thus these arcs which
broader but also corresponds well with the eastward were on presumably closed field lines disappear. The
edge of the bulge. These currents are closed via a west- polar arc in the top left of Figure 7 illustrates, however,
ward electrojet current through the substorm bulge that these features can play an important role in the
which gives rise to the ground magnetic perturbation auroral pattern during substorms. A high-latitude polar
associated with the substorm process (i.e., a more neg- arc system is visible in the dusk sector between 1812 and
ative AL index). Pellinen et al. [1995] have recently 1840 UT. By the time the substorm bulge has appeared,
shown that the downward current is established within a the polar arc system has faded so that it is below the
minute or so after the upward current and that it is threshold of visibility but not necessarily gone. Toward
carried dominantly by cold ionospheric electrons. The the end of the substorm expansion the polar arc reapidea
of the SCW is also consistent with the differing peared close to its previous location. Thus, if the arc
patterns of proton and electron auroras in the substorm system was on closed field lines initially, the region was
bulge discussed by Fukunishi [1975] (i.e., protons are probably on closed field lines throughout the expansion
associated with the eastern part of the bulge and a phase interval. This disagrees with previous views condownward
current, and electrons are associated with the cerning the development of the open field line region
upward current to the west).
during substorms.

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 187
DMSP-F7 and Viking traversed the dawn and dusk poleward edge of the auroral distribution. For this case
sectors prior to onset, and their trajectories have been the diffuse poleward boundary has remained fairly conprojected
to the ionosphere (Figure 7). The DMSP stant (about 70ø), but the bright auroral features have
neural network boundaries [Newell et al., 1991] indicate shown first an equatorward motion followed by a "leap"
that low-altitude boundary plasma sheet particles existed or intensification of aurora at the poleward boundary.
throughout the dawn sector. The Viking satellite re- Note, however, that no motion of the poleward boundcorded
trapped particles far equatorward of the main ary of the weaker aurora was involved. This illustrates
UV oval and structured electron precipitation extending that the poleward system during a double oval configuup
to the polar arc. Sunward convection existed through- ration can develop activations independently of the
out this region. There was a gap in both the electron and more equatorward system. In some circumstances these
proton precipitation in a small area between the polar might be mistaken for poleward leaps. We shall see later
arc system and the main UV oval. These observations that these events (section 3.4) sometimes do correspond
support the view that some polar arcs are the extension to the recovery of the plasma sheet and thus at least in
of the normal auroral distribution to high latitudes and principle are consistent with the retreat of a neutral
that the main UV oval lies equatorward of the openclosed
field line boundary. Adopting this view means
line.
that it is incorrect to evaluate the changing size of the 1.6. Summary
open field line region by measuring the area bounded by We have attempted to illustrate in a short space a few
the auroral oval, such as has been done in the past of the broad correspondences which exist between the
[Frank and Craven, 1988].
aurora and other geophysical ionospheric observations.
Poleward leap and electrojet fine structure. Ob- The examples were chosen to illustrate simple aspects of
servations concerning the poleward leap of the westward the aurora as well as to show features which may not be
electrojet and the aurora have been used as support for commonly recognized by magnetospheric substorm rethe
near-Earth neutral line (NENL) model (see "reconnection"
in the glossary and section 4.4) of substorms
[e.g., Hones et al., 1985; Hones, 1992]. The retreat of the
searchers.
neutral line in the magnetosphere in the recovery phase
is assumed to map into the ionosphere as a poleward
2. MODULAR CONCEPT
motion of the westward electrojet in the ionosphere. The What is a global auroral substorm? As the title of this
expected consequence of this is that the aurora should article suggests, there is some disagreement concerning
also "leap" poleward as the neutral line retreats tail- what is and is not an auroral substorm event. With a
ward. Nielsen et al. [1988] have looked for evidence for a limited data set, what one researcher might call a subpoleward
leap of the auroras but did not find any evi- storm, another might claim is a pseudobreakup or somedence
for a motion separate from the irregular motions thing unrelated to the substorm process [e.g., Pellinen et
seen during the expansion of the auroral bulge. al., 1990]. Pudovkin [1991, p. 17] has referred to the
Plate 3 shows an example of a rapid change in the UV substorm as a wide variety of phenomena which "may
aurora and electrojets after the initial expansion of the differ... in the sequence of the physical processes probulge.
This example has been derived from the fine-scale ducing them." Pi 2 pulsations, dipolarizations, and enelectrojet
structure calculated by Kotikov et al. [1993]. ergetic particle injections are frequently used as sub-
On the left are keogram plots of the electrojet intensity storm indicators in the magnetosphere, and yet these
and auroral intensity centered around 2300 MLT (ec- same events would often not be characterized as "subcentric
dipole 1985 coordinates). On the right the cor- storms" in the ionosphere. One may wonder to what
responding large-scale auroral patterns are displayed. extent a substorm event must resemble the schematics
For this case a westward electrojet existed near the shown in Figures la, lb, and 2 to be accepted as an
poleward boundary of the aurora between 1600 and 1630 auroral substorm. A local definition of a substorm is
UT. This corresponds to the intense arc system which limited and says nothing about whether the global prowas
situated at the poleward edge of the bulge at the cess illustrated by the general schematics is occurring
beginning of substorm recovery. The electrojet began to [Elphinstone et al., 1995a]. Theories, in turn, have tended
weaken, and the bright aurora drifted equatorward until to be developed on the basis of a particular data set
about 1630 UT, when a second equatorward electrojet which is generally biased in location and/or in spatial/
began to intensify. This corresponds to the main UV temporal resolution.
oval seen at 1652:13 UT (Plate 4, middle right). At this As a consequence many researchers try to explain
point the double electrojet and oval structure are clearly entire substorm schematics using a particular theory that
evident. Between 1645 and 1700 UT the electrojet near is based upon limited types and numbers of observa-
68 ø suddenly intensified concurrently with a rapid inten- tions. This situation reminds one of the classical story of
sification of the poleward arc system. This is an example the blind men who encountered an elephant: they all
of a sudden intensification of a westward electrojet ac- arrived at different conclusions regarding the nature of
companied by an intensification of the aurora near the the elephant depending upon whether the individuals

188 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
had touched the tail, the legs, the trunk, etc. It is impor- This development can be very valuable in the undertant
to remember that with auroral substorms, just as standing of the processes involved in the substorm cycle.
with elephants, descriptions based upon limited obser- Substorm phases can also be somewhat ambiguous.
vations are incomplete; they are, however, complemen- Oguti [1981, 1992] has noted that pulsating auroras
tary and must be synthesized in order to provide a should not be considered a substorm recovery phase
complete and accurate description or theory of the phe- phenomenon but rather a pattern which begins after an
nomenon being investigated.
auroral breakup. He noted that the phases of recovery
The substorm as set out in the short paper byAkasofu and expansion can occur at the same time but in differ-
[1964] was a collection of ground-based observations put ent regions. This is reminiscent of the argument used
into a unified whole. It allowed for the interpretation of against the existence of a growth phase (i.e., that the
what was seen at various locations to be put in a frame- growth phase occurred at the same time as expansion
work which made them comprehensible. It has been very but at a different location). Similar arguments can be
successful in describing what can happen during auroral made when trying to describe when the recovery phase
activity. There are, however, variations to this behavior begins. What local time range should be used to deterwhich
are useful to understand in their own right. A wide mine when the aurora begins to move equatorward?
range of phenomena exists in the magnetosphere and When the auroral bulge is ill-defined, this description
ionosphere which support and allow for the existence of results in an equally ill-defined time for when the recovthe
global structure known as the auroral substorm. ery phase begins. Similarly, if pulsating aurora is used,
A knowledge and study of the smaller activity cycles then the distinction between expansion and recovery
within the larger substorm cycle can aid in the interpre- phases is probably meaningless.
tation of the aurora substorm. Pseudobreakups, precur- We argue that there are important aspects of auroral
sors, dayside activity, multiple onset substorms, substorm activity which can become lost in the traditional depicsequences,
different types of onsets, auroral spirals, and
tion of an auroral substorm. Cycles exist in auroral
the vast number of other auroral forms seen at various activity which couple to allow the global concept of an
points during auroral activity are elements of a substorm
auroral substorm to exist. These cycles do not always
have to act together to give a traditional global auroral
scheme. How they interact, when they occur, and their
substorm. The cycles themselves, in turn, are a result of
fundamental importance to the thing known as an audifferent
magnetospheric processes coupling in a repeatroral
substorm varies from event to event. The problem
able manner. Occasionally, one or another of these
in magnetospheric physics then becomes what counter-
"modules" may not be of sufficient strength to initiate
part each of these phenomena have in the magnetothe
next portion of the cycle. At this point the chain of
sphere and how these interact together. The emphasis
events leading to a complete cycle is broken. This, of
can then shift from a debate over which theory explains course, influences the future course of events with the
the auroral substorm to the following question: Where
possibility of new modular elements becoming active. As
does each individual piece fit in the overall puzzle?
we shall see in the last section before the glossary, these
While it is to be expected that not all substorms are
cycles and the modules from which they are constructed
the same, it is becoming increasingly clear that funda- allow one to understand how different observers with
mentally different activations of the aurora occur and
different magnetospheric regions of interest can conshould
be distinguished from one another. Montbriand
clude conflicting ideas about the substorm process.
[1992] has discussed some differences which exist be- This section will outline some auroral features which
tween the various schematics. The changing schematics are examples of the ionospheric signatures of these
shown in Figures la, lb, and 2 and discussed in section modular elements. It will also describe a little of what is
1.2 reflect an ambiguity which exists from event to event currently known concerning them. The emphasis will be
and makes the characterization of a particular event on individual phenomenological auroral modules which
from the general schematic rather difficult. One sche- can be fit into a particular auroral event rather than on
matic shows arcs poleward of the expanding bulge. An- trying to construct a global scenario which will not repother
does not. This particular ambiguity exists because resent the details of individual cases. Auroral patterns
the double oval was observed on some occasions but not are used to understand the modules involved, since these
on others. This difference provides important mapping are the only means at present to get very high temporal
information which is not otherwise available. Another and spatial resolution with a global ionospheric perspecschematic
shows polar arcs fading before onset. Others tive. The reader should understand that each of these
do not. We now know polar arcs can appear and disap- has an associated process with which it can be identified.
pear at virtually any time during the expansion of the In the last section of the paper we give some simple
auroral bulge and, of course, at most any time during examples of how this modularization can be used in
auroral activity. In the schematics, one sees a rather understanding some of the underlying magnetospheric
chaotic set of arcs in the panels after breakup. This does cycles which play a role in auroral activity.
not depict in a useful way the order and the individual Several ideas bear directly on the modular concept.
features which usually exist in one manner or another. Figure 8 illustrates some of the fundamental views which

34 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.- MODULAR AURORAL ACTIVITY ß 189
1630UT
ELECTROJETS
170OUT
1630UT 1700UT
INTENSITY(23MLT-ecc)
70MLAT
60MLAT
70MLAT
60MLAT
1.'x -: UT
165213 UT
17OO16 UT
Plate 3. The development of the fine structure of the westward electrojet (data from Kotikov et al. [1993]):
(top left) keograms of the electrojet development and (bottom left) auroral intensity at 23 MLT. White and
red indicate the highest intensities of both auroral and westward electrojet activity. Observations at right show
the larger-scale auroral development. The white boxes (lines) show the locations of the westward electrojets.
This illustrates that the poleward intensification of either the electrojet or aurora does not imply a poleward
leap of the aurora.
are altering the standard model of auroral activity. Sergeev
and Yahnin [1979] reported that the substorm expansion
phase is a series of pulses which combine to produce
the bulge. These "quantum elements" are considered to
be bursts of reconnection occurring in the tail. In this
view each substorm bulge is different to the extent that
the number, size, and duration of these elements determine
the qualitative nature of a particular auroral activation.
The amount of rainfall and the size of the drops
during a rain shower therefore becomes a reasonable
analogy for the level of auroral bulge development. Thus
one might expect to see events varying in their ionospheric/magnetospheric
impact. This idea helps explain
the variety in intensity and duration exhibited by the
expansion bulge and the WTS.
Kennel [1992] introduced what is known as the
"strong" Kiruna conjecture, which essentially decouples
the supposed reconnection-based convection from the
mechanism of near-Earth substorm onset. One auroral
equivalent of this concept could be associated with the
different locations at which seemingly independent intensifications
occur on the double-oval distribution. This
first step toward modularization using separate auroral
activations was introduced by Cogger and Elphinstone
[1992]. This view, combined with the work mentioned
above, led Elphinstone et al. [1994a, 1995a] to the modular
view of the auroral substorm which is presented
here.
Lui [1991] provided an additional component
to this
idea through his synthesis model of the magnetospheric
substorm. In the general form of this scheme a near-
Earth plasma instability initiates a rarefaction wave

190 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY
V. SERGEEV C. KENNEL
MICRO-SUBSTORMS
ANALOGY TO
MANY DIFFERENT
TYPES OF RAIN
DISTANT TAIL
AND NEAR EARTH
PROCESSES
INDEPENDENT
T. LUI
SYNTHESIS
MODEL
USING DIFFERENT
ASPECTS OF
MANY MODELS
NEAR EARTH
ONSET
NEUTRAL LINE
LATER
34, 2 / REVIEWS OF GEOPHYSICS
Figure 8. Three key elements associated with the
modular concept of auroral activity.
which propagates tailward, eventually triggering recon- roral activity, but when it does occur, it gives some
nection via a reduced Bz in the neutral sheet. In this predictive power concerning onset. It is also a good time
substorm scenario a coupling exists which begins near marker by which other observations can be placed in
the Earth and creates effects in the deeper tail. This idea context.
helps one to understand the poleward expanding bulge. Figure 9 illustrates dayside auroral patterns, most of
Other substorm explanations reverse this concept with which develop during active conditions (the exception to
the more tailward regions generating near-Earth conse- this might be the high-latitude dayside form in the top
quences [Baker et al., 1993]. These examples help to middle). Only the last pattern (bottom right) is confined
illustrate the variety of both interpretation and obser- to time periods after a substorm onset. The other patvation
which need to be accommodated in advancing a terns have all been known to develop shortly before
generally acceptable substorm theory. The modular nightside substorm onset as well as after. These dayside
concept of auroral activity adopts an alternative ap- activations are probably directly driven by the solar wind
proach whereby each of the above two scenarios can and can therefore occur at any time during an auroral
occur and contribute to our understanding of auroral nightside event. What makes them interesting is that
activity.
they can be precursors to subsequent nightside events.
As well as a very basic system of modules to describe Elphinstone et al. [1991a, Plate 2 and Figure 7] have
essential elements of activity and/or the associated cou- shown a clear example of this type of activity beginning
pling, there are also secondary modules which may be prior to any nightside breakup. In their paper a clear
identified with specific processes which may become case of when fan arcs develop during a double oval
active at different times. These secondary modules are configuration (i.e., recovery phase of a previous submore
likely to appear at a variety of times during some storm) is also shown. This dayside activity gradually
cycle of auroral activity. The reader should refer to spreads to the nightside, AAFs develop, and finally a
section 5 for a description of how these modular ele- substorm bulge occurs. Thus the dayside activity can
ments fit into larger cycles of magnetospheric activity. occur during different phases of one substorm but still be
a growth phase activity for subsequent intensifications.
2.1. Primary Auroral Modules
Figure 5 shows that this precursor activity appears to
This section contains some of the fundamental fea- have a causal relation to substorm onset, and so from a
tures which should be factored into any complete de- practical viewpoint this phenomenon is extremely imscription
of auroral activity. These auroral modules have portant in future studies which aim to do space weather
been labeled "primary" ones in the sense that they predictions of substorm onset. Events of this nature
provide time markers with which to reference the occur- allow one to follow the step-by-step motion which conrence
of other observations.
nects these dayside events with the nightside. They leave
Auroral growth and dayside patterns. As men- little doubt as to a growth phase existing in the ionotioned
in the Introduction, it has taken some time for the sphere. Some of these brightenings can proceed exgrowth
phase to become established as a separate sub- tremely rapidly from the dayside to the nightside so that
storm phase. Today this phase is well-established in the in about 5-10 min the entire progression is over. This
magnetosphere but less well understood in its iono- can be superimposed on a slower, probably convectionspheric
signatures both on the dayside and the nightside. driven enhancement of the auroras which can last for an
Similar to other modules, growth phase may not be a hour. An interesting aspect of this activity is an apparent
necessary element for the instigation of nightside au- coupling between the 1400 and 0300 MLT sectors during

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 191
0 Impulsive
>0 12 By delendent Low Latitude 12
"High Latitude ..,,' VD amsei Form Pulsations I "fan arcs"
Dayside /"'""'"' Form" urora {r .,,9 at--3 minutes
' By dependent pOlar'arc
Afternoon
Spiral Forms
__X= 5
D
12 Omega-like Forms
Post Substorm 9
Figure 9. Dayside morphologies during active times [from Elphinstone et al., 1993a].
this slow auroral enhancement. This again is not neces- brightening. The latter definition has significant ambigusary
for an onset but needs to be understood as part of ity associated with it [Elphinstone et al., 1995a].
the possible range of auroral behavior.
lypes of auroral onset. Another obvious time
The short-timescale event mentioned above may be marker is the beginning of auroral breakup or pseudorelated
to a compression of the dayside magnetosphere breakup. We shall consider onset to be the explosive
which can generate fast-mode compressional waves at activation of the aurora in the evening sector. This event
the magnetopause which propagate into the inner mag- may or may not develop into a major disturbance, denetosphere.
These, in turn, couple to shear Alfven waves pending on its coupling to other modules. This activation
which initiate disturbances at low altitude. An example takes on several forms. Some of these patterns are
of such an event is given by Erlandson et al. [1991]. This definitely due to different processes, while others may be
occurred shortly before a substorm onset and had asso- separate ionospheric manifestations of the same magneciated
signatures in the dayside auroral distribution [El- tospheric process.
phinstone et al., 1993a].
While there are many forms onset can take, Figure 10
On the nightside it is known that arcs move equator- shows two of the fundamentally different types of onset.
ward prior to onset and can (but not always) fade before The first (Figure 10, top) involves periodically spaced (in
onset [Pellinen and Heikkila, 1978]. The fading may azimuth) auroral forms known as AAFs. Detailed examindicate
that electric field fluctuations trigger breakup. ples of these forms have been given by Elphinstone et al.
Atkinson [1993] has used the observation that the growth [1995a]. The faint forms seen at 0459:06 UT in Plate la
phase equatorward motion tends to zero just prior to may be the ground-based equivalent of the AAFs. One
onset to suggest that breakup is associated with the or more of these forms develops explosively into a subconvection
and topology of the magnetosphere ap- storm bulge (Figure 10, top right). AAFs can cover many
proaching a steady state. This configuration is then un- hours of local time and can occur along with Pc 5
stable to small perturbations. AAFs can develop 20 min magnetic pulsations and sometimes omega bands. They
prior to an explosive onset, and pseudobreakups can
--1
travel eastward, east of onset, at speeds of 1-2 km s
periodically occur for about an hour prior to a bulge and westward, west of onset. They are not auroral curls
forming. These suggest that a wave instability and ion- (see below for a description) and, in general, are not
ospheric preconditioning are important in triggering a auroral spirals although they have been observed to
major onset. These precursor observations make it nec- develop into these forms. Their striking regularity in
essary to define optical onset as the time of explosive longitude leads one to speculate on a wave-associated
poleward motion rather than as the time of an arc mechanism for their generation. Figure 11 illustrates the

192 ß Elphinstone et al ß MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
Types of Substorm Onset
Onset A. expansion
80* 70* 60*
Equatorward boundary
can be dramatically
(:hanged or not
B. altered at all
18 MLT 6 MLT
\ ß Activity develop
o
Figure 10. Two types of onset [from Elphinstone et al., 1994a, 1995a].
mode number (the number of waves found within a unit The second form of activation is shown in the bottom
circle) measured in the few minutes surrounding sub- of Figure 10. This type of event involves an eastward
storm onset. The mode number peaks near 100 and then propagating (approximately 5-10 km s -1) intensification
shows a fairly sharp cutoff at higher mode numbers. beginning near 2100 MLT at or near the main UV oval.
Multiple onsets observed in the Viking dhta set have The activity spreads eastward with vortex streets of spibeen
known to occur from separate parts of the long rals developing [Murphree and Elphinstone, 1988; Pellistring
of AAFs so that auroral bulges can form in dif- hen et al., 1995].
ferent local time sectors and create independent effects Other onset types can be important for evaluating
in each region. Here we can return to the question posed substorm theories, although these may be minor variaby
Elvey [1957] (see section 1.2) and answer that breakup tions on the versions described above. There have been
activity seen locally may not be causally linked to each
other. This is in contrast to the standard substorm
model.
many reports of the bulge-type activation occurring from
the main UV oval during a double oval configuration
[Elphinstone et al., 1991a, b; Murphree et al., 1991; Cogger
and Elphinstone, 1992]. These onsets occur during the
recovery phase of a previous substorm [Kamide et al.,
1977b; Elphinstone and Hearn, 1993; Murphree et al.,
MODE
20
I
NUMBER
40 60
I I
AT SUBSTORM
80 100 120
I I I
ONSET
140
I
1993], and approximately 20% of all onsets involve this
form of activation [Elphinstone et al., 1995a]. In the
¸
015
1o
20
z 5
Peak in mode number at 84 to 104
latitude = 63.8 MLAT
[
20 40 60 80 100 120 140
MODE NUMBER
Figure 11. Observations of the mode number of azimuthally
spaced auroral forms (AAFs) at onset (derived from work by
Elphinstone et al. [1995a]).
20
15
10
future we need to understand which of these onsets are
triggered and which are spontaneous or internally
driven. Pudovkin et al. [1991] has outlined the differences
in character and development between triggered and
spontaneous onsets.
Quiet time intensifications can occur on the poleward
boundary of the auroral distribution [de la Beaujardiere
et al., 1994]. These are not technically substorms, but
they nevertheless play an important role in substorm
development. Cogget and Elphinstone [1992] have shown
that such intensifications are one means by which auroral
activity develops after substorm expansion. One
such event shown in their paper occurs coincident with
the recovery of the plasma sheet [Hones et al., 1987].

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 193
Low latitude
)ral form
12
Multiple arc
westward
of surge
St;
N-S structures
Dayside spiral
Dayside Active
without
4,a Double oval every
. / developes
2:!prefetentially in morning)
Traversing arc
sector gap
18 MLT Double Oval 6 MLT
Eastward motion
Omega-like
bands
Main UV oval
Collapsed "BPS"
Figure 12a. Double-oval configurations [from Elphinstone et al., 1994a].
1. Most poleward arc intensifies with or
vortex street periodically
-10 minutes.
2. Emissions within bulge fade. Local
H bays recover with fading.
Formation of double electrojet begins.
Main jet in poleward oval.
3. Poleward motion gradually ends.
Recovery strongly local time
dependent.
Periodic -10 minute
N-S "injections"
A fundamental question exists when one tries to tion does not determine subsequent development but
decide precisely what a weak substorm is versus a that the previous magnetic field configuration might.
pseudobreakup. Koskinen et al. [1993] and Nakamura et Precursor activity in this view is simply an onset module
al. [1994] note that there are no real observational or pseudobreakup which does not couple directly to
differences between pseudobreakups and breakups ex- other modules. They may, however, play a key role in
cept for intensity and spatial scale. They believe that the conditioning the system for a more major event.
instabilities involved are suppressed in pseudobreakups Auroral bulge. This module allows one to distinbefore
reaching some critical level. Local or instrumen- guish optically whether the previous event is a pseudotal
definitions cannot usually answer whether an event is breakup or not. If an arc brightening is followed by an
a pseudobreakup or a weak substorm, since both can explosive poleward motion, then one would generally
have essentially the same signatures. Perhaps one should tend to call the event a substorm onset (although strictly
consider the pseudobreakup to be associated with the speaking it still may not fulfill the global characteristics
onset module of a substorm which activates without shown in Figures 1 and 2). This module is composed
further large-scale consequences. This is consistent with partially of many secondary modules given below, and it
the notion that an instability does not reach sufficient is not clear if it could exist without them.
strength to have larger-scale influences. In our substorm lhe double oval and its variations. This distribu-
"ecosystem," onsets and pseudobreakups may be lik- tion is important in the substorm process for underened
to two lions. One lion is hungry and therefore eats, standing the mapping of auroral forms [Elphinstone et
while the other is not hungry and therefore does not eat. al., 1995b, and references therein]. It exists mostly dur-
This behavior is determined by the lion's previous inter- ing time periods when the AL is in recovery (i.e., inaction
with other elements in the ecosystem. Similarly, creasing). In many cases the double oval forms when the
whether an event becomes a "true" onset versus a aurora locally reaches its most poleward extent. At this
pseudobreakup is probably determined by other modu- time the aurora immediately equatorward within the
lar elements. These could include things such as thinning bulge begins to fade (see, for example, Figure 11 of
and earthward motion of the plasma sheet, AAFs, ion- Elphinstone et al. [1993b]). Figure 12a shows a few of the
ospheric conductivity, convection and its relation to the double oval patterns which can develop after this time.
conductivity distribution, etc. Ohtani et al. [1993], for Many of these patterns are derived by including the
example, found that the spatial scale of current disrup- secondary modules given in the next subsection.

194 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF 'GEOPHYSICS
Activation of
most poleward oval
(Boundary Layer Substorm)
Dayside
'auroral
asymmetry
Routes to New Morphology
Radial expansion from
Double Oval main UV oval
(Near Earth Substorm)
ost Orlers p. g iade vni!ili! lad e
without polar arcs
Throat for
By

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 195
1 5U1'
0327 UT
031,: UT
t
7)
Plate 4. An example of double oval coupling. The most poleward are system splits and a streamer moves
equatorward to the main UV oval where anAAF onset occurs [from Elphinstone et al., 1995b]. Particle regions
based on neural network particle regions are shown in the bottom right. 0, void; 1, BPS; 2, CPS.
on closed field lines. It was a source for auroral kilomet- this sense they do not necessarily exist independently of
ric radiation very similar to that associated with night- the above modules but do appear at such a wide variety
side arcs, and the electron precipitation showed that the of times and locations that they cannot be easily labeled
fluxes of trapped particles exceeded the field-aligned as occurring specifically at some point in a substorm's
ones up to energies of 22 keV [Eliasson et al., 1987]. As development. These modules, as we shall see in section
mentioned in association with Figure 7 this result has
profound implications for the interpretation of the polar
region bounded by the auroral oval. In general, this area
is probably not representative of the region of open field
lines (see also section 3).
Since polar arcs are one manifestation of more general
convection and mapping considerations, they play
an important role in understanding substorm development.
As such, they have been given a primary module
status.
5, help give clues to the mechanisms operating during a
given cycle of auroral activity. We have included many
types of modules in this category, since past experience
has shown that focusing on certain phenomena and
excluding others can prove misleading for future studies.
Vortex streets, traversing arc systems, and poleward
arc undulations. Anger and Murphree [1976] noted
that an arc system can exist at the poleward edge of the
substorm bulge which extends into the morning sector.
These bridging, or traversing, arcs have some interesting
2.2. Secondary Modules
characteristics which warrant them being associated with
These features usually occur after, or in association a separate module. They can brighten at about the time
with, one of the primary modules mentioned above. In the double oval begins to form, and these arcs have

196 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
elements which could be interpreted as poleward leaps
from a scanning photometer viewpoint. The vortex street
which often accompanies the brightening has spiral
forms which could traverse across the scan of a photometer
and give the impression of a rapid poleward leap.
These vortex streets can also form wavelike undulations
on the poleward boundary (see the bottom right of
Figure 12a).
Explanations for these brightenings must be able to
account for coherence which occurs over several hours
article by Oguti [1981], the arc splitting associated with
Sergeev's elementary poleward progression of the substorm
bulge is outlined in detail. The basic time evolution
is shown schematically on the right of Figure 13a
(downward represents increasing time in Figure 13a, and
the two arrows illustrate the clockwise sense of development
of the feature). In this same article, Oguti describes
a process whereby the most poleward oval of the
double oval splits (using current terminology), and an
auroral streamer is initiated which drifts to the main oval
local time and vortex streets which last for less than 1 and begins an event there which results in pulsating
min [Murphree and Elphinstone, 1988]. Differences in diffuse aurora. This same process can also initiate AAF
Alfven transit times to the ionosphere over this broad events, as seen in Plate 4. Oguti et al. [1988] describes
local time range make this a nontrivial observation. If these features as an essential part of the expansion of the
these forms are in some manner associated with the bulge. They showed these could be related to rapid
plasma sheet boundary layer (PSBL), then the 1-min magnetic field changes at geosynchronous altitude.
coherence puts a severe constraint both on formation These perhaps are the signatures of disturbances propand
propagation to the ionosphere. It therefore seems agating earthward initiated by reconnection deeper in
more likely that these are related to instabilities which the tail. Perhaps they are related to the earthward propform
closer to the auroral ionosphere but which are agating "protoplasmoids" observed in the midtail by
ultimately driven from a process in the plasma sheet Moldwin and Hughes [1994]. Alternatively, these may be
boundary layer (PSBL).
related to the bursty bulk flows seen in the plasma sheet
Detached arc. These forms were first reported by [Baumjohann et al., 1990; Angelopoulos et al., 1992,
Anger et al. [1978] using imaging from the ISIS 2 satellite. 1994].
They occur in the afternoon and evening sectors equa- The right side of Figure 13a also shows the clockwise
torward of the main auroral distribution. Later work motion of the developing feature (viewed in the direcshowed
them to be associated with harder precipitation tion of the magnetic field). This is a general characterthan
is normally seen in the diffuse aurora [Wallis et al., istic of the discrete aurora in the evening sector and is
1979], and they were found to occur 10-12 hours after a probably associated with the electric field convergence
peak in the AE index [Moshupi et al., 1979]. Their (and thus a clockwise flow reversal and an upward fieldgeographic
longitude dependence [Anger et al., 1978] aligned current) which one expects to find near the
supports the view that they are associated with weak auroral arc system. Note, however, that the developing
diffusion of electrons into the loss cone [Wallis et al., auroral form has the reverse sense of winding. This
1979]. A schematic of such a form is shown in the lower counterclockwise auroral form has been called the auleft
of Figure 12a. This low-latitude system can pulsate roral spiral by Davis and Hallinan [1976]. This feature
with periods in the Pc 5 range. It does this in conjunction occurs throughout the auroral distribution and can be
with auroral pulsations within the dayside oval. Thus this found at nearly any local time and at most activity levels.
feature may represent a coupling between processes The particular one that occurs at the west tip of an
occurring at the magnetopause and those in the inner expanding substorm bulge has historically been termed a
magnetosphere. One such process might be the cavity "westward traveling surge." This terminology is perhaps
mode resonance suggested by Kivelson and Southwood no longer useful, since this form frequently is not alone
[1985]. In order for this to be viable, field lines in the in occurrence. Thus for ill-defined substorm bulges it is
near-Earth region need very long resonant periods (i.e., difficult to decide which spiral is the WTS. Also, since a
high mass density in order to increase the Alfven transit spiral is not always found at the westward termination of
time). This would be consistent with associating them a substorm bulge, this is not a good argument for giving
with regions of detached cold plasma [Wallis et al., 1979]. this spiral a special designation.
Another mechanism might be one proposed initially by The term "spiral," on the other hand, represents a
Coroniti and Kennel [1970] and modified by Perona generic feature of the aurora and has along with it a
[1972] for sudden commencement events. Perona [1972] probable means of generation. Hallinan [1976] has prefound
that it may be possible to get longer-period pre- sented the essence of an important ideawhich seems to
cipitation oscillations resulting from the interaction of be a good explanation of these forms. The magnetic
very low frequency (VLF) waves with the magnetic field vorticity and shear associated with an upward fieldcompressions
associated with the solar wind variations. aligned current is in an anticlockwise sense when viewed
Bulge evolution, streamers, curls, and spirals. in the direction of the magnetic field. Hallinan [1976]
Plate 4 shows how a streamer can couple the two ovals of showed that for an upward field-aligned current sheet
the double oval distribution. The left of Figure 13a with perturbations at one end (the equatorial plane), the
shows how auroral streamers break off from a develop- perturbations will grow along the field line, resulting in a
ing spiral form [Nakamura et al., 1993]. In an excellent large spiral form in the ionosphere. The convective flow

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 197
D
E
quiet auroral arc
'"'>=:':':': :-:-:->:-: .>--->--'.-:q-:-:-'.-'.-:-:::
::-' 5:;.-:: :5-:-:-:-:-:-:-:-:-:-:-'-':
'."-":-':: 5-': ': ': '" '"' :'"; :; ':' x .: ;.'-'
"'" 111. ====================================
..................
------: ::i!. ::ii::..===========================================
MAY 18 1971 SYOWA
Figure 13a. The development of spirals, streamers, and arc splitting: (left) from Nakamura et al. [1993] and
(right) from Oguti [1981].
around the sheet is clockwise, consistent with the con- existence of two different vortex structures. These can be
vergent electric field usually needed to support an up- distinguished from one another both by their scale sizes
ward field-aligned current. Because of the vortical na- and their sense of wrapping up. Figure 13b, from Davis
ture of these forms, the Kelvin-Helmholtz instability [1978], illustrates the primary differences between these
(KHI) is frequently invoked to explain them, and a basic auroral forms. The rotation sense of a spiral can be
substorm theory has even been generated around them explained by magnetic vorticity, whereas the reverse sense
[Rostoker and Eastman, 1987]. These features however of a curl is probably due to electric field instabilities.
probably have more to do with the upward field-aligned Pulsating aurora. This form of aurora is linked in
currents and magnetic vorticity than to do specifically some respects to the module described above. The
with the KHI.
north-south structures which begin at the most poleward
Another auroral form which winds up in the same oval move equatorward and according to Oguti [1981,
sense as the flow (clockwise, viewed in the direction of 1992] eventually turn to auroral pulsations. Streamers
the magnetic field) is the auroral curl. A point of con- therefore play an important role in understanding the
fusion concerning auroral forms arises because of the transition from discrete to diffuse pulsating aurora. The

198 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
'sf ' CURL
1984, 1986]. (One would expect differing travel times
from the equatorial plane to result in much longer delay
times.) The magnetic pulsations instead appear to be
associated with conductivity variations. Good support
for this can be seen in the simulations of real data by
Oguti and Hayashi [1985].
Oguti [1992] has noted that numerous misconceptions
exist regarding auroral pulsations. The periods vary considerably
even though the statistical average is about
10 s. The fast pulsations are not actually oscillations in
DIA. 20-I, FOOKI
luminosity but instead are rapid changes from one state
(off) to another (on). These forms of pulsation activity
ß
do not show a latitudinal variation of the frequency (with
5P/RA L
higher frequency at lower latitudes) on a case-to-case
basis even though such a relationship seems to hold
Figure 13b. The difference between spirals and curls, from
Davis [1978]. Reprinted by permission of Kluwer Academic
Publishers.
statistically. The periods do, however, vary according to
activity level. The higher the level (and thus the aurora
is occurring at a lower latitude), the shorter the period of
the pulsation consistent with the above mentioned statistics.
Figure 14a is a general schematic of where pulsating
aurora and omega bands are found under active
wide range of pulsating aurora types has been docu- conditions [from Oguti, 1981]. Note that the pulsations
mented and defined by Yamamoto and Oguti [1982] and are found in a region associated with the main 0val.
Oguti [1981, 1982]. The first work of relating optical Omega bands. Omega bands are also found on the
pulsations to magnetic ones was done by Campbell and equatorward distribution, generally in the morning sec-
Rees [1961]. This relationship varies greatly from case to tor. They drift with an eastward speed approximately
case possibly because of the fact that magnetic pulsa- equal to the E x B convective flow in the morning sector
tions represent an integrated effect from a large region, [Opgenoorth et al., 1983; Andre and Baumjohann, 1982].
whereas the optical pulsations are related to small-scale Buchert et al. [1990] have shown that a redistribution of
patches which pulsate independently of one another field-aligned currents from an east-west sheet into pole-
[Oguti and Hayashi, 1984].
ward tonguelike structures is sufficient to explain the
Precipitation associated with pulsating auroras does magnetic pulsation observations (Ps 6 type). As disnot
have a peak in flux at a particular energy but instead cussed by Baumjohann and Kamide [1981], this east-west
has a broad high-energy tail (with energies of tens of
-1
current sheet and its drift (at speeds of up to 1.5 km s
keV [Bryant, 1969; Sandahl et al., 1980] but which can eastward [Elphinstone et al., 1995a]) is probably associalso
be as low as 2 keV [Yau et al., 1981]). Energy ated with drift precipitation of energetic electrons.
dispersion which is observed with these forms implies an Buchert et al. [1988] noted that the neutral wind shear
equatorial source region [McEwen et al., 1981], and flow mechanism proposed by Lyons and Walterscheid
other observations support an association with VLF [1985] does not yield sufficient phase velocities to explain
wave bursts [e.g., Johnstone, 1983]. They occur at any Kp the observations.
level and 95 to 100% of the time for local times after This topic is not entirely independent of the previous
0400 MLT [Oguti, 1981]. Similar to AAFs, they drift module since omega bands also involve pulsating aurowestward
in the evening and eastward in the morning. ras (see Figure 14b, from Oguti et al. [1981a]). Oguti and
They divide into two latitudinal bands after 0400 MLT Watanabe [1976] noted that a form of pulsating aurora
for Kp values between 2 and 3.
occurs within the omega band. This is a quasi-periodi-
Essentially, two mechanisms have been proposed for cally repeating (10-30 s), poleward propagating, mushthe
pulsating precipitation involved in this auroral fea- room shape. This shape moves poleward at speeds of
ture. Coroniti and Kennel [1970] require compressional 10-30 km s -. As shown in Figure 14, there is consid-
MHD waves to regulate the energetic electrons in the erable structure inside an omega band which has not
equatorial plane. The pitch angle modulations cause been thoroughly studied. It is not clear what causes the
VLF wave activity to grow which then causes diffusion of two separate regions of an omega band, one of which
electrons into the loss cone. The second mechanism does pulsates and has striations. These latter features (multinot
require the active participation of the MHD waves ple arcs and striations within the diffuse aurora) may be
[Trefall et al., 1975]. The MHD wave hypothesis may related to a mirror instability [Chiu et al., 1983].
only be supported for some types [Oguti, 1981]. For Omega bands are also closely tied to the AAF events
others it is clearly not operative, since the magnetic discussed in terms of auroral breakup [Elphinstone et al.,
pulsations and auroral pulsations have ionospheric delay 1995a]. These connections therefore give omega bands a
times relative to one another of less than 1 s [Oguti et al., new sense of importance, since understanding them may

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 199
-- .,"''
/'. ' "-
' '-'--'""' '5\
/ ;,',\
EQUA TORWARD
PROPAGATION
'qto, ',, !
' "' ' ;/" ff -t'/
FLICKERIN // 0-" l / ISTREAMiN G
VARIATIONS

200 ß Elphinstone et al.- MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
b
':':,.:::.:f'' ... :.. .":': .....
.' ":.":

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 201
1995c]. A simple analysis of the pulsations in the lobe occurrence during auroral activity. These can move
region showed that the source of these must have been equatorward [Persson et al., 1994; Gazey et al., 1995] or
located tailward of XOSE = --28 RE. The conclusion poleward [Nielsen et al., 1993] (Plate lb) during growth,
reached by Elphinstone et al. [1995c] was that these expansion, or recovery. These parallel arc systems are
pulsations were driven by a reconnection process in the frequently found westward of the substorm bulge. Remidtail
to deep tail and were possibly associated with a ports of these features date back at least to work by
plasmoid leaving the magnetosphere around this time. Anger and Murphree [1976].
Figure 15b illustrates the subsequent development of It is now known that at least some of these features
this surge from 0818 to 0830 UT. At 0818 UT a station- are associated with Pc 5 magnetic pulsations as they
ary surge form can be seen just to the east of 2200 MLT propagate poleward and fade, and new ones appear
(top left). At 0830 UT (bottom right) a large-scale sys- from further equatorward [Nielsen et al., 1993]. It has
tem which might be considered to be a surge form exists been proposed [Elphinstone and Hearn, 1993] that these
well to the west of 2200 MLT. This pattern resembles the features are related to a disturbance moving tailward,
schematic in the middle of Figure 12a. One could con- generating alternating sheets of field-aligned current.
clude on the basis of only these two images (top left and The propagating source and the Alfven transit time
bottom right), that the surge form in the upper right effects can give rise to these and other forms such as fan
propagated to the west, and a north-south feature de- arc systems. Phase mixing of these waves results in the
veloped from the most poleward oval and drifted equa- arc systems approaching one another over time. Thus a
torward. This is not what occurred (see below), and so mechanism with fixed wavelength in the magnetosphere
the dynamics of this event would be completely lost if (and fixed period pulsation) can manifest itself in the
one had only these two images to study.
ionosphere with a wide range of arc separations and
One important form to follow in the sequence is the widths. The width of the resulting ionospheric arc sysnorth-south
feature which is clearly seen in the lower tems is limited only by whether or not one can distinrow
of Figure 15b. In the middle left panel at 0823 UT guish different Alfven transit times between the arcs. If
this north-south form was actually part of the auroral one can, the phase mixing continues, and the arcs will
surge and formed a connection between the main UV
oval and the more poleward system. Thus a portion of
the surge pattern was left behind as the surge form
continue to narrow.
disintegrated and gave rise to the north-south-aligned 3. THE AURORA AS A MAGNETOSPHERIC
structure. Another auroral form to follow in this event is
the one which began to the west of the auroral surge at
0821:32 UT (top right). This form was also north-south
ROAD MAP
The aurora is one of the few means by which large-
scale ionospheric dynamics can be monitored. Further-
aligned but protruded into the region which 3 min before
would be normally considered as an open field line more, using these data, we can accurately identify the
region. This form rotated in a clockwise sense (right- ionospheric location of substorm onset. It is important
hand side when viewed in the direction of the magnetic for us to be able to place these observations into a
field) and formed a bridge across to the surge form by magnetospheric context so that they can be compared to
0827 UT. Thus this development was initiated to the high-altitude local observations of the magnetospheric
west of the surge form, and the progression was to the substorm. An essential aspect of this involves the queseast
until the bridge was complete. The north-south tion of where auroral arcs map to in the magnetosphere.
feature is a marker of the previous link that the station- This mapping can be done in several ways, each of which
ary surge made between the main UV oval and the more has advantages. One commonly used method involves
poleward system. Note also that the dynamics of the the tracing of individual field lines using empirical magevent
include the multiple arc systems to the west of the netospheric models (see work by Jordan [1994] for a
surge (see observation at 0821 UT).
review of these models). For reasons outlined below, this
These observations put into question the existence of is not generally a productive exercise unless the tracings
open field lines to the west of the surge [see Murphree et are done within a more global context. Global maps
al., 1993], since in only 2-3 min this region can be filled from these models can be used with some success so long
with discrete arcs which have been shown in other cases as the researcher does not take the results too literally.
to have a PSBL source. One can only speculate that The ideal use of such mapping is to combine it with
these stationary surge forms (SSFs) are very different global data such as global auroral images and determine
from the normal smaller-scale, westward traveling surges what meaningful relationships can be deduced. Subseen
from the ground. These stationary surge forms may storms inherently deal with highly dynamic conditions
in fact link near-Earth processes to those occurring for which statistical models offer only a broad base of
deeper in the tail.
interpretation. Mappings dealing with substorms must
I. atitudinally propagating multiple arcs. Multiple therefore explain how these dynamics are incorporated
arc systems in the evening sector, such as those illus- into the use of these models.
trated in the bottom right of Figure 12a, are a common Another primary means of mapping involves inferring

202 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
magnetospheric source regions from low-altitude parti- magnetosphere's minimum B surface (approximately
cle observations. These observations combined with the- the equatorial plane) along field lines into the ionoory
allow inferences to be made concerning the source of sphere (adapted from Elphinstone et al. [1991a, b]). The
auroral breakup. A technique which is just beginning to distribution of features was chosen to resemble known
be extensively used is the detailed comparison between auroral forms which exist during substorms. Plates 5a
high- and low-altitude observations combined with ei- and 5b show mappings using the highest magnetic activther
or both of the techniques listed above.
ity version of the T87 and T89 models, respectively.
This section outlines a potential "road map" between The applicability of the rules stated above can be seen
the ionosphere and the magnetosphere which can be most easily by looking at the red and white features in
used by researchers to understand the diverse observa- the right of Plate 5a (near Xosa = -50 Rr ). Moving
tions given in the previous section. This general road from red to white (i.e., closer to the magnetopause)
map has the advantage that a specific case does not moves one closer to local noon in the ionosphere (left
necessarily need to be actually mapped because on a panel between 70 ø and 80 ø in the dawn sector). Moving
large scale it has features which identify it with parts of earthward of these points (i.e., to the region near Xsa
the road map.
= -10 Rr) takes one equatorward in the ionosphere.
Two additional rules are useful when considering map-
3.1. General Mapping Principles
ping:
General principles can be used when mapping using 3. Magnetic flux is conserved between the ionoan
empirical magnetospheric model. These principles sphere and the magnetosphere so that locations with low
are based on two important properties of the magneto- B fields in the equatorial plane map to narrower (latispheric
magnetic field that distinguish it from a simple tudinal) structures in the ionosphere than do regions of
centered dipole field. First, this field is confined within higher magnetic field (if they have the same longitudinal
the magnetopause. Second, the magnetospheric field is extent).
drawn out into a long tail on the nightside. In a com- 4. Regions which are associated with low B fields
pletely closed magnetosphere (no normal component are unlikely to be modeled accurately in global statistical
through the magnetopause) all points on the magneto- models. These regions therefore are the most likely to
pause map into the northern and southern ionospheres change their projections to the ionosphere with different
to two points called the northern and southern cusps. As models. For the same reason, we can expect large varimagnetic
field normal to the magnetopause is added, the ations in the projections of these regions as a consemagnetosphere
becomes more open (i.e., some field quence of changing magnetospheric conditions.
lines become linked to the solar wind) and the iono- These rules have obvious implications when one looks
spheric cusp region becomes broader. We know through at the width of the deep-tail region in the ionosphere
studies of cusp precipitation that the magnetospheric compared with the near-Earth region (compare the two
cusp regions map to ionospheric areas confined to about regions near midnight in the left panel of Plate 5a). The
2.5 hours of local time centered about magnetic noon high-latitude region should also be more variable than
[Newell and Meng, 1992]. This result implies that for the the equatorward area and will be more model sensitive.
purposes of this paper the real magnetosphere resem- Plate 5b shows that using a different magnetic field
bles a closed type of magnetosphere. In this relatively model (based on the same data set but different funcclosed
type of magnetosphere two rules can be used in tional forms to represent the current elements) gives
mapping forms to and from the magnetosphere [Elphin- results which are in many ways qualitatively similar (on
stone et al., 1994b].
a large scale) to those in Plate 5a. A closer examination,
1. Closer to the magnetopause translates in the ion- however, shows a multitude of differences which would
osphere to closer to local noon. This rule indicates that profoundly affect local field line tracings particularly at
in contrast to a simple dipole field mapping, features at high ionospheric latitudes. Field-aligned currents can
a constant local time in the magnetosphere do not follow also alter mapping by an hour of local time or more
a constant local time in the ionosphere (i.e., field lines [Donoran, 1993]. Observationally, geosynchronous mapare
not lines at constant longitude).
pings can change by 300 to 400 km [Oguti et al., 1991b].
2. Closer to the Earth at a fixed distance from the These changes are comparable with changes resulting
magnetopause results in an equatorward displacement from the more basic choice of model current functions.
in the ionosphere.
One example of the differences can be seen in the
These rules result in mappings quite different from projection to the ionosphere of the feature along Xos M
what one would achieve using a dipole field, primarily = -26 Rr near the dawn magnetopause. In the T87
because of the distortions introduced by the magneto- mappings it strongly resembles one element of the fan
pause and tail currents. Empirical magnetospheres such arcs seen in Figure 9, while in the other (T89) it lies
as the Tsyganenko [1987, 1989] models tend to have this several hours westward and closer to the more equatorclosed
type of configuration. (The two Tsyganenko mod- ward arcs. In both cases, though, features exist resemels
will be called T87 and T89 in what follows.) To bling fan arcs. Bim et al. [1991] have shown that when B z
illustrate this, Plate 5 shows features projected from the is large along the flanks of a magnetosphere, closed field

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 203
NLAT
ML
' 'MLAT
1 MLT
0 MLI"
IONOSPHERE
1'87 SEPTEMBER 23 1 ß :. 21UT Kp- 5
60 MLAT
: ' LAT
,18 ML "ML¾ ,
IONOSPHERE
o MLT
..,...,,-'
T: ß SEPTEMBER 23, 1 ß:. 21UT Kp= 5
0
VGSM Re
30 0 -9O
YGSM (Re)
Plate 5. (a) A mapping of large-scale features (right) from the equatorial plane to the (left) ionosphere
using the Tsyganenko [1987] (T87) Kp 5 model (revised from Elphinstone et al. [1991a]). (b) Same as for Plate
5a but using the Tsyganenko [1989] (T89) model. The same equatorial locations have been mapped. Large
differences appear primarily at high latitudes in the dawn and dusk sectors.

204 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
lines extend to higher latitudes in the dayside sector. We
see that these deep-tail flank regions project to features
resembling polar arcs and fan arcs when B z is large (i.e.,
the T87 model has this characteristic in its magnetic field
configuration). Elphinstone et al. [1994b] used this reasoning
to suggest that the KHI along the dayside and
flanks of the tail is responsible for these fan arc systems.
This would suggest that when B z is decreased on the
flanks of the magnetotail, polar arcs would tend to
disappear and the dayside polar region would become
more open. It would also explain why polar arcs sometimes
disappear prior to auroral breakup. If this were
true, then this is inconsistent with models which portray
KHI in the LLBL as the source for omega bands, auroral
surges, and auroral breakup [e.g., Rostoker and Eastman,
1987; Rostoker and Samson, 1984]. Here we have an
excellent example of how general mapping principles
can help resolve outstanding questions in substorm re-
search.
5
[] []
20 IMAGES R: .72 688 POINTS F: 734
1/2 HR. AVERAGES 0-24 MLT
SLOPE = .63 (.02)
INTERCEPT = 32 (2)
I I I
60 70 80
OBSERVED OVAL LOCATION(MLAT)
20 IMAGES R: .94 626 POINTS F: 4600
1/2 HR. AVERAGES 0-23.5 MLT
SLOPE: ,87 (,01) D...._.
INTERCEPT: 9.7 (0.8)
5057 BOUNDARIES
USE [] _
Auroral conjugacy also plays an important role in
understanding mapping and the applicability of rule 4,
stated above. Stenbaek-Nielsen et al. [1972] noted that
there are two important types of auroral arcs to consider
when studying conjugacy. In the first type, dipole conjugacy
between hemispheres is maintained. This corresponds
to the equatorward system of electron arcs which
lie within and just poleward of a proton precipitation
arc. The second type lies poleward of the first, and
LI3
5e
I
60
I
7e
I
eo 9
conjugacy for this type can vary dramatically from event
to event. These two types are consistent with the main
OBSERVED OVAL LOCATION[MLAT)
UV oval and the more poleward oval, respectively, and Figure 16. The correspondence of the main UV oval and the
are consistent with rule 4 for the mapping of auroral inner edge (peak) of the cross tail current calculated in the
features.
Tsyganenko [1987] empirical magnetic field model [from El-
Comparing the maps in Plate 5 with the data in Plate phinstone et al., 1991b]: (top) the model open-closed field
2 and the schematics in Figures 9, 10, and 12 enables one
to make fairly obvious connections between some of the
auroral features and potential sources in the magnetosphere.
For instance, the main UV oval is unlikely to
map into the deep tail because of its location, its width,
and its relatively stable nature. To investigate this, Elboundary
generally lies far poleward of the main UV oval;
(bottom) the peak in the main UV oval position can be
predicted quite accurately on a global basis from the peak in
the T87 nightside magnetospheric volume current density projected
to the ionosphere.
phinstone et al. [1991b] measured about 5000 UV Viking When this result was found, it was a great surprise to
auroral boundaries for a wide range of local times, find that the open-closed field line boundary did not
dipole tilt angle, and magnetic (Kp) activity level and seem related to the steady state main UV auroral oval.
attempted to compare them with the T87 empirical Other model parameters were also correlated with the
magnetic field model. Averages of these boundaries over observables, and eventually it was found that the region
0.5 MLT helped eliminate mapping inaccuracies associ- where the peak in the cross-tail current projects to (the
ated with field-aligned currents. The peak in the main general region between Xos M = -4 and -12 Re) cor-
UV oval intensity and the poleward boundary of this relates extraordinarily well with the main UV oval (botmain
UV oval (not including polar arcs and other high- tom of Figure 16). This result indicated that much of
latitude auroral phenomena) were the observational what is referred to as the auroral oval must originate in
boundaries studied. When statistical comparisons were the inner plasma sheet. Elphinstone et al. [1991b] used
made between the peak in the main UV oval (or the this correlation to conclude that substorm onset must
other more poleward observed boundary) and the open- originate close to Earth owing to the onset's arc associclosed
field line boundary (open field lines were defined ation with this peak in the main UV oval.
by field lines exiting the model boundaries before clos- The above correlations strongly support the premise
ing) in the T87 model, it was found that this model that auroral features seen on the main UV oval occur in
boundary generally lay far poleward of the main UV the near-Earth region near midnight and closer to the
oval (top of Figure 16).
magnetopause at magnetospheric local times away from
90

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 205
midnight (XosM from about 4 to 12 Rr). We know that ble explanation for the fading of arcs within the bulge as
the main UV oval maintains its existence throughout the the double oval forms is that this bulge region is initially
substorm, and so its near-Earth source mechanism is not within an auroral source region generated by a nearlikely
to rapidly change. This knowledge allows one to Earth disturbance. As the disturbance (whether a neumake
a similar connection with auroral features occur- tral line or a disturbance front) moves down the tail, the
ring on this distribution without actually having per- source is no longer affecting the near-Earth region, and
formed the specific substorm mappings. For example, so the aurora connected to this region (i.e., within the
the association of omega bands with the most equator- bulge) begins to fade. This near-Earth region no longer
ward oval during a double-oval configuration, combined supports a substorm current wedge, and so the more
with their association with auroral pulsations and VLF equatorward westward ccuojct ^ .... : will also iau,. ' r,;o ß 111o
waves, indicates that omega bands map to the near- could give rise to the broad correspondence between the
Earth region and not out to the boundary layers. This is ground magnetic recovery and the thickening of the
consistent with the omega band mappings shown by plasma sheet.
Pulkkinen et al. [1991b] and is inconsistent with the The above example illustrates that it is important to
mappings put forth by Rostoker and Samson [1984]. understand how changes to the magnetospheric current
Similarly, the variability of the most poleward arc systems can alter the mapping of features from the
system in the double oval (near midnight), its width, as magnetosphere to the ionosphere. Figures 17a and 17b
well as its location, all support its mapping further tail- illustrate how mapping at midnight (in eccentric dipole
ward and its association with the deep tail or at least the coordinates) can be altered via changes to the ring
PSBL (as opposed to the LLBL) [Elphinstone et al., current and via cross-tail current systems in the T87
1995c].
model for two levels of Kp activity (3 and 5). These
The split between the dayside and nightside auroral parametric curves could in principle be used to construct
systems near 1500 MLT can also be understood in this a time-dependent mapping to the ionosphere based on
scheme as being due to a division between processes hypothetical current changes in the magnetosphere. The
originating in the dayside boundary region and those curves at the top of each graph represent the XosM
coming from the nightside. Thus the dayside spirals seen location (left axis) associated with the ionospheric proin
the lower left panel of Figure 9 probably are associ- jection displayed further down (right axis). The ionoated
with the dayside boundary layers, while the arcs spheric points were chosen to be separated each by 1 ø of
which lie further poleward may be connected to separate latitude with the most equatorward point being in all
nightside processes. These are a few examples of how cases the projection of the peak in the equatorial plane
the large-scale auroral distribution can be used in com- volume current density. (This is the parameter which
bination with the maps given in Plate 5. Other geophysi- correlated well with the peak in the main UV oval.) The
cal parameters can then be interpreted on the basis of most equatorward line is therefore an illustration of how
their locations relative to this large-scale distribution. the main UV oval peak might be expected to move
under different conditions. This allows one to then un-
3.2. Temporal Variations
derstand the meaning of the projection of the 4 ø of
Lui [1978] was probably the first researcher to inves- latitude immediately poleward. If this region projects to
tigate the necessary changes to the cross-tail current a wide range in XosM, then 4 ø in the ionosphere spans a
during a growth phase followed by a dipolarization. large portion of the central plasma sheet and the asso-
Since then, time-dependent adaptations to the Tsyg- ciated "oval" is narrow. If, on the other hand, these
anenko magnetic field models have been used to inves- points do not extend far tailward, then the central
tigate the properties of the tail current sheet prior to plasma sheet region will map to a broad region in the
onset [Sergeev et al., 1990; Pulkkinen et al., 1991a]. In ionosphere (i.e., a wide "oval" in the below discussion).
these studies the original field model parameters are If no points exist at a particular value of the parameter,
altered, and new current systems added to try to dupli- then the points do not have an equatorial plane projeccate
observed changes to the magnetic field configura- tion.
tion at several high-altitude satellites. Pulkkinen et al. Figure 17a shows changes resulting from ring current
[1992] have performed this calculation for several case variations, while Figure 17b shows changes induced by
studies and found that the tail current sheet may become moving the inner edge of the current sheet (top), insufficiently
thin to make thermal electrons chaotic late in creasing the strength of the cross-tail current (middle),
the growth phase. One generally needs to increase the and altering the thickness of the current sheet (bottom).
strength of the cross-tail current (by typically a factor of The current sheet thinning employed is from Sergeev et
2) and thin the current sheet down to between about 0.1 al. [1990]. The ring current panels illustrate that periods
and 0.5 Rr. These studies show that statistical models with high ring current intensity (i.e., magnetic storm
are not sufficiently stretched to be used for the mapping time periods) imply that the main UV oval will move
of auroral features prior to onset.
equatorward. The bottom of Figure 17a shows that un-
Temporal variations of the magnetic field topology der fairly normal conditions, increasing the strength of
can help explain some auroral observations. One possi- the ring current will make the oval broad in the iono-

206 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 207
-4
-13
-22
-40
-15
-30
-45
-60
-4
-13
-22
-31
-40
Kp = 3 peak of volume current density varies from 1.29 to 2.37 nA/m2
circles,solid line:AT CURRENT MAX ." ."
:
squares,solid line:CURRENT MAX+I degree : ..'
: ß
solid line only:CURRENT MAX+2 degree : ..'
circles,dashed line:CURRENT MAX+3 degree
:
..
ß
:.
squares,dashed line:CURRENT MAX+4 de f :'
. . .
- 10 -8 -6 -4 -2
xGSM LOCATION OF INNER EDGE OF CURRENT SHEET(RE)
1.00 1.50 2.00 2.50
MULTIPLICATIVE INCREASE IN CROSS TAIL CURRENT TERM
Kp = 3 peak of volume current density varies from 1.46 to 48.2 nA/m2
circles,solid line:AT CURRENT MAX
squares,solid line:CURRENT MAX+ 1 degree
solid line only:CURRENT MAX+2 degree
circles,dashed line:CURRENT MAX+3 degree
squares,dashed line:CURRENT MAX+4 degree
'.;, '. ............... ;; ................ '.-. ................ ;1..
1 2 3 4 5
THICKNESS OF CURRENT SHEET(Re)
3.00
90
74.
90
a66
58
90
" 58
6'

208 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
will naturally broaden the projected region tailward of it.
If the peak in the current sheet is found too far tailward,
then the strong current will generate regions of negative
B z in the tail and a neutral line forms.
storm process. Initially, one might have a double-oval
distribution like that shown in Plate 4 with the discrete
Increasing the strength of the cross-tail current (midaurora
associated with the deeper tail and the diffuse
precipitation equatorward associated with the main UV
oval. Eventually, however, it is this main UV oval which
dle of Figure 17b) will not change the location of the activates in an auroral breakup event, and so this region
main UV oval but will cause the oval to narrow, and can rapidly change character and become associated
eventually a neutral line will form. It is interesting to see with discrete aurora [Elphinstone et al., 1995b].
that changing the thickness of the current sheet (bottom Low-energy (up to 40 keV) ion precipitation can also
of Figure 17b) can drastically change the associated help link the aurora to processes in the magnetotail so
volume current density without dramatically affecting that a mapping can be inferred. The velocity dispersed
the large-scale mapping (the oval will thin somewhat but ion signature (VDIS) is perhaps the most useful of these
not in comparison with altering the current intensity). signatures. There are two places where dispersion signa-
Thus a thin current sheet could develop in the near- tures frequently exist, one at the equatorward edge of
Earth tail which would result in intense local volume the double oval and the other at the poleward edge. The
current densities but which would not significantly alter latter feature has been associated with ion beams (4-20
the mapping of the main UV oval.
keV) precipitating from the PSBL and becoming dis-
One could easily construct a substorm sequence persed in energy as a function of latitude via an E x B
based on these parameterizations. A strong ionospheric drift [Zelenyi et al., 1990]. This feature frequently lies just
growth phase condition could be realized if the inner poleward of the most poleward arc system but can be
edge of the tail current sheet moved earthward and the coincident with this arc as it is just developing [Elphincurrent
sheet intensified. This would result in a thinning stone et al., 1995c]. This again identifies the most poleof
the oval and a movement equatorward prior to onset. ward arc syste. m (and the auroral activations associated
If the dipolarization at onset then injected and energized with it) with either the PSBL or its inner edge. The other
particles into the ring current, the oval would thicken VDIS lies at the equatorward edge of the main UV oval
and the main UV oval would move further equatorward. and is a low-energy (less than a few keV) ion dispersion
As the ring current intensified, the active auroral region signature (LEIDS). Sauvaud et al. [1981] explain the
would become broad. With the decay of the ring current LEIDS as being associated with the drifts of ions relative
the oval would eventually return to its previous config- to an injection source. They may also be associated with
uration. This simple example shows how the curves in upwelling ions at an inverted V event located near the
Figures 17a and 17b can be employed to understand how inner edge of the cross-tail current sheet.
general magnetotail configuration changes can alter the High-energy particle isotropic boundary mappings.
midnight mapping of the auroral region.
Lui and Burrows [1978], Kirkwood and Eliasson [1990],
and Elphinstone et al. [1995a] have all shown that auroral
3.3. Particle Observations and Their Relation breakup occurs adjacent to the energetic (40-80 keV)
to Substorm Mapping
electron isotropic boundary. That is, energetic electrons
Feldstein and Galperin [1985] have given an excellent equatorward of that point are generally trapped (i.e.,
summary of the reasoning used to support the argument moving perpendicular to the field line), whereas polethat
the discrete auroral region is associated with the ward of that point there are equal fluxes along the field
whole central plasma sheet. If one accepts this, then if line as there are perpendicular to it (i.e., isotropic).
the arc associated with breakup is the most equatorward Elphinstone et al. [1995a] has shown that an AAF event
one, then this arc must lie at the inner portion of the took place poleward of the 80- to 250-keV ion isotropic
central plasma sheet. On the basis of high-altitude sat- boundary but equatorward of the location where the
ellite data, Lopez and Lui [1990] also argue auroral number of these ions began to alter rapidly with latitude.
breakup is initiated in the near-Earth tail. In order for The former location probably maps to where ions are
this to be true, discrete auroral arcs must have their scattered into the loss cone by motion within a curved
source in this region just prior to substorm onset. magnetic field geometry [Sergeev et al., 1983, 1993], while
As is frequently done, Sergeev et al. [1991] have inter- the latter marks the location where the plasma pressure
preted a double-oval configuration in terms of the most rapidly begins to increase near the Earth. These obserpoleward
system being discrete and mapping to the vations precisely pinpoint the onset location relative to
PSBL, and the more equatorward system being purely specific electron and ion precipitation boundaries. Note,
diffuse precipitation. One reason researchers argue however, that this only places onset near the Earth
about this near-Earth onset location is that the more (within a region of large magnetospheric pressure graequatorward
region, where onset will eventually begin, is dients) in a region tailward of where ions begin to be
frequently of a diffuse auroral character (e.g., see Plate scattered by the magnetic field configuration.
9 of Elphinstone et al. [1993b]). These contrasting views Space weather, MHD predictions, and mapping.
of where arcs map to initially seem irreconcilable but in The ability to predict substorms and conditions at geofact
can be explained through the dynamics of the sub- synchronous satellites can in principle be a means to

34 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.- MODULAR AURORAL ACTIVITY ß 209
;. ?..'.,:...'. :,I( '..-'".!:'
....... ;:-,;i:.!%.,:. .... .....,r-'-,½::, ...:;:½- ,1 :!.
.
21 "/
,..
1H:" 19 LIT 100541 UT
:: :'... . .:½'
.: .;:.;::;::::::::;,, .:: ......... :.. ; ..... /.::; ...:.. ß
;..
: :e.r:?f;-: ..
.:
.: WT$
.
;. 1' ;'. ,,
101224 UT
Figure 18. An auroral substorm event studied in detail by Hones et al. [1987]. At about the time ISEE
records the recovery of the plasma sheet (bottom) a westward traveling spiral develops on the poleward edge
of the aurora. Forty-five degree field of views at (left) Inuvik, (middle) Fort Yukon, and (left) Barrow are
shown as circles. MLATs 68 and 72 are shown along with 2100 and 2200 MLT. Observations at Barrow could
be misinterpreted as a poleward leap of the aurora at this time.
reduce satellite malfunctions and prevent electrical
power blackouts associated with the large substorminduced
currents (see introduction). To do this, of
course, means that precursor events must be identified
either at the Sun or in the solar wind, the magnetosphere,
or the ionosphere. This in turn will eventually
require an understanding of the onset mechanism for
the activity and a real-time means of predicting it.
A first step toward this goal has been made by Fedder
et al. [1995], who have shown that an MHD simulation of
the magnetosphere driven by observed solar wind conditions
can reproduce many of the important features of
the auroral distribution prior to substorm onset. The
event chosen for study is the one given in Figure 5. In the
process of accurately reproducing some of the features
of auroral growth phase, Fedder et al. have independently
helped substantiate the view that the auroral oval
lies deeply embedded within the closed field line region,
that onset is initiated within closed field lines, and that
polar arcs lie adjacent to the open-closed field line
boundary. This predictive aspect of space physics should
be a major goal for research in the International Solar
Terrestrial Program (ISTP) period and will no doubt
have spinoffs which will help resolve conflicting substorm
theories as well as answer many important mapping
questions.
3.4. High-Altitude Comparisons With Auroral Data
General. Many authors have found that ionospheric
substorm signatures can be seen on the ground
while not at geosynchronous orbit. Singer [1983] noted
this for the occurrence rate of Pi 2 signatures of onset.
About 10% of these types of event are seen at low
altitude (and midlatitude) but not at high altitude, and
vice versa [Yeoman et al., 1994]. Sauvaud et al. [1987]
found that geosynchronous satellites do not always register
pseudobreakups, and Lui et al. [1976] also found
this to be the case for a contracted oval event. Onset is
therefore not always within geosynchronous orbit or at
least can be a localized phenomenon at these radial
distances [Nakamura et al., 1994].
Lopez et al. [1993] have shown that there is a reasonable
correspondence between the poleward expansion of
the auroral activity and the tailward expansion of the
current disruption region in the vicinity of the CCE
satellite (apogee at 8.8 RE). Scholer et al. [1984] have
given an example of a poleward leap of the electrojet
approximately coincident with the recovery of the
plasma sheet both at ISEE 1 and ISEE 3.
Double-oval spiral events, substorm recovery, and
fast flows in the PSBI_. Figure 18 shows a developing
auroral bulge for an event studied by Hones et al. [1987].

210 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
In the first four observations the substorm bulge develops
and is filled with auroral emissions. Consistent with
Figure 12a, as the bulge reaches its most poleward
extent, the emissions within the bulge fade (middle row
of Figure 18) and the local magnetic recovery begins.
After this time, and more or less coincident with the
recovery of the plasma sheet at the ISEE satellite, an
auroral spiral develops near Barrow station (Figure 18,
bottom). This event is an example of the type of activation
given on the left side of Figure 12b and in this case
occurs along with plasma sheet recovery. The magnetic
signature at Barrow was consistent with the passage of a
"WTS." These spirals associated with the thickening of
the plasma sheet may be related to the results of Lyons
and Huang [1992], who found isolated H bay activity
associated with the recovery of the plasma sheet. In that
study, four events were used. Two of these had activity
immediately preceding the event in question, one had no
information in the critical local time sector 2100 to 2400
MLT and one had significant activity in the previous
4 hours. If these events did occur in true isolation from
JII © out
JII © in
(a)
Y, Z plane
at X -70 R e
other auroral breakups, then they would be very strong
support for the modular concept. If, on the other hand,
they occurred in association with other activity, they still
support the separate activation of events near the poleward
boundary of the aurora.
Figure 19, from Bythrow and Potemra [1987], illus-
(b)
trates the view that surge forms are distortions occurring
near the open-closed field line boundary. The spiral
Figure 19. A schematic showing one view of how surge forms
may map to the magnetotail [from Bythrow and Potemra, 1987].
events mentioned here are likely to be of that nature.
Simulations by Yamamoto et al. [1993] show that these
spirals can be explained if they occur adjacent to the Plate 6 shows coincident observations in the northern
open-closed field line boundary. Note, however, that this
ionosphere. The white line near 0100 MLT shows the
is not the entire story, since many of these spiral forms mapping of the entire plasma sheet tailward of the Xos M
can occur deeply embedded within the closed field line position of the IRM satellite. This entire region was
region (see the right side of Figure 12b).
mapped to the ionosphere in order to convey that the
These spiral events can also be identified with fast precise mapping is not important for this case study. At
flows in the PSBL. Figure 20, from Baumjohann [1988], the beginning of the event (2228-2246 UT) a double
shows an example of a IRM satellite fast flow event seen oval with no intensifications on it can be seen. By 2253
in the PSBL at high altitude in the midnight sector. This UT (Plate 6, top right) the double oval has activated in
event took place during a quiet magnetic activity interval the general projected vicinity of the fast flow occurrence.
when the AE was less than 100 nT. It occurred after the This gradually develops into two "boundary" region
AE had recovered from a slightly active period (200-300 spirals (third row) similar to those shown on the left of
nT) at about 2000 UT on May 3, 1986. This recovery was Figure 12b. They have faded completely by the last panel
consistent with the plasma sheet recovering at the ISEE at 2312:58 UT. This case illustrates a correspondence in
1 satellite at that time.
both time and space between the fast flow events seen at
On a longer term, the AE had just recovered from a high altitude and the activations of the most poleward
major storm period with the AL index reaching about oval of a double-oval configuration. To the best of our
-2000 nT. The Dst index was recovering from a storm in knowledge, this is the first time that such a double
which the Dst reached -80 nT and was greater than -40 correspondence has been made. If the fast flow were due
nT (i.e., recovering) at the time of the observations. This to a reconnection event this is again consistent with the
event was used by Baumjohann [1988] as a quiet time scenario outlined above whereby reconnection in the tail
fast flow observed in the PSBL. The PSBL was encoun- supports the spiral generation at high latitudes in the
tered at about 2215 UT, and a fast flow of more than 400 ionosphere.
km s - earthward began at 2245 UT and lasted until Further support for the poleward oval being associabout
2305 UT (other fast flows occurred, but these ated with reconnection can be found in the event decannot
be compared with low-altitude data).
scribed in section 2.2 (under "stationary surge") and
Noon

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.- MODULAR AURORAL ACTIVITY ß 211
AMPTE/IRM :3 MAY 1986
-,,],,i,,i,,[,,i,,i,,i,,i,,i,,i,,i,,i,,i,]l,,I ,, _
_
25--
: E
_
190
_
- -
- ""- '&'/'=- GSM
_
B 180:
170 - , ,.,- "-- ,[ - _- -
_
k B o 5 GSM
p 1o-2 10--1
6OO
200
0
-200
v{ -4o0
-600 ,,
UT 21 00 22'00 23'00 0:00 1 00
17 94 16 29
ZNS -5 91 -3.87
LT 0 16 ]00 A 200 0 50
Figure 20. A fast flow event (2245 to
2305 UT) in the plasma sheet boundary
layer (PSBL) seen at the Active Magnetospheric
Particle Trace Explorer
Ion Release Module (AMPTE/IRM)
during quiet times [from Baurnjohann,
9881.
shown in Figure 15. This association of the most pole- ciated more equatorward aurora fades. Eventually, in
ward arc system with a resonant phenomenon in the the midtail region, reconnection occurs, and at some
outer plasma sheet may help explain the heating of the point the last closed field line reconnects. In the ionoplasma
sheet seen after onset by Huang et al. [1992]. sphere this may correspond to poleward oval activations
Many of the heating events from that study occurred and a new electrojet forming at this higher latitude (i.e.,
after a second intensification had occurred.
the traversing arc). This might then trigger the recovery
Slavin et al. [1989] have investigated two events when of the plasma sheet as the neutral line begins to retreat
auroral images could be compared with deep-tail obser- in response to the energy released at this time. This
vations by ISEE 3. They found the results to be consis- magnetospheric scenario explains why the fading of the
tent with the plasmoid model for substorm tail dynamics. more equatorward electrojets (along with the intense
Note that in both cases the observations of a plasmoid precipitation there) could occur at about the same time
occurred in association with the formation of what is as the most poleward system is activating with a new
now termed the double-oval distribution (see their electrojet. This view would have the reconnection of the
Plates 1 and 2). The second event occurred after a time last closed field line occurring 5-30 min after the exploof
previous activity when the most poleward system was
still active. Previous to this, several traveling compression
regions were observed (perhaps the signature of
sive onset.
plasmoids seen in the tail lobes). It may be that these 4.
features were signatures associated with intensifications
IMPLICATIONS FOR SUBSTORM MODELS
on this poleward boundary.
We have seen that the wide variety of auroral activity
These types of magnetospheric events may be implies that no single mechanism or theory is likely to
strongly related to auroral activations found on the most explain everything. One of the fundamental problems
poleward oval of a double-oval configuration as well as still existing in magnetospheric physics is to determine
to the fading of emissions equatorward of this activating the mechanism of substorm onset. There are a variety of
arc (see also Plate 3). As mentioned in the module existing theories, each one of which is purported to give
section, these events can occur as the bulge reaches its the correct answer to this question. One would think that
most poleward extent. In the magnetosphere the iono- sufficient computer time, along with the knowledge of
spheric poleward progression of the bulge can be the requisite physics, now exists to resolve this problem.
equated with the tailward motion of the current disrup- Unfortunately, the magnetosphere involves physical protion
region. As this disruption region moves tailward, it cesses which exist at such a wide range of spatial and
decouples from the near-Earth region, and so the asso- temporal scales such that a single type of simulation is

212 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34 2 / REVIEWS OF GEOPHYSICS
MAY 3, 1': ß
1
2240 to 2246 UT 225311 UT
225613 UT 33 UT 225813 UT
....
23(X)54 UT 230415 UT ,. ß UT
Plate 6. The auroral dynamics associated with the fast flows shown in Figure 17. Spirals develop near 0200
MLT (between 2253 and 2310 UT) on the most poleward system of a double-oval configuration. The white line
near 0100 MLT is the T87 mapping of the plasma sheet tailward of AMPTE/IRM.
unlikely to encompass all of the desired scales. Cross- with the scarcity of in situ measurements makes for a
coupling between processes where individual particle difficult task in evaluating the various substorm models.
motion becomes important and those involving large- Most of the models presented here have originated as
scale magnetohydrodynamic fluid motions, makes con- an attempt to explain a particular set of observations
clusive simulations of the magnetosphere substorm a (usually magnetospheric versus ionospheric). Gradually,
very difficult problem.
however, it is becoming recognized that these models
For example, some theories invoke reconnection as a must merge together and interact to give a proper undriver
for most of the substorm process. These theories derstanding of what actually occurs during an auroral/
are perhaps the most evolved from the viewpoint of magnetospheric substorm. Even if a model of the magbeing
able to generate computer simulations and explain netospheric substorm successfully reproduces the effects
many of the observables. Unfortunately, these simula- seen in the magnetosphere, there is a further problem in
tions rely partly on microphysics, which is introduced translating this into what should be seen in the ionointo
the simulation either.explicitly or via the particular sphere. This problem is particularly evident when one
numerical scheme used (see work by Raeder e! al. [1996] considers that even theories for "simple" auroral arcs
for a description of some of the problems). Other the- are not yet adequate to explain arc observations
ories have tended to focus primarily on an onset mech- [Borovsky, 1993].
anism and so do not have well-developed explanations A start toward achieving a synthesis of models and
for the broader range of observations. This combined how they relate to a classical auroral substorm has been
.....

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 21 3
Ionosphere M aonetosDhere
_
Growth Noon-midnight
Oval boundaries cross-section
move equatorward
Onset Current
Equatorial
projection
Initial arc brightening (decrease disruption
Exoansion
wave
L,,,. '""'"' ' Convection
Convection
surge Rarefaction
wave
Late Exoansion/Recovery_ Neutral line
'
- 80R E
Figure 21. A synthesis view of the magnetospheric causes of the auroral substorm [from Lui, 1991].
made by Lui [1991]. His view is presented in Figure 21. tion could occur independently and drive near-Earth
On the left is a simplified view of what happens in the conditions into an unstable state (or it may not have
aurora as each magnetospheric process (shown on the sufficient strength to do so). Either of these scenarios in
right) becomes active. Kropotkin [1972] initially pro- turn might affect the tail flanks of the magnetosphere
posed that a rarefaction wave propagated tailward after and trigger disturbances in the high-latitude dawn and
the development of a flute instability in the near-Earth dusk ionospheric sectors. Solar wind conditions, on the
tail. He suggested that the tailward propagation of this other hand, can drive disturbances directly there (and on
wave might instigate reconnection in the tail. This con- the dayside magnetopause) and cause "growth" phase
cept was later adopted by Chao et al. [1977] to explain activity to appear at various times. This directly driven
plasma sheet thinning. This later became a key ingredi- activity in turn can create conditions appropriate for new
ent of the synthesis model put forth by Lui [1991]. activations. Some of these possible modular cycles are
The modular view of the substorm would modify this discussed along with their ionospheric signatures in the
model such that different portions of the magnetosphere next section.
could become active independently of the near-Earth On the basis of this modular concept and of a knowlinstability.
Further, a near-Earth instability may not be edge of how the aurora relates to the magnetosphere,
sufficient to trigger a rarefaction wave that moves all the the various substorm models can be evaluated and
way down the tail, or else conditions may not be correct placed in context. This is done below for each of the
for it to invoke reconnection. Alternatively, reconnec- major substorm models.

214 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
4.1. Near-Earth Onset Models
ents necessary to trigger onset. They also demonstrated
These models involve a diversion of the cross-tail that wavelengths less than the ion gyroradius were stacurrent
at the inner edge of the central plasma sheet ble. This latter result was found to be consistent with the
which is associated with some plasma instability. This sharp cutoff in AAF mode numbers associated with
idea dates back at least to work by Atkinson [1967] and substorm onset (see Figure 11). Ivanov et al. [1992] have
has been revived numerous times over the years [Hoff- extended this analysis to the ballooning instability with a
man and Burch, 1973; Lui, 1991]. The current disruption finite plasma [3. Vetoulis and Chan [1994] have detercan
also occur through what has been termed a "current mined that Alfven-ballooning modes are likely to have
sheet catastrophe" [Burkhart et al., 1992]. Tearing modes mode number for peak growth of about 100. This is
instigated near the Earth, as proposed by Buchner and almost exactly the mode numbers found observationally
Zelenyi [1987], have been supported by the near-Earth just prior to onset (Figure 11). This combined with the
field configuration found by Pulkkinen et al. [1991a, observations that AAFs brighten along with onset make
1992] at the end of growth phase.
this instability a strong contender for actively participat-
Local onset models. Some researchers view the ing in some forms of auroral breakup. Shear flow Alfven
current disruption region to be a turbulent area with ballooning may also play a role in the undulations seen
localized (< 1 Rœ) structure [Lui et al., 1988]. The non- at the equatorward edge of the diffuse oval [Vigas and
local nature of the onset region as seen from the ground Madden, 1986].
in AAF events implies that a more global mechanism
must also be active [Elphinstone et al., 1995a]. This
supports the view that if a small-scale local process is
4.2. Boundary Layer Models
involved in the breakup mechanism, then it must be fed LLBL model. The model by Rostoker and Eastman
by a larger magnetotail instability [Hesse and Bim, 1993] [1987] has inherent in it the topological connection
or the microinstability must somehow trigger the other between the LLBL and the PSBL (see their Figure 3).
larger-scale one.
This model is the least developed of the substorm mod-
Cross-field current instabilities. Lui et al. [1991] els and invokes the KHI as the governing mechanism for
suggested that a cross-field current instability is respon- substorm onset. According to this scheme, the low-altisible
for auroral breakup. As noted, there may be more tude boundary plasma sheet (BPS) and central plasma
than one mechanism involved. This may be a candidate sheet (CPS) precipitation signatures correctly identify
for certain forms of near-Earth onsets, although it can their high-altitude counterparts [Eastman et al., 1988].
only play a partial role for AAF-type onsets, since these Most researchers now agree that this cannot be the case.
events almost certainly also involve a more large-scale In this model the KHI serves as the mechanism to
(probably MHD) wave instability. The association of the produce the vortex streets seen in auroral data. As noted
main UV oval (and hence the breakup arc) with the peak by Lysak et al. [1995] and Elphinstone et al. [1993a], the
in the cross-tail current [Elphinstone et al., 1991b] lends sense of auroral spirals is opposite to the sense found in
some support to this model.
the KHI flow but consistent with the sense found in the
MHD instabilities: Ballooning, flute, interchange in- magnetic shear of an upward field-aligned current. Since
stabilities. Roux et al. [1991a, b] were the first re- magnetic shear can stabilize the KHI, one must be
searchers to show observations which indicated the bal- careful in attributing too much correspondence between
looning instability may be important for auroral the existence of auroral spirals and the vortex streets in
breakup. Ullaland et al. [1993] have attributed an auroral a KHI.
breakup related to a sudden commencement as being Further, Nishida et al. [1988] have shown that the
due to a ballooning instability. Controversy does exist as PSBL is not a distinct entity in which similar events
to whether this is a viable mechanism. Ohtani and could be found in both the near Earth and more distant
Tamao [1993] argue that previous evidence for balloon- tail. This view supports the mapping shown in Plate 6, in
ing has not taken into account plasma compressibility. which the disturbance involves a large region of the tail
Chan et al. [1994] on theoretical grounds showed that (the active plasma sheet) and not just the PSBL.
ballooning might occur when Pñ > 2Pll. Lui et al. [1992] Mapping considerations imply that the KHI may be
showed that observationally just prior to disruption the important in the dawn and dusk sectors and may play an
total particle pressure is either isotropic or Pñ > Pll' essential role in the understanding of fan arcs and those
Ullaland et al. [1993] and Daglis et al. [1991] have found polar arcs which are extensions of the auroral oval. This
the opposite to be the case, although one must be careful implies that while the KHI may play an interesting role
to evaluate when the observations took place relative to in the evolution of auroral patterns, it is unlikely to play
the exact time of onset.
an important role in understanding nightside auroral
Iranov and Pokhotelov [1987] showed that it was eas- activations.
ier to trigger the flute instability when there exist field- Surface waves in the outer plasma sheet. Surface
aligned currents which have associated with them azi- waves observed at high altitude at the boundary between
muthal pressure gradients in the equatorial plane. These the lobe and plasma sheet close to the time the plasma
gradients ease conditions on the radial pressure gradi- sheet recovers [Forbes et al., 1981] may be related to the

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 215
spirals and poleward undulations seen at this time (see
Figure 12a, bottom right, and Figure 12b, left).
4.4. Reconnection
4.3. Ionospheric Contributions
The most poleward arc system and reconnection.
Many events supposedly occurring in the aurora have
Daglis et al. [1990, 1991, 1994] have presented evi- been interpreted as the signature of reconnection (e.g.,
dence that during growth phase the ionosphere may play the poleward leap of the aurora [Hones et al., 1985] and
an active role in preconditioning the magnetosphere for the splitting of an arc as the bulge develops [Atkinson et
an onset event. Upwelling of O * from the ionosphere al., 1989; Atkinson, 1992]). The development of the
may result in cigarlike pitch angle distributions of ener- double oval independently from the substorm bulge supgetic
ions in the near-Earth region. The associated ports a developing neutral line tailward of 30 Re [Elanisotropic
pressure distribution could then lead to an phinstone et al., 1995c]. The associated spirals and unenhanced
cross-tail current. As Cladis and Francis [1992] dulations at this time are consistent with this view. In
and Daglis andAxford [1996] have suggested, this implies some cases these are known to occur just as the plasma
that the ionosphere can play an active role in magneto- sheet recovers, and we have seen that these can occur in
tail dynamics as well as provide a source for ions. Persson association with fast bulk flows in the PSBL. Associating
et al. [1994] found supporting evidence for this by ob- these events and the streamers with reconnection would
serving ion outflow close to the moment of onset on the imply that reconnection is not steady but occurs in
breakup arc itself.
bursts. We also know that this most poleward system
Conductivity may also play a role, as noted byAkasofu has, at least sometimes, a magnetospheric source tailand
Kan [1982]. Conductivity may influence where an ward of 30 Re. Therefore we also have a possible
onset occurs and could help explain the local time asymlocation
in the tail for this activity. The results of Slavin
metry which places onset in the premidnight sector and
et al. [1984, 1993] which show traveling compression
at the west edge of AAF events [Elphinstone et al.,
regions appearing in the deep tail 10-30 min after sub-
1995a]. Elphinstone et al. [1992] worked with the assump- storm onset are not inconsistent with the view that
tion that precipitation giving rise to ionospheric conducreconnection
and the resulting plasmoid are sometimes
tivity can be regulated by the partial filling of loss cones
initiated shortly after onset. Perhaps these tail events are
due to pitch angle scattering of particles near the peak in
more closely tied to the intensifications on the most
the cross-tail current. The Earth's internal magnetic field
poleward system of the double-oval distribution rather
has asymmetries which modify the precipitation maps.
than onset itself.
These patterns show peak precipitation at different local
times as a function of universal time. These results
Near-Earth neutral line model. Some versions of
this model tie auroral breakup closely to the last closed
agreed with actual observations of onset. This work
demonstrated that there may be local time asymmetries
field line reconnecting. These are clearly not correct on
the basis of observations of the onset taking place on the
to onset (and the substorm bulge) which depend on the
Earth's internal field.
main UV oval during a double-oval configuration. Some
versions of this model, however, have reconnection ini-
PiB pulsations are associated with onset, and their
cause may also play a role in conditioning the iono- tiated within the closed field line region [e.g., Baker and
sphere. In this context, Lysak et al. [1992] has investi- McPherron, 1990, and references therein]. On the basis
gated a resonant cavity between the ionosphere and a of auroral data during active intervals there must be a
point along the field lines where the Alfven speed peaks. time period of about 4-15 min after onset before the last
It is clear that the coupling between the ionosphere and closed field line reconnects. This progression of the
magnetosphere is important. What is not so clear is bulge from the main oval to the poleward portion of the
whether the ionosphere plays a direct role in causing double oval has been reported many times [Sergeev and
auroral breakup. Hesse and Bim [1991] argue that the Yahnin, 1979; Elphinstone et al., 1991a; Murphree and
ionosphere only changes timescales but does not alter Cogger, 1992; Cogger and Elphinstone, 1992; Elphinstone
the qualitative pattern which is driven by reconnection in and Heam, 1993; Gazey et al., 1995]. We also know that
the magnetotail. Raeder et al. [1996], on the other hand, this time progression appears to depend on the level of
show that ionospheric conductivity may be fundamental activity. That is, high activity implies a short time interval
to understanding tail dynamics.
for the reconnection to reach the last closed field line
Kan et al. [1988], Kan [1993], and Kan and Sun [1985] (5-15 min), whereas low activity implies a longer time
take the opposite view and propose a model in which the interval (15-30 min).
magnetosphere plays a relatively minor role. It is not It is clear, however, that the arrival of the bulge at this
likely that the magnetosphere remains a completely pas- boundary frequently has very specific and clear auroral
sive player in the process. Nevertheless, the success of signatures [Elphinstone and Heam, 1993; Cogger and
their simulations in reproducing the substorm bulge is Elphinstone, 1992; Elphinstone et al., 1995c]. It may be
quite impressive. It has demonstrated very well the need that this later auroral signature is the event associated
to take into account the dynamics of Alfven wave cou- with the last closed field line reconnecting. This is not
pling to the ionosphere.
inconsistent with high time resolution data which have

216 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
put the plasma sheet dropout 4.5 min after onset [Hones
et al., 1986].
Baker et al. [1993] propose that these types of auroral
12
observations can be explained by a coupling between the
reconnection regions and the near-Earth region. We
have seen that there can be some coupling from the
outer tail to the inner for some events, but it is clear even
for those events that the nearer-Earth onset (AAF type)
is attributable to a separate instability.
1
3
Birn and Hesse [1994] showed that an increase of
about a factor of 3 for B z from midnight to the flanks
may suppress the resistive tearing instability. This may
an important result when mapping is considered. From
mapping we know that strong Bz along the flanks of the
magnetotail is probably associated with closed field lines
extending to high latitude at dawn and/or dusk in the
18
-:?. ':øøo 06
ionosphere. This implies the existence of polar arcs at a
time when resistive tearing is suppressed. The disappearance
of polar arcs prior to onset (which is sometimes
5
O0
6
observed) may be evidence that B is decreasing along
the flanks of the tail (see mapping section above). If this
were the case, then this would support the view that
when B is decreased, one might expect the magneto-
Figure 22. A schematic view of the active auroral modules
and some possible associated mechanisms. The auroral modules
shown are as follows: 1, dayside spirals; 2, detached arc; 3,
pulsating aurora; 4, fan arcs and polar arcs; 5, equatorward
sphere to be in a better state to initiate the resistive undulations; 6, AAFs; 7, double oval spirals; 8, streamer; and
tearing instability.
9, polar arc/stationary surge. The corresponding possible associated
mechanisms are as follows: 1, upward field-aligned current
instability (dayside boundary layer); 2, Pc 5 resonance
5. DISCUSSION AND CONCLUSIONS
(cavity mode or ring current instability) pitch angle diffusion?;
3, VLF wave instability?; 4, broadband compressional waves
and resonances?, KHI?; 5, shear flow Alfven ballooning?; 6,
An overview of active auroral modules has been preballooning
with azimuthal pressure gradients or anisotropic
sented to aid magnetospheric substorm researchers in
pressure; 7, upward field-aligned current instability (fast flow
identifying events which may be applicable to their par- in PSBL)?; 8, bursty bulk flows in CPS?; and 9, disturbance in
ticular data sets. Figure 22 is a schematic representation flanks of tail?
of a few of the auroral signatures associated with various
modules. These have been superimposed on a single
diagram for brevity. Tables 1 and 2 summarize many of ical studies have the drawback that processes tend to be
the principle points of this paper by listing the modular put to the side in the effort to accurately describe the
elements. The tables are a first attempt at associating system. In general, however, researchers have some coneach
module with a process, its signature in the aurora, cept of how a particular auroral form relates to a magits
source in the magnetosphere, and its coupling with netospheric process and how this process has links back
other modules. The tentative associations made here into the solar wind and to preconditioning by other
presumably will be tested and explored in the future. In means. These ideas provide a starting point from which
this section we expand our view to try and understand theories can be developed and substantiated.
the underlying processes involved in these modules and This section is based on work described in the modhow
various cycles of activity arise.
ules and mapping sections above, and so the reader
should refer there for references and details. The fol-
5.1. Processes and Cycles Within the Modular View lowing represents a synthesis of the observations and
Many of the figures in this paper (Figures la, lb, 2, 9, modeling efforts described above and illustrates the use-
12a, 12b, 14, and 22) illustrate a problem which is faced fulness of the modular concept in "building" one's own
by most researchers involved in phenomenological stud- substorm. The examples presented below are specific
ies of two-dimensional time-dependent "qualitative" instances of a much broader set of possibilities. As such,
data. The data are generally too complex and varied to they should not be considered as "canonical" descripbe
presented by themselves, and so schematics are de- tions of auroral activity cycles but rather as examples of
vised to give the "essence" of the data from the re- the possible interactions allowed (and known to exist
searcher's view. This can be very useful, but in the case phenomenologically) within this complex system.
of the aurora, even relatively simple events become The left schematic of Figure 23a illustrates the prodifficult
to characterize in a way that illustrates the cess/module correspondence for a general case. A modimportant
parts of the scheme. Further, phenomenolog- ule can be thought of as a magnetospheric process which

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 21 7
TABLE 1. Primary Modules
Auroral Possible
Module Module Mechanism
1 Intensification
la Main UV oval onset
lb Eastward propagation of
arc brightening
lc Poleward oval onset
ld Intensification near the
foot of polar arc
2 Bulge
3 Dayside patterns
3a Fan arcs
3b Day-night split
3c 1400 MLT spirals
3d Impulsive low-latitude
form
3e FTE-like
4 Position and width changes to
oval
near-Earth
instability
reconnection
reconnection/fast
flow in PSBL
convective changes
tailward propagating
rarefaction wave
and/or reconnec-
tion, SCW
near-Earth
Magnetospheric
Source Types of Coupling
in dusk sector and at
intermediate distance
PSBL or separatrix
intersection of cross-tail
current and tail LLBL
all of midnight plasma
sheet and/or
topological
reconfiguration of
magnetosphere
5 Double oval tail boundary regions PSBL (midnight) and
become active
usually when
plasma sheet
has recovered
LLBL (dawn/dusk)
6 Changes to polar arcs
6a Dawn-dusk-expanded ovals KHI on flanks of tail tail flanks with large B z
6b Polar arcs out of bulge either same as (a) or
effect in tail lobe
tail lobe
if no coupling exists to other
modules, then these are pseu-
do breakups
from inner to outer regions of
magnetosphere
KHI or broadband morning sector part of directly driven signatures of
compressional dayside magnetosphere solar wind changes; can also
waves coupling and along the flank of be coupled to changes into
Alfven waves
tail
duced from other modules
decoupling of dayside and nightside affecting these regions
dayside/nightside
generators
boundary layers
upward field-aligned
current generator
dayside boundary layer
KHI with Alfven near subsolar dayside
transit time effects boundary layer
reconnection or KHI dayside magnetopause
temporal changes to dependent on locations dependent on how the
magnetospheric
magnetospheric currents are
topology; strongly
altered; this can be desolar
wind
termined by other modules
controlled
and also determine whether
other modules become active
can affect most other modular
development
can affect convection and cause
polar arc intensification
(type ld); IMF Bz related
See section 5 for more information. PSBL, plasma sheet boundary layer; LLBL, low-latitude boundary layer; SCW, substorm current wedge;
KHI, Kelvin-Helmholtz instability; FTE, flux transfer event.
is operating owing to some variation in the solar wind. and the traversing arc may be the alternative ionospheric
Whether or not these variations result in the module signature.
becoming active will be related to preconditioning of the Proceeding in this spirit, cycles of auroral behavior
magnetosphere-ionosphere system and interactions with can be understood and described within the modular
other processes or modules. Magnetosphere-ionosphere flamework. For this, a symbolic representation of the
coupling determines how this process appears in the modularization will be adopted. Throughout Figures
ionosphere: as aurora, field-aligned currents, convec- 23a-23c three symbols are used to describe different
tion, etc. Thus, while we started this modular concept levels of modular elements. Circles represent primary
from a phenomenological view, one could just as easily modules (PM), and squares represent secondary moddevelop
new modules based on a cycle one expects in the ules (SM). Within these are numbers from Tables 1 and
magnetosphere and make a prediction of the result in 2 stating which module is involved. Hexagons will repthe
ionosphere. The prediction of the poleward leap of resent some composite form (CM) of modular elements
the aurora from the near-Earth neutral line model is an (i.e., a macro or cycle or phase) which occurs frequently
example of just such a reasoning. While this prediction enough to warrant a definition. Bidirectional arrows
did not turn out to be entirely true, the splitting of arcs indicate some type of feedback, switching between

218 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
TABLE 2. Secondary Modules
Auroral Possible Magnetospheric
Module Module Mechanism Source Types of Coupling
1 Traversing arc reconnection of last
closed field line and
also M-I coupling
2 Detached arc pitch angle diffusion and
possibly cavity mode
resonance
3 Splitting arc reconnection or M-I
coupling effect
4 Spirals Field-aligned current
instability
5 Curls Electric field instability
6 Streamers Fast flows in central
plasma sheet
near separatrix
dusk ring current
region
neutral line M-I
coupling
nearly any location
associated with
auroral arcs
nearly any location
associated with
auroral arcs
central plasma sheet
7 Pulsating aurora VLF activity and pitch near-Earth
angle diffusion
8 Omega bands
9 AAFs
10 Undulations
flute, ballooning energetic near-Earth morning
drifting electrons sector
Ballooning with azimuthal near-Earth
pressure gradients or (usually morning)
pressure anisotropies
shear flow Alfven near-Earth
ballooning
can initiate streamers
couples to dayside spirals
can initiate streamers and type l d
onsets; these couple to form bulge
when this occurs at the west tip of
the bulge it is the WTS; it is primary
signature of type lc onsets
not known
couples to 1, 3, 4, and 7 secondary
modules; outer to inner tail
coupling
couples to 6, 8, and 9 secondary
modules; relation to onset not yet
known
possible coupling to AAF main UV
onsets and to module 7, above
is the dominant form of onset type
l a; may couple with modules 4, 7,
and 8, above
not known
11 Stationary surge partially related to spirals; most of dusk sector couples near-Earth and deep-tail
may be related to tail regions
12
waveguide resonances
Latitudinally propagating Alfven waves coupling to central plasma sheet couples inner and outer regions of
arcs propagating source tail
See section 5 for more information. M-I, magnetosphere-ionosphere; WTS, westward traveling surge; AAF, azimuthally spaced auroral form.
states, or the states existing at the same time. Unidirectional
arrows give a sense of temporal development.
Under this representation, classical expansion and
recovery phases can be described by the two macros
given in the bottom right schematic of Figure 23a, and
the entire traditional auroral substorm cycle can be
drawn simply, as in the top right schematic of Figure 23a.
In this view, we obtain the phenomenological picture of
the substorm. This begins with the change in polar arcs
(primary module 6, or PM-6) followed by an onset
(PM-la), then a bulge (PM-2), omega bands (SM-8),
and pulsating aurora (SM-7). The recovery phase is
characterized primarily by a continuation of omega
bands and pulsating aurora and a gradual "recovery" of
the position and width of the oval to a quiet time
configuration (PM-4). Equatorward drifting arcs are also
seen during this time (SM-12).
Referring to Tables 1 and 2, processes can be identified
with each of the above phenomenological descriptions,
and from these a simplified process-driven expla-
nation of the substorm is evident. As one thinks about
this, it becomes clear that other cycles of activity are
quite likely. For example, a very developed growth phase
(denoted by CM-G and shown in the top left of Figure
23b) could logically occur as follows (note that individual
portions of this scheme do not have to necessarily occur):
IMF Bz changes from positive to negative and
loads the magnetotail with magnetic flux via reconnection
on the dayside magnetopause (PM-4 and PM-3e).
Initially, as a result of this, the dayside and nightside
generators of field-aligned current may become decoupled
(PM-3b). The dayside reconnection in turn enhances
the cross-tail current and reconfigures the magnetosphere
so that Bz is no longer large along the flanks
of the magnetotail (PM-6). Solar wind pressure changes
could also occur driving KHI in the boundary layers
(PM-6; PM-3a, c, d), and delivering compressional wave
energy into the magnetosphere. Depending on ionospheric
conditions, this might couple to Alfven waves
(PM-3a) or even to a cavity mode resonance (SM-2).
These changes (in a variety of ways) could initiate a
tearing instability and reconnection in the midtail and
deep-tail regions (PM-lb, c) and/or trigger a near-Earth
instability (PM-la) such as ballooning (SM-9). The
deeper-tail events might then couple to near-Earth activity
possibly resulting in coupling signatures in the
ionosphere (SM-12).
These types of events would be pseudobreakups and

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.' MODULAR AURORAL ACTIVITY ß 219
INTERACTIONS
AND
PRECONDITIONING
VIA OTHER
MODULF.
$OKS, R
VARIATIONS
SHOCK Bow
MAGNETOSPHERIC
PROCESS
M-I COUPLING MODULE
1ONOSPHERIC
SIGNATURF, S
MODULE #
O -Primary
-Secondary
-Macro
TRADITIONAL
AURORAL SUBSTORM
E= EXPANSION PHASE R=RECOVERY PHASE
Figure 23a. Schematic representations of a few magnetospheric activity cycles based on the modular view of
auroral activity. Bidirectional lines represent events which contain feedback, or the time sequence can vary
(the system can flip between the two states). Singly directed lines imply a temporal sequence. This representation
allows a complex set of events to be described in relatively little space. The left chart illustrates how a
module is described equivalently by either its phenomenological aspect or by its underlying process. Which
approach is used depends on the application and the level of understanding of the underlying physics. The
bottom right schematics (enclosed by hexagons denoting cycles or partial cycles of activity) show how the
traditional expansion and recovery phases are represented using the modular view. In these schematics, circles
represent the primary modules, and squares the secondary ones (which are indexed according to the numbers
given in Tables 1 and 2). The top right schematic illustrates how the traditional auroral substorm can be
represented using this approach.
could recur unless the instabilities were strong enough to (which are perhaps related to reconnection processes at
affect other regions of the magnetosphere. In such a case the dayside magnetopause) would appear in the dayside
a revised expansion phase might result (CM-E1), such as (PM-3e), and a split between the dayside and nightside
that shown in the top right schematic of Figure 23b. Such sectors would occur (PM-3b). Fan arcs (PM-3a) would
a case involves the instability reaching some critical appear first on the dayside and propagate to the nightlevel,
which then results in an expansion phase bulge side accompanied by Pc 5 magnetic pulsations and west-
(CM-E) related to the propagation of a rarefaction wave ward electrojet enhancements. Enhanced upward fielddowntail
(PM-2). While a sufficient level of growth ac- aligned currents in the 1400 MLT sector would cause
tivity will presumably trigger a substorm, lower levels will spirals to appear (PM-3c), and if Alfven transit times
result in the growth cycle to appear isolated from an were appropriate, a low-latitude impulsive feature might
expansion phase.
appear (PM-3d). As the auroral disturbances reach the
This sequence of events (CM-G and CM-E1 in the nightside ionosphere, several things are possible. Intentop
of Figure 23b) under appropriate conditions for the sifications could form near the open-closed field line
M-I coupling to occur would have its auroral counter- boundary (PM-lc), an eastward propagating activation
parts as follows (assuming, of course, that the mecha- might appear (PM-lb), or an onset from the main UV
nisms chosen in Tables 1 and 2 have the corresponding oval could occur (PM-la). If the coupling to the ionosignatures):
Polar arcs would fade (PM-6), convection sphere was appropriate in the latter case, AAFs could
would be enhanced, and the oval would move equator- occur (SM-9), and equatorward propagating arcs might
ward (PM-4). Flux transfer event (FTE) candidates move toward this inner region (SM-12).

220 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
G=Full Growth
DO= Double Oval
PSBL= Poleward oval activity
\/
Fq
El= Revised full expansion phase
CO= Coupled ovals
Figure 23b. Schematic representations of a few magnetospheric activity cycles based on the modular view
of auroral activity. Shown are five "macro" elements (abbreviated G, DO, PSBL, El, and CO) which are
observed to occur in the auroral distribution. See Figure 23a caption for additional information.
The revised expansion phase (CM-E1) shown in the trons into the ring current region, as well as enhanced
top right of Figure 23b involves several additional second- convection in the inner magnetosphere, could initiate
ary modules but also includes the possibility of an impor- the shear flow ballooning instability and wavelike ripples
tant temporal progression beginning with an AAF activa- near the equatorward edge of the diffuse aurora (SMtion
(SM-9: implying a wave instability existing just prior to 10). These particles could also precipitate in the afteronset)
and a traversing arc (SM-i: implying reconnection noon sector to form diffuse aurora far from the regular
or a disturbance reaching the open-closed field line bound- auroral oval (SM-2). The mechanism for the formation
ary) as the bulge reaches its most poleward extent. It also of a stationary surge (SM-11) is so ill defined, it is
includes the possibility of polar arcs intensifying directly perhaps best to just mention that the process involves
out of the substorm bulge (PM-6) which indicates an in- most of the duskside plasma sheet and must couple in
teresting coupling between substorm and polar arc phe- some manner the near-Earth region to the deeper tail.
nomena which has previously not been recognized. When the tail boundary regions are active (PM-5),
Immediately below the diagram for full growth phase one might expect reconnection to play a role in the
in Figure 23a is one illustrating a common pattern which substorm dynamics (SM-3). This reconnection process in
frequently forms late in a substorm cycle (CM-DO). At turn may drive bursty bulk flows in the central plasma
the time when the oval is in a state of recovery (CM-R, sheet (SM-6) and field-aligned currents (SM-4). This
from Figure 23a), a double oval tends to develop (PM- combination of modules is diagramed in the bottom left
5). Along with this feature, undulations (SM-10), curls schematic of Figure 23b (CM-PSBL). The auroral equiv-
(SM-5), and detached arcs can appear (SM-2). Station- alent (seen schematically in the bottom right of Figure
ary surges can form out of the most poleward system 12b) would be the periodic auroral arc splitting (SM-3)
(SM-11).
on the most poleward arc of a double oval (PM-5), which
In a process-based description, the tail boundary lay- is frequently followed by an auroral streamer (SM-6),
ers (both the LLBL and the PSBL) become active in and auroral spirals (SM-4).
response to the magnetotail reconfiguring itself as a While the above cycle (CM-PSBL) is very common, it
disturbance leaves the tail (PM-5). The strong electric only occasionally results in a substorm onset (PM-la) if
fields give rise to instabilities from which the auroral the bursty bulk flows trigger a near-Earth instability such
curls arise (SM-5). Injection of energetic ions and elec- as ballooning (SM-9). This may be linked to the devel-

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 221
also help researchers understand the so-called steady
magnetospheric convection events as being controlled by
smaller cycles such as CM-PSBL events. This modular
view also helps avoid semantic difficulties about what is
an onset event and what is not and makes it clear that
auroral activity cycles are history dependent. This has
previously not been adequately taken into account and
has led to mistaken interpretations of the timing of some
magnetospheric events (such as plasma sheet recovery)
relative to some auroral activations.
5.2. Questions for the Future
POSSIBLE GLOBAL AURORAL SUBSTORM
A stereotyped auroral substorm isolated and existing
by itself is a rare and unlikely event. Cases which enter
Figure 23c. Schematic representations of a few magnetospheric
activity cycles based on the modular view of auroral
activity. Shown is an idealized "substorm event" which illustrates
the types of activations which occur and some of the
more obvious cycles which have been observed, This is meant
as an illustration of the modular representation rather than as
a canonical "auroral substorm." See Figure 23a caption for
the grey area of events which do not conform with the
global development expected from the classical auroral
substorm are more likely. Our understanding both of the
patterns and the processes involved in the aurora have
evolved such that new descriptions of active auroral
phenomena have become necessary. Auroral activity inadditional
information.
volves cycles of development which deserve to be analyzed
separately in order to better understand the underlying
magnetospheric processes.
opment of pulsating aurora from streamers (SM-7). This Researchers can now ask questions different than
tail-to-near-Earth onset coupling is diagramed in the "Did we observe a global auroral substorm onset?" and
bottom of Figure 23b (CM-CO), and a prime observa- enquire into what type of onset module was seen and
tional example of this is given in Plate 4.
whether this module triggered larger, more global cycles
With the above introduction, one can construct pos- of activity: If so, why? If not, why? This new view could
sible generic auroral substorm scenarios which could not be reflected in future studies concerning magnetospheric
easily be sketched or described in detail by other means. phenomena. Future simulations should be able to in-
Figure 23c is a view of just one possible complex cycle clude the ionosphere as a key element in their developbuilt
up from modules coupling together in prescribed ment so that the links between the auroral and magneways.
Here we see elements of the traditional substorm tospheric substorm can be further understood. We shall
scheme but now fundamentally new activations can oc- end the paper with questions which hopefully can be
cur at various points in the cycle. In the old flamework, answered more fully after the acquisition and analysis of
one might call these activations or intensifications, but data collected from the future missions of the Internathey
held little or no special significance. However, some tional Solar Terrestrial Program.
of these can drastically affect the subsequent evolution What differences exist between substorms where only
of the general cycle and determine a proper understand- a few modules occur and ones where many modules are
ing of what drives various activities. For instance, an manifested?
isolated cycle of CM-G --> CM-E1 --> CM-DO during What specific magnetospheric mechanism can be
quiet times does not give the same information as a identified with each of the auroral patterns?
similar event during which the double oval is previously When one cycle of activity has occurred, what is the
present. In the latter case, we have the potential oppor- likelihood it will be followed by some other cycle? For
tunity to observe in the ionosphere very clear inner-to- example, if a double oval forms, how often is it followed
outer magnetospheric coupling followed by a progres- by PSBL-type cycles or coupled oval cycles? If a particsion
from the outer regions of the magnetosphere to the ular type of onset occurs, how often does it result in
near-Earth region (or vice versa).
bulge formation and why? How often do these in turn
In such a case, growth phase activity may exist simul- form double oval configurations?
taneously with other activity making the "phase" of the Are PSBL cycles the driving process for steady consubstorm
an ambiguous term. Two onsets may exist in vection events?
different local time sectors and may have quite different There are many processes/events which do not follow
character. It was these cases of onset before previous field lines and so do not have a simple mapping from the
activity had died away which led researchers to under- ionosphere to the magnetosphere. How important are
stand that substorm onset had a source location in the these, and when do they affect the interpretation of
near-Earth region of the magnetosphere (by knowing ionospheric data?
onset began from the peak in the main UV oval). How often do certain modules occur only in preferred
This more complex view of auroral activity cycles may cycles? While some do tend to occur at a particular point

222 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
in a larger cycle (e.g., the double-oval module), on Auroral oval: A term used to refer to the shape of
occasion they appear without this temporal connection the dominant portion of the auroral distribution. The
(the double oval sometimes appears without a preexist- term originates from statistical studies in the early 1960s
ing substorm bulge). What clues does this give us as to and hence refers primarily to the distribution of discrete
the auroral module's generation mechanism?
auroras.
What criteria determine the direction of the coupling Auroral substorm: In its original form [Akasofu,
between the inner and outer magnetosphere? 1964] it designated the auroral activity composed of two
Can substorm onset sometimes be predicted by day- phases called expansion and recovery. Expansion inside
or nightside magnetospheric-ionospheric activity volved poleward motion, and recovery was a return
occurring in the 10-15 min prior to onset (as suggested equatorward to the previous state.
by Figure 5b) ?
Auroral zone: Prior to Russian work, primarily by
We know from auroral observations (section 4.4) the Feldstein [1960, 1963], the aurora was believed to exist
temporal progression of the substorm bulge poleward is primarily in a zone of constant magnetic latitude. Daydependent
on the magnetic activity level. Can reconnec- side aurora was believed not to occur frequently as a
tion simulations reproduce this result? What causes this result of this misconception.
difference?
Azimuthally spaced auroral form (AAF): Periodi-
Do microinstabilities couple to larger MHD-type in- cally spaced (in longitude) auroral forms frequently seen
stabilities to result in substorm onset?
just prior to auroral breakup. They should, in general, be
What other cyclical auroral behavior can be found distinguished from spirals, curls, and omega bands, alwhich
will yield clues to underlying processes? though they have been known to change into spirals and
Why do substorms not release their energy all at once can exist at the same time as omega bands [Elphinstone
but rather through multiple events?
et al., 1995a]. Several of their characteristics support that
How do variations in the AL/A U index get translated a ballooning instability in the magnetosphere may be
into the longer-scale variations of the Dst (i.e., which associated with their generation.
particular modules must be active in association with the Breakup: A dynamic change to the aurora which
storm process)?
was known to occur at least as far back as [Stagg, 1937;
What is the cause of substorm polar arcs (i.e., the Elvey, 1957]. It describes the disorder which appears
ones that intensify directly out of the substorm bulge)? when previously quiet arcs get disrupted by an onset of
Are these arcs on open field lines as opposed to the the most equatorward arc system.
more typical expanded oval arc systems?
Bulge: The shape of the poleward moving aurora in
Can auroral indices similar to those described in the the midnight sector during the substorm expansion
introduction be calculated from global auroral data such phase. When the bulge gets larger (usually by its polethat
they can be implemented to run continuously? To ward edge moving poleward), it is said to be expanding.
do this would require a continuous global auroral mon- This has led to the term expansion phase of the auroral
itor.
GLOSSARY
substorm.
Convection in magnetohydrodynamic (MHD) fluids:
The bulk motion of a plasma in MHD. Convection in
ordinary fluids arises from temperature gradients,
whereas in MHD fluids it arises from the electric field
Arcs: Definitions for this term vary considerably, being transformed away in the plasma frame of referdepending
on the instrumentation used [see, e.g., Shep- ence. In ideal MHD the magnetic fields are "frozen" to
herd et al., 1988]. We shall use an optical definition such the plasma motion, leading to the idea that magnetic
that an arc is the narrowest longitudinally extended field lines can also convect. Field-aligned currents flowauroral
feature that shows spatial or temporal structure ing from the magnetosphere provide a coupling to the
in a given region. A set of arcs occurring together will be ionospheric electric fields and convection. The energy
called an arc system. The typical measured thickness of source for this convection is the solar Wind as it flows
an auroral arc is about 100 rn [Borovsky et al., 1991]. around the magnetosphere.
Auroral indices' These are generally measures of Corrected geomagnetic coordinates (CGM): An
magnetic activity. They include indices associated with ionospheric coordinate system used to order auroral
auroral electrojets (AL for westward, ,4 U for eastward, observations and eliminate the effects of deviations in
and AE for both), the global system (Kp), and ring the Earth's internal magnetic field from a simple dipole
current/magnetospheric storm-related phenomena field. See work by Heam et al. [1993] for a description of
(Ost).
ionospheric coordinate systems used to order geophysi-
Auroral modules: A set of auroral patterns ob- cal quantities.
served during active times which can be combined in Cross-tail current: A nightside current near the
series or parallel to construct one instance of activity. Earth's equatorial plane which is directed from the dawn
The method in which they get assembled describes how to the dusk sector of the magnetosphere. This current is
the global magnetospheric system couples together. associated with the existence of the magnetotail.

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 223
Curl: A small-scale (2-10 km) auroral feature Ionosphere: A portion of the atmosphere which is
which is wound up clockwise when viewed in the direc- permanently ionized, primarily owing to photoionization
tion of the magnetic field.
from solar ultraviolet rays and particle precipitation.
Dayside auroral breakup: A term used to describe The low-altitude boundary of this region is near 60 km,
intensifications and poleward moving auroral forms on and its high-altitude boundary is near 3-4 R e near the
the dayside auroral distribution [e.g., Sandholt et al., 1990]. equator. Three subregions have been named the D layer
Day-night split: A separation (usually in latitude) (below 90 km), the E layer (from 90 to 150 km) and the
which can occur during active times between the after- F layer (above 150 km).
noon and evening auroras.
Detached arc system: A broad arc system which
lies equatorward of the main auroral distribution in the
afternoon and evening sectors.
Diffuse aurora: Generally, a nonstructured auroral
form. A strict definition would be precipitation which
does not have field-aligned acceleration associated with
tow-altitude boundary plasma sheet (I. ABPS): A
low-altitude (less than 2 Re) particle precipitation exhibiting
changing flux and energy characteristics and
usually identified with auroral arcs.
tow-altitude central plasma sheet (I. ACPS): A lowaltitude
particle precipitation associated with diffuse aurora.
it (this, however, cannot, in general, be determined). Low-energy ion dispersion signature (tEIDS): An
Dipolarization: A high-altitude signature usually ion signature found at low altitudes, similar to a VDIS
associated with substorm onset. It signifies a change in
but usually found more equatorward and of a lower energy.
the nightside magnetospheric magnetic field toward a
It is frequently found equatorward of inverted V events.
more dipolar configuration.
tow-latitude impulsive dayside form (LLIDF): An
Discrete aurora: Usually associated with auroral
auroral feature which appears first equatorward of the
arcs but strictly defined is the aurora associated with
1200-1500 MLT auroral oval and then progresses up to
field-aligned acceleration. The precipitation usually has
the oval [Elphinstone et al., 1993a].
a prominent peak in electron flux at a particular energy.
tow-latitude boundary layer (LLBL): A magneto-
Double oval: Two large-scale intensity peaks spanspheric
region near the Earth's equatorial plane where
ning several hours of local time, parallel to each other,
the plasmas from inside and outside mix.
but separated by about 5ø-10 ø of latitude. The equator-
Magnetic local time (M[I): A coordinate in which
ward peak is the main UV oval. The region between
them is filled with smaller-scale arcs and diffuse aurora longitudes defined from a magnetic pole location are
used to define a 24-hour clock: 0000 MLT is in the
[Elphinstone and Heam, 1992; Elphinstone et al., 1995b,
antisunward direction, 0600 MLT is a dawn meridian,
c].
1200 MLT is near noon for an observer, and 1800 MLT
Electrojets' Horizontal currents flowing at ionospheric
heights in the region of the aurora, primarily in
is in the vicinity of dusk. See work by Heam et al. [1993]
the east/west direction.
for a more detailed explanation.
Fan arcs: Multiple arcs spaced in latitude which Magnetic pulsation: Periodic or quasi-periodic
converge near 1200 MLT. They are usually seen in the variations in the Earth's magnetic field. Frequencies
0600-1200 MLT sector.
range from 5000 mHz (0.2-s period and called a Pc 1) to
Field-aligned currents: Currents which flow along beyond 1.67 mHz (600-s period and called a Pc 5) (see
magnetic field lines and provide a means of coupling the work by Jacobs et al. [1964] for a description of the
magnetosphere to the ionosphere. These currents are various types of pulsations.) Pi 2 values are irregular
also called Birkeland currents, after the person who first pulsations used to define onset times and have frequensuggested
their existence.
Geosynchronous orbit: A satellite orbit of 6.6 Re
in the equatorial plane which allows the satellite to
cies from about 6.7-25 mHz.
Magnetohydrodynamics (MHD): The treatment of
a collection of charged particles as a fluid (a magnetoremain
at the same geographical point above the Earth. hydrodynamic fluid). MHD fluids involve the collective
Hall currents: Ionospheric currents which flow per- motions and interactions of the particles, the electrical
pendicular to the ionospheric electric field.
currents, and the magnetic fields. Restrictions exist as to
Injection event: A rapid enhancement of energetic its applicability to space plasmas; these restrictions are
particles in the vicinity of geosynchronous orbit. This is related to the spatial and temporal scales of the individfrequently
used as a definition of substorm onset. ual particle motions as compared to the macroscopic
Interplanetary magnetic field (IMF): The magnetic parameters. Ideal MHD (which is sometimes useful)
field generated at the Sun and associated with interplan- occurs when the conducting fluid has infinite conductivetary
space. Southward fields (IMF Bz) in the IMF have ity.
been found to be associated with auroral activity. Magnetopause: The surface which bounds the mag-
Inverted V: An electron precipitation pattern which netosphere. A useful operational definition (used in the
shows an increase, then decrease, in characteristic en- mapping section in this paper) is that it is the outer
ergy as a function of latitude. Usually associated with boundary of the locus of field lines which connect to the
discrete arcs.
Earth.

224 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
Magnetosphere' The cavity formed from the inter- position of the IMF with the Earth's magnetic field will
action of the solar wind with a planetary magnetic field. create points with zero magnetic field called neutral
Magnetotail' The region of the magnetosphere points (or lines if they have some extent in space). In
which stretches out in the form of a "tail" away from the order to have "open" field lines, some form of merging
Sun.
between the two fields must occur. The magnetic field
Magnetotail lobes' Two low plasma density open energy associated with two field lines reconnecting is
field line regions of the magnetotail (one in the north used in accelerating particles.
and the other in the south). The energy density in these Ring current: The azimuthally flowing current asregions
is given predominantly by the magnetic field. sociated with the relative drift motions of electrons and
Main (UV) oval' The feature of the UV and optical ions in the near-Earth region of the magnetosphere. The
aurora which can almost always be observed. This results majority of this current flows westward, while a small
from a combination of both discrete and diffuse auroral
fraction in its inner portion flows eastward.
processes. This is the equatorward portion of the double Spiral: An auroral form (20-1500 km) which wraps
oval distribution.
up counterclockwise when viewed in the direction of the
Modular element' Amagnetosphericprocesswhich magnetic field.
is operating due to variations in the solar wind and Stationary surge form' An auroral form which recoupling
to other processes (see Figures 23a-23c). Au- sembles the WTS but remains stationary. Some types of
roral modules may appear in the ionosphere if the magthis
pattern may be quite different from the classical
netosphere-ionosphere coupling is appropriate.
WTS (see text).
Near-Earth' In this paper "near-Earth" refers to
Sun-aligned arcs: High-latitude polar arcs which
radial distances less than 10 R E.
are aligned in a noon-midnight direction.
Omega bands: "Luminous tongues of emission ex-
Storm (magnetic)' A time period lasting days durtending
poleward from a band of diffuse aurora in the
ing which the magnetosphere is first compressed and the
morning sector of the auroral oval with scale sizes of a
ring current is enhanced. The terms "storm" and "subfew
hundred kilometers." [Buchert et al., 1990, p. 3733].
storm" are historical ones, and they should not imply
Onset: In this paper, onset refers to the time when
that the storm is some direct manifestation of a series of
the aurora begins an explosive poleward motion in the
substorms. The term dates back prior to work by Adams
nightside sector.
[1892].
Open field line' A magnetic field line which has
Streamer: A particular north-south auroral strucone
end intersecting the Earth and one end connected to
the IMF.
ture which forms at the poleward edge of the double oval
Pealerson currents' Ionospheric currents which
configuration and drifts to the south-west (in the northflow
parallel to the ionospheric electric field.
ern hemisphere).
Plasma sheet: The central region of the magneto- Substorm current wedge (SCW): An upward fieldtail
which is bounded by the magnetotail lobes. The aligned current in the western portion of the bulge is
boundary layer closest to the magnetotail lobes is known linked to a downward field-aligned current in the eastern
as the plasma sheet boundary layer (PSBL).
portion via a westward electrojet in the auroral bulge
Polar arc systems (high latitude)' A generic term itself. The closure current (eastward) in the magnetofor
all arc systems which occur poleward of the "auroral sphere's equatorial plane effectively diverts the westward
cross-tail current.
oval." This encompasses other, more specific terms such
as teardrop (or horsecollar), theta, and Sun-aligned arcs. leardrop-shaped aurora (also called horsecollar)'
Polar cap: In this paper this refers exclusively to the A system of high-latitude polar arcs which exist in the
region of open magnetic field lines in the vicinity of the
dawn and dusk sectors in such a manner as to form a
magnetic pole.
teardrop-shaped "polar cap," with the narrowest part of
Pseudobreakup: Elvey [1957, p. 69] defined this the teardrop pointed toward magnetic noon. When the
term to mean "a disruption of one of the arcs near to the auroral oval is included, the pattern also resembles a
center of the auroral zone." Akasofu [1964] made an horse collar.
association of these events with a surge which propa- lheta aurora: A polar arc system for which there is
gates along a brightened arc without affecting others. evidence that open field lines lie on either side of it and
Ohtani et al. [1993] distinguish these from onsets accord- that it occurs on closed field lines. It therefore bifurcates
111[ tU WlIK;,tlIK,I d [IUUI:11 ,¾1It lCltCl U,CUI3. IOOdy tlIC
definition seems to be based on whether or not what
follows is termed a substorm.
occurs. Theta aurora should therefore be a very specific
form of the more generic term "high-latitude polar arc."
Pulsating aurora: Aurora which switches on and off Traversing arc system' An arc system which brightrapidly
with frequencies of 0.05-2 Hz (0.5- to 20-s peri- ens at or near the end of the bulge's poleward motion. It
ods) [Yamamoto and Oguti, 1982].
usually "traverses" across the midnight meridian and
Reconnection, or merging' A theory which de- spans several hours in local time [Murphree and Elphinscribes
the intersection of magnetic field lines. A super- stone, 1988].

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 225
Velocity-dispersed ion signature (VDIS): A ion Anger, C. D., and J. S. Murphree, ISIS-2 satellite imagery and
signature at low-altitude currently identified with the auroral morphology, in Magnetospheric Particles and Fields,
edited by B. M. McCormac, pp. 223-234, D. Reidel, Norionospheric
projection of the plasma sheet boundary well, Mass., 1976.
layer. Its signature is higher energy (about 10-40 keV) Anger, C. D., T. Fancott, J. McNally, and H. S. Kerr, ISIS-II
at higher latitudes which decreases with decreasing lat- scanning auroral photometer, Appl. Opt., 12, 1753-1766,
itude.
1973.
Westward traveling surge (WtS): Originally, this
referred to the single westward traveling spiral form at
the west end of the auroral bulge. It is now known,
however, that it is frequently not a single form and that
it moves westward episodically.
Anger, C. D., M. C. Moshupi, D. D. Wallis, J. S. Murphree, L.
H. Brace, and G. G. Shepherd, Detached auroral arcs in the
trough region, J. Geophys. Res., 83, 2683-2689, 1978.
Anger, C. D., J. S. Murphree, and L. L. Cogger, Satellite
observations of spatial and temporal fluctuations in postmidnight
auroras, Can. J. Phys., 59, 1137-1142, 1981.
Anger, C. D., et al., An ultraviolet auroral imager for the
Viking spacecraft, Geophys. Res. Lett., 14, 387-390, 1987.
ACKNOWLEDGMENTS. The authors would like to ex- Atkinson, G., An approximate flow equation for geomagnetic
press their gratitude to Pao Qi for his work in providing the flux tubes and its application to polar substorms, J. Geophys.
Canopus data. The Canopus project is supported by the Ca- Res., 72, 5373-5382, 1967.
nadian Space Agency. The first author would also like to thank
Atkinson, G., Mechanism by which merging at X lines causes
D. Hearn, C. Kennel, and V. Sergeev for discussions which
discrete auroral arcs, J. Geophys. Res., 97, 1337-1344, 1992.
Atkinson, G., How does magnetospheric convection relate to
were useful in formulating this paper and E. Donovan and A.
the expansion onset of substorms?, J. Atmos. Terr. Phys., 55,
T. Y. Lui for suggesting changes to the manuscript. The au- 1151-1157, 1993.
thors would also like to thank the two referees for suggestions Atkinson, G., F. Creutzberg, R. Gattinger, and J. S. Murphree,
which have helped clarify portions of the text. This work was Interpretation of complicated discrete arc structure and
supported under grants from the Natural Sciences and Engi- behavior in terms of multiple X lines, J. Geophys. Res., 94,
neering Research Council of Canada.
5292-5302, 1989.
T. E. Cravens was the Editor responsible for this paper. He Aubry, M.P., C. T. Russell, and M. G. Kivelson, On inward
thanks the technical reviewer, Dan Baker, and the cross-disci- motion of the magnetopause before a substorm, J. Geophys.
plinary reviewer, Alan Robock.
Res., 75, 7018-7031, 1970.
Baker, D. N., and R. L. McPherron, Extreme energetic particle
decreases near geostationary orbit: A manifestation of cur-
REFERENCES
rent diversion within the inner plasma sheet, J. Geophys.
Res., 95, 6591-6599, 1990.
Baker, D. N., P. R. Higbie, E. W. Hones Jr., and R. D. Belian,
Adams, W. G., Comparison of simultaneous magnetic distur- High-resolution energetic particle measurements at 6.6 R e ,
bances of several observatories, Philos. Trans. R. Soc. Lon- 3, Low-energy anisotropies and short-term substorm predon
A, 183, 131-140, 1892.
dictions, J. Geophys. Res., 83, 4863-4868, 1978.
Akasofu, S.-I., The dynamical morphology of the auroral po- Baker, D. N., E. W. Hones Jr., P. R. Higbie, R. D. Belian, and
laris, J. Geophys. Res., 68, 1667-1672, 1963.
P. Stauning, Global properties of the magnetosphere during
Akasofu, S.-I., The development of the auroral substorm, a substorm growth phase: A case study, J. Geophys. Res., 86,
Planet. Space $ci., 12, 273-282, 1964.
8941-8956, 1981.
Akasofu, S.-I., Polar and Magnetospheric Substorms, D. Reidel, Baker, D. N., T. I. Pulkkinen, R. L. McPherron, J. D. Craven,
Norwell, Mass., 1968.
L. A. Frank, R. D. Elphinstone, J. S. Murphree, J. F.
Akasofu, S.-I., Energy coupling between the solar wind and the Fennell, R. E. Lopez, and T. Nagai, CDAW 9 analysis of
magnetosphere, Space Sci. Rev., 28, 121, 1981.
magnetospheric events on May 3, 1986: Event C, J. Geo-
Akasofu, S.-I., and J. R. Kan, Importance of initial ionospheric phys. Res., 98, 3815-3834, 1993.
conductivity on substorm onset, Planet. Space Sci., 30, 1315-
1316, 1982.
Baumjohann, W., The plasma sheet boundary layer and magnetospheric
substorms, J. Geomagn. Geoelectr., 40, 157-175,
Akasofu, S.-I., and A. L. Snyder, Comments on the growth 1988.
phase of magnetospheric substorms, J. Geophys. Res., 77,
6275-6277, 1972.
Baumjohann, W., and Y. Kamide, Joint two-dimensional observations
of ground magnetic and ionospheric electric
Allen, J. H., and H. W. Kroehl, Spatial and temporal distribu- fields associated with auroral zone currents, 2, Three-ditions
of magnetic effects of auroral electrojets as derived mensional current flow in the morning sector during subfrom
AE indices, J. Geophys. Res., 80, 3667-3677, 1975. storm recovery, J. Geomagn. Geoelectr., 33, 297-318, 1981.
Andre, D., and W. Baumjohann, Joint two-dimensional observations
of ground magnetic and ionospheric electric fields
Baumjohann, W., G. Paschmann, and H. Luhr, Characteristics
of high-speed ion flows in the plasma sheet, J. Geophys.
associated with auroral currents, 5, Current system associ- Res., 95, 3801-3809, 1990.
ated with eastward drifting omega bands, J. Geophys., 50, Birn, J., and M. Hesse, Influence of plasma sheet thickening
194-201, 1982.
toward the tail flanks on magnetotail stability and dynamics,
Angelopoulos, V., W. Baumjohann, C. F. Kennel, F. V. Coroniti,
M. G. Kivelson, R. Pellat, R. J. Walker, H. Luhr, and
J. Geophys. Res., 99, 5847-5853, 1994.
Birn, J., E. W. Hones Jr., J. D. Craven, L. A. Frank, R. D.
G. Paschmann, Bursty bulk flows in the inner central Elphinstone, J. S. Murphree, and D. P. Stern, On open and
plasma sheet, J. Geophys. Res., 97, 4027-4039, 1992.
Angelopoulos, V., C. F. Kennel, F. V. Coroniti, R. Pellat,
closed field line regions in Tsyganenko's field model and
their possible associations with horse-collar auroras, J. Geo-
M. G. Kivelson, R. J. Walker, C. T. Russell, W. Baumjo- phys. Res., 96, 3811-3817, 1991.
hann, W. C. Feldman, and J. T. Gosling, Statistical charac- Borovsky, J. E., Auroral arc thicknesses as predicted by various
teristics of bursty bulk flow events, J. Geophys. Res., 99, theories, J. Geophys. Res., 98, 6101-6138, 1993.
21,257-21,280, 1994.
Borovsky, J. E., and D. M. Suszcynsky, Optical measurements

226 ß Elphinstone et al ß MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
of the fine structure of auroral arcs, in Auroral Plasma
Dynamics, Geophys. Monogr. Set., vol. 80, edited by R. L.
Lysak, pp. 25-30, AGU, Washington, D.C., 1993.
Brekke, A., On the evolution in history of the concept of the
auroral oval, Eos Trans. AGU, 65(38), 705-707, 1984.
Brekke, A., and A. Egeland, The Northern Light, p. 62, Springer-Verlag,
New York, 1983.
Bryant, D. A., G. M. Courtier, and A.D. Johnstone, Modulation
of auroral electrons at large distances from the Earth,
J. Atmos. Terr. Phys., 31,579-592, 1969.
Buchert, S., W. Baumjohann, G. Haerendel, C. La Hoz, and H.
Luhr, Magnetometer and incoherent scatter observations of
an intense Ps6 pulsation event, J. Atmos. Terr. Phys., 50, 357,
1988.
Buchert, S., G. Haerendel, and W. Baumjohann, A model for
the electric fields and currents during a strong Ps 6 pulsation
event, J. Geophys. Res., 95, 3733-3743, 1990.
Buchner, J., and L. M. Zelenyi, Chaotization of the electron
motion as the cause of internal magnetotail instability and
substorm onset, J. Geophys. Res., 92, 13,456-13,466, 1987.
Burkhart, G. R., R. E. Lopez, P. B. Dusenbery, and T. W.
Speiser, Observational support for the current sheet catastrophe
model of substorm current disruption, Geophys. Res.
Lett., 19(16), 1635-1638, 1992.
Bythrow, P. E., and T. A. Potemra, Birkeland currents and
energetic particles associated with optical auroral signatures
of a westward traveling surge, J. Geophys. Res., 92,
8691-8699, 1987.
Caan, M. N., R. L. McPherron, and C. T. Russell, Characteristics
of the association between the interplanetary magnetic
field and substorms, J. Geophys. Res., 82, 4837-4842,
1977.
Campbell, W. H., and M. H. Rees, A study of auroral coruscations,
J. Geophys. Res., 66, 41-55, 1961.
Chan, A. A., M. Xia, and L. Chen, Anisotropic Alfv6n-ballooning
modes in Earth's magnetosphere, J. Geophys. Res.,
99, 17,351-17,366, 1994.
Chao, J. K., J. R. Kan, A. T. Y. Lui, and S.-I. Akasofu, A model
for thinning of the plasma sheet, Planet. Space Sci., 25,
703-710, 1977.
Chapman, S., An outline of a theory of magnetic storms, Proc.
R. Soc. London A, 95, 61-77, 1919.
Chapman, S., Earth storms: Retrospect and prospect, J. Phys.
Soc. Jpn., 17, 6, 1962.
Chiu, Y. T., M. Schulz, J. F. Fennell, and A.M. Kishi, Mirror
instability and the origin of morning-side auroral structure,
J. Geophys. Res., 88, 4041-4054, 1983.
Cladis, J. B., and W. E. Francis, Distribution in magnetotail of
O + ions from cusp/cleft ionosphere: A possible substorm
trigger, J. Geophys. Res., 97, 123-130, 1992.
Cogger, L. L., and R. D. Elphinstone, The Viking auroral
substorm, Substorms-l, Proceedings of the First International
Conference on Substorms, Eur. Space Agency Spec. Publ.,
ESA SP-335, 77-82, 1992.
Cogger, L. L., and J. S. Murphree, The UV auroral distribution:
Its impulsive nature, Adv. Space Res., 10, 167-177,
1990.
Coroniti, F. V., and C. F. Kennel, Electron precipitation pulsations,
J. Geophys. Res., 75, 1279-1289, 1970.
Craven, J. L., CtllU I-I 1111%., Lld110I) Of i:IUI Ol Daglis, I. A., E. T. Sarris, and G. Kremser, Indications for
ionospheric participation in the substorm process from
AMPTE/CCE observations, Geophys. Res. Lett., 17(1), 57-
60, 1990.
Daglis, I. A., E. T. Sarris, and G. Kremser, Ionospheric contribution
to the cross-tail current enhancement during the
substorm growth phase, J. Atmos. Terr. Phys., 53, 1091-1098,
1991.
Daglis, I. A., S. Livi, E. T. Sarris, and B. Wilken, Energy
density of ionospheric and solar wind origin ions in the
near-Earth magnetotail during substorms, J. Geophys. Res.,
99, 5691-5703, 1994.
Davis, T. N., The morphology of the auroral displays of 1957-
1958, 2, Detail analyses of Alaska data and analyses of
high-latitude data, J. Geophys. Res., 67, 75-110, 1962.
Davis, T. N., Observed characteristics of auroral forms, Space
Sci. Rev., 22, 77-113, 1978.
Davis, T. N., and T. J. Hallinan, Auroral spirals, 1, Observations,
J. Geophys. Res., 81, 3953-3958, 1976.
Deehr, C. S., Ground-based optical observations of hydrogen
emissions in the auroral substorm, in Proceedings of the
ICS-2, edited by J. R. Kan, J. D. Craven, and S.-I. Akasofu,
pp. 229-236, Geophys. Inst., Univ. of Alaska, Fairbanks,
1994.
de la Beaujardiere, O., L. R. Lyons, J. M. Ruohoniemi,
E. Friis-Christensen, C. Danielsen, F. J. Rich, and P. T.
Newell, Quiet-time intensifications along the poleward auroral
boundary near midnight, J. Geophys. Res., 99, 287-298,
1994.
Donovan, E. F., Modeling the magnetic effects of field-aligned
currents, J. Geophys. Res., 98, 13,529-13,543, 1993.
Eastman, T. E., G. Rostoker, L. A. Frank, C. Y. Huang, and
D. G. Mitchell, Boundary layer dynamics in the description
of magnetospheric substorms, J. Geophys. Res., 93, 14,411-
14,432, 1988.
Edgar, B.C., W. T. Miles, and A. E. S. Green, Energy deposition
of protons in molecular nitrogen and applications to
proton auroral phenomena, J. Geophys. Res., 78, 6595-
6606, 1973.
Eliasson, L., R. Lundin, and J. S. Murphree, Polar cap arcs
observed by the Viking spacecraft, Geophys. Res. Lett.,
14(4), 451-454, 1987.
Elphinstone, R. D., and D. J. Hearn, Mapping of the auroral
distribution during quiet times and substorm recovery, in
Substorms-l, Proceedings of the First International Conference
on Substorms, Eur. Space Agency Spec. Publ., ESA
SP-335, 13-18, 1992.
Elphinstone, R. D., and D. J. Hearn, The auroral distribution
and its relation to magnetospheric processes, Adv. Space
Res., 13(4), 17-27, 1993.
Elphinstone, R. D., J. S. Murphree, L. L. Cogger, D. Hearn,
and R. Lundin, Observations of changes to the auroral
distribution prior to substorm onset, in Magnetospheric Substorms,
Geophys. Monogr. Ser., vol. 64, edited by J. R. Kan,
T. A. Potemra, S. Kokubun, and T. Iijima, pp. 257-275,
AGU, Washington, D.C., 1991a.
Elphinstone, R. D., D. J. Hearn, J. S. Murphree, and L. L.
Cogger, Mapping using the Tsyganenko long magnetospheric
model and its relationship to Viking auroral images,
Uyllallllt; J. Geophys. Res., 96, 1467-1480, i991b.
using global auroral imaging with emphasis on large-scale Elphinstone, R. D., J. S. Murphree, and D. J. Hearn, Three
evolutions, in Auroral Physics, edited by C.-I. Meng, M. J. auroral source regions: Implications for substorm onset, in
Rycroft, and L. A. Frank, pp. 273-288, Cambridge Univ. Recueil d'actes du seminaire du Gdr Plasma, edited by
Press, New York, 1991.
A. Berthelier, R. Pottelette, and C. Huc, pp. 45-50, Cent.
Cresswell, . R., and T. N. Davis, Observations on pulsating de Rech. en Phys. de l'Environ., Gresse en Vercors, France,
aurora, ]. Geophys. Res., ?l, 3155-3161, 1966.
1992.
Daglis, I. A., and W. I. Axford, Fast ionospheric response to Elphinstone, R. D., D. J. Hearn, J. S. Murphree, L. L. Cogger,
enhanced activity in geospace: Ion feeding of the inner M. L. Johnson, and H. B. Vo, Some UV dayside auroral
magnetotail, ]. Geophys. Res., 101, 5047-5065, 1996.
morphologies, in Auroral Plasma Dynamics, Geophys.

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 227
Monogr. Ser., vol. 80, edited by R. L. Lysak, pp. 31-45, ing instrumentation for the Dynamics Explorer mission,
AGU, Washington, D.C., 1993a.
Space Sci. Instrum., 5, 369-393, 1981.
Elphinstone, R. D., J. S. Murphree, D. J. Hearn, W. Heikkila, Fukunishi, H., Dynamic relationship between proton and elec-
L. L. Cogger, and I. Sandahl, The auroral distribution and tron substorms, J. Geophys. Res., 80, 553-574, 1975.
its mapping according to substorm phase, J. Atmos. Terr. Gazey, N. G. J., M. Lockwood, P. N. Smith, S. Coles, R. J.
Phys., 55(14), 1741-1762, 1993b.
Bunting, M. Lester, A.D. Aylward, T. K. Yeoman, and H.
Elphinstone, R. D., D. J. Hearn, L. L. Cogger, I. Sandahl, Luhr, The development of substorm cross-tail current dis-
D. Klumpar, and M. Shapshak, The double oval UV auroral ruption as seen from the ground, J. Geophys. Res., 100,
distribution: Implications for the substorm process, in Pro- 9633-9648, 1995.
ceedings of the ICS-2, edited by J. R. Kan, J. D. Craven, and Goertz, C. K., and R. A. Smith, The thermal catastrophe
S.-I. Akasofu, pp. 413-420, Geophys. Inst., Univ. of Alaska, model of substorms, J. Geophys. Res., 94, 6581-6596, 1989.
Fairbanks, 1994a.
Goertz, C. K., L. H. Shan, and R. A. Smith, Prediction of
Elphinstone, R. D., D. J. Hearn, and J. S. Murphree, Dayside geomagnetic activity, J. Geophys. Res., 98, 7673-7684, 1993.
aurora poleward of the Main Auroral Distribution: Impli- Gonzalez, W. D., J. A. Joselyn, Y. Kamide, H. W. Kroehl,
cations for convection and mapping, Physical Signatures of G. Rostoker, B. T. Tsurutani, and V. M. Vasyliunas, What
Magnetospheric Boundary Layer Processes, edited by J. A. is a geomagnetic storm?, J. Geophys. Res., 99, 5771-5792,
Holtet and A. Egeland, pp. 189-200, Kluwer Acad., Nor- 1994.
well, Mass., 1994b.
Grafe, A., Freja electron precipitation as a hint for an explo-
Elphinstone, R. D., et al., Observations in the vicinity of sive development of the eastward electrojet, Proceedings of
substorm onset: Implications for the substorm process, J. the ICS-2, edited by J. R. Kan, J. D. Craven, and S.-I.
Geophys. Res., 100, 7937-7969, 1995a.
Akasofu, pp. 391-398, Geophys. Inst., Univ. of Alaska,
Elphinstone, R. D., et al., The double oval UV auroral distri- Fairbanks, 1994.
bution, 1, Implications for the mapping of auroral arcs, J. Grant, I. F., and G. B. Burns, Observation and modeling of
Geophys. Res., 100, 12,075-12,092, 1995b.
correlated Pi B magnetic and auroral luminosity pulsations,
Elphinstone, R. D., et al., The double oval UV auroral distri- J. Geophys. Res., 100, 19,387-19,404, 1995.
bution, 2, The most poleward arc system and the dynamics Grant, I. F., G. B. Burns, and K. D. Cole, Pi(b) and Pi2
of the magnetotail, J. Geophys. Res., 100, 12,093-12,102, geomagnetic pulsations associated with substorm onset, in
1995c.
Anare Research Notes, vol. 69, edited by M. Conde and H.
Elvey, C. T., Problems of auroral morphology, Proc. Natl. Beggs, pp. 71-80, Aust. Upper Atmos. and Space Phys.
Acad. Sci. U.S,4.., 43, 63-75, 1957.
Res., Kingston, Tasmania, Australia, 1988.
Erlandson, R. E., D. G. Sibeck, R. E. Lopez, L. J. Zanetti, and Hallinan, T. J., Auroral spirals, 2, Theory, J. Geophys. Res., 81,
T. A. Potemra, Observations of solar wind pressure initi- 3959-3965, 1976.
ated fast mode waves at geostationary orbit and in the polar Hansteen, C., On the polar light or Aurora Borealis and
cap, J. Atmos. Terr. Phys., 53, 231-239, 1991.
Australia, Philos. Mag., 333-334, Nov. 1827.
Fairfield, D. H., Advances in magnetospheric storm and sub- Hearn, D. J., R. D. Elphinstone, J. S. Murphree, and L. L.
storm research, J. Geophys. Res., 97, 10,865-10,874, 1992. Cogger, Geographic asymmetries of the Viking auroral
Fairfield, D. H., and L. J. Cahill Jr., Transition region magnetic distribution: Implications for ionospheric coordinate sysfield
and polar magnetic disturbances, J. Geophys. Res., 71, tems, J. Geophys. Res., 98, 1653-1667, 1993.
155-169, 1966.
Hesse, M., and J. Birn, Magnetosphere-ionosphere coupling
Fedder, J. A., S. P. Slinker, J. G. Lyon, and R. D. Elphinstone, during plasmoid evolution: First results, J. Geophys. Res.,
Global numerical simulation of the growth phase and ex- 96, 11,513-11,522, 1991.
pansion onset of a substorm observed by Viking, J. Geophys. Hesse, M., and J. Birn, On the energy budget in the current
Res., 100, 9083-9094, 1995.
disruption region, Geophys. Res. Lett., 20(14), 1451-1454,
Feldstein, Y. I., Geographical distribution of aurorae and 1993.
auroral arcs azimuths, in Investigations of Aurorae, edited by Hewson, E. W., A survey of the facts and theories of the
B. A. Bagarjatsky, pp. 61-78, Publ. House Acad. of Sci., aurora, Rev. Mod. Phys., 9, 403-431, 1937.
Russia, 1960.
Hoffman, R. A., and J. L. Burch, Electron precipitation pat-
Feldstein, Y. I., Some problems concerning the morphology of terns and substorm morphology, J. Geophys. Res., 78, 2867auroras
and magnetic disturbances at high latitudes, Geo- 2884, 1973.
magn. Aeron., 3, 183-192, 1963.
Hones, E. W., Jr., Poleward leaping auroras, the substorm
Feldstein, Y. I., Modelling of the magnetic field of magneto- expansive and recovery phases and the recovery of the
spheric ring current as a function of interplanetary medium plasma sheet, in Substorms-I, Proceedings of the First Interparameters,
Space Sci. Rev., 59, 83-165, 1992.
national Conference on Substorms, Eur. Space Agency Spec.
Feldstein, Y. I., and Y. I. Galperin, The auroral luminosity Publ., ESA SP-335, 477-483, 1992.
structure in the high-latitude upper atmosphere: Its dynam- Hones, E. W., Jr., T. J. Rosenberg, K. A. Anderson, G. K.
ics and relationship to the large-scale structure of the Parks, and H. J. Singer, The poleward leap of the auroral
Earth's magnetosphere, Rev. Geophys., 23, 217-275, 1985. electrojet as seen in auroral images, J. Geophys. Res., 90,
Feldstein, Y. I., and G. V. Starkov, Dynamics of auroral belt 5333-5337, 1985.
and polar geomagnetic disturbances, Planet. Space Sci., 15, Hones, E. W., Jr., T. A. Fritz, J. Birn, J. Cooney, and S. J.
209-229, 1967.
Bame, Detailed observations of the plasma sheet during a
Forbes, T. G., E. W. Hones Jr., S. J. Bame, J. R. Asbridge, substorm on April 24, 1979, J. Geophys. Res., 91, 6845-6859,
G. Paschmann, N. Sckopke, and C. T. Russell, Substorm- 1986.
related plasma sheet motions as determined from differen- Hones, E. W., Jr., C. D. Anger, J. Birn, J. S. Murphree, and
tial timing of plasma changes at the ISEE satellites, J. L. L. Cogger, A study of a magnetospheric substorm re-
Geophys. Res., 86, 3459-3469, 1981.
corded by the Viking auroral imager, Geophys. Res. Lett.,
Frank, L. A., and J. D. Craven, Imaging results from Dynamics 14(4), 411-414, 1987.
Explorer 1, Rev. Geophys., 26, 249-283, 1988.
Hopcraft, K. I., and P. R. Smith, Polarization characteristics of
Frank, L. A., J. D. Craven, K. L. Ackerson, M. R. English, Pi 2 oscillations, Planet. Space Sci., 34, 1259-1261, 1986.
R. H. Eather, and R. L. Carovillano, Global auroral imag- Horowitz, J. L., The substorm as an internal magnetospheric

228 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
instability: Substorms and their characteristic time scales Koskinen, H. E. J., R. E. Lopez, R. J. Pellinen, T. I. Pulkkinen,
during intervals of steady interplanetary magnetic field, J. D. N. Baker, and T. Bosinger, Pseudobreakup and sub-
Geophys. Res., 90, 4164-4170, 1985.
storm growth phase in the ionosphere and magnetosphere,
Huang, C. Y., L. A. Frank, G. Rostoker, J. Fennell, and D. G. J. Geophys. Res., 98, 5801-5813, 1993.
Mitchell, Nonadiabatic heating of the central plasma sheet Kotikov, A. L., A. V. Frank-Kamenetsky, Y. O. Latov, O. A.
at substorm onset, J. Geophys. Res., 97, 1481-1495, 1992. Troshichev, E. M. Shishkina, J. S. Murphree, and R. D.
Hudson, M. K., A.D. Kotelnikov, X. Li, I. Roth, M. Temerin, Elphinstone, Filamentary structure of the westward electro-
J. Wygant, J. B. Blake, and M. S. Gussenhoven, Simulation jet in the midnight sector auroral distribution during subof
proton radiation belt formation during the March 24, storms: Comparison with Viking auroral observations, J.
1991, SSC, Geophys. Res. Lett., 22(3), 291-294, 1995. Atmos. Terr. Phys., 55(14), 1763-1774, 1993.
Iijima, T., and T. A. Potemra, Large-scale characteristics of Kropotkin, A. P., On the physical mechanism of the magnetofield-aligned
currents associated with substorms, J. Geo- spheric substorm development, Planet. Space Sci., 20, 1245phys.
Res., 83, 599-615, 1978.
1257, 1972.
Ismail, S., and C.-I. Meng, A classification of polar cap auroral Lester, M., W. J. Hughes, and H. J. Singer, Polarization patarcs,
Planet. Space Sci., 30, 319-330, 1982.
terns of Pi 2 magnetic pulsations and the substorm current
Ivanov, V. N., and O. P. Pokhotelov, Flute instability in the wedge, J. Geophys. Res., 88, 7958-7966, 1983.
plasma shell of the Earth's magnetosphere, Sov. J. Plasma Lester, M., M. Lockwood, T. K. Yeoman, S. W. H. Cowley, H.
Phys., Engl. Transl. 13, 833-838, 1987.
Luhr, R. Bunting, and C. J. Farrugia, The response of
Ivanov, V. N., O. P. Pokhotelov, F. Z. Frigin, A. Roux, and S. ionospheric convection in the polar cap to substorm activity,
Perraut, Magnetosphere with irregular pressure and a finite Ann. Geophys., 13, 147-158, 1995.
beta, Geomagn. Aeron., 32, 211-216, 1992.
Lindqwister, U. J., C. H. Ho, A. J. Mannucci, and B. T.
Iyemori, T., and S. Tsunomura, Characteristics of the associa- Tsurutani, Forecasting ionospheric storms using GPS
tion between an SC and a substorm onset, Mere. Natl. Inst. global network, Eos Trans. AGU, 76(46), Fall Meet. Suppl.,
Polar Res. Spec. Issue Jpn., 26, 139-184, 1983.
433-434, 1995.
Jacobs, J. A., T. Kato, S. Matsushita, and V. A. Troitskaya, Lopez, R. E., and A. T. Y. Lui, A multisatellite case study of
Classification of geomagnetic micropulsations, J. Geophys. the expansion of a substorm current wedge in the near-
Res., 69, 180-181, 1964.
Earth magnetotail, J. Geophys. Res., 95, 8009-8017, 1990.
Johnstone, A.D., The mechanism of pulsating aurora, Ann. Lopez, R. E., H. E. J. Koskinen, T. I. Pulkkinen, T. Bosinger,
Geophys., 1,397-410, 1983.
R. W. McEntire, and T. A. Potemra, Simultaneous obser-
Jordan, C. E., Empirical models of the magnetospheric mag- vation of the poleward expansion of substorm electrojet
netic field, Rev. Geophys., 32, 139-157, 1994.
activity and the tailward expansion of current sheet disrup-
Joselyn, J. A., Geomagnetic activity forecasting: The state of tion in the near-Earth magnetotail, J. Geophys. Res., 98,
the art, Rev. Geophys., 33, 383-401, 1995.
9285-9295, 1993.
Kamide, Y. P., D. Perreault, S.-I. Akasofu, and J. D. Winning- Lui, A. T. Y., Estimates of current changes in the geomagneham,
Dependence of substorm occurrence probability on totail associated with a substorm, Geophys. Res. Lett., 5(10),
the interplanetary magnetic field and on the size of the 853-856, 1978.
auroral oval, J. Geophys. Res., 82, 5521-5528, 1977a.
Lui, A. T. Y., A synthesis of magnetospheric substorm models,
Kamide, Y., S.-I. Akasofu, and E. P. Reiger, Coexistence of
J. Geophys. Res., 96, 1849-1856, 1991.
two substorms in the midnight sector, J. Geophys. Res., 82,
Lui, A. T. Y., and J. R. Burrows, On the location of auroral
1620-1624, 1977b.
arcs near substorm onsets, J. Geophys. Res., 83, 3342-3348,
Kan, J. R., A global magnetosphere-ionosphere coupling 1978.
model of substorms, J. Geophys. Res., 98, 17,263-17,275,
Lui, A. T. Y., S.-I. Akasofu, E. W. Hones Jr., S. J. Bame, and
1993.
C. E. Mcllwain, Observation of the plasma sheet during a
Kan, J. R., and W. Sun, Simulation of the westward traveling
surge and Pi 2 pulsations during substorms, J. Geophys. Res.,
contracted oval substorm in a prolonged quiet period, J.
90, 10,911-10,922, 1985.
Geophys. Res., 81, 1415-1419, 1976.
Kan, J. R., L. Zhu, and S.-I. Akasofu, A theory of substorms:
Lui, A. T. Y., C.-I. Meng, and S. Ismail, Large-amplitude
Onset and subsidence, J. Geophys. Res., 93, 5624-5640, undulations on the equatorward boundary of the diffuse
1988.
aurora, J. Geophys. Res., 87, 2385-2400, 1982.
Kaneda, E., Dayside auroral activity and its relation to sub- Lui, A. T. Y., R. E. Lopez, S. M. Krimigis, R. W. McEntire,
storm, Rep. Ionos. Space Res. Jpn., 27, 209-212, 1973. L. J. Zanetti, and T. A. Potemra, A case study of magne-
Kaneda, E., M. Takagi, and N. Niwa, Vacuum ultraviolet totail current sheet disruption and diversion, Geophys. Res.
television camera, paper presented at 12th International Lett., 15(7), 721-724, 1988.
Symposium of Space Technology and Science, Tokyo, 1977. Lui, A. T. Y., C.-L. Chang, A. Mankofsky, H.-K. Wong, and D.
Kennel, C. F., The Kiruna conjecture: The strong version, in Winske, A cross-field current instability for substorm ex-
Substorms-l, Proceedings of the First International Confer- pansions, J. Geophys. Res., 96, 11,389-11,401, 1991.
ence on Substorms, Eur. Space Agency Spec. Publ., ESA Lui, A. T. Y., R. E. Lopez, B. J. Anderson, K. Takahashi, L. J.
SP-335, 599-601, 1992.
Zanetti, R. W. McEntire, T. A. Potemra, D. M. Klumpar,
Kern, J. W., and E. H. Vestine, Theory of auroral morphology, E. M. Greene, and R. Strangeway, Current disruptions in
Kirkwood, S., and L. Eliasson, Energetic particle precipitation 1461-1480, 1992.
in the substorm growth phase measured by EISCAT and Lyons, L. R., and C. Y. Huang, Observations of plasma sheet
Viking, J. Geophys. Res., 95, 6025-6037, 1990.
expansion at substorm onset, R = 15 to 22 Rœ, Geophys.
Kivelson, M. G., and D. J. Southwood, Resonant ULF waves: Res. Lett., 19(18), 1807-1810, 1992.
A new interpretation, Geophys. Res. Lett., 12(1), 49-52, Lyons, L. R., and R. L. Walterscheid, Generation of auroral
1985.
omega bands by shear instability of the neutral winds, J.
Kokubun, S., R. L. McPherron, and C. T. Russell, Triggering Geophys. Res., 90, 12,321-12,329, 1985.
of substorms by solar wind discontinuities, J. Geophys. Res., Lysak, R. L., J. Grieger, and Y. Song, Fast ionospheric feed-
82, 74-86, 1977.
back instability and substorm onset, in Substorms-I, Pro-

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 229
ceedings of the First International Conference on Substorms, Nakamura, R., and T. Oguti, Drifts of auroral structures and
Eur. Space Agency Spec. Publ., ESA SP-335, 231-236, 1992. magnetospheric electric fields, J. Geophys. Res., 92, 11,241-
Lysak, R. L., Y. Song, and J. C. Grieger, Coupling of the 11,247, 1987.
magnetopause to the ionosphere by means of Allyen waves Nakamura, R., T. Oguti, and GADC Research Group, Drifts
and field-aligned currents, in Physics of the Magnetopause, of auroral structures and their relationship to geomagnetic
Geophys. Monogr. Ser., vol. 90, edited by P. Song, B. U. O. activity, in Proc. Natl. Inst. Polar Res. Symp. Upper Atmos.
Sonnerup, and M. F. Thomsen, pp. 385-393, AGU Wash- Phys., 2, 96-102, 1989.
ington, D.C., 1995.
Nakamura, R., T. Oguti, T. Yamamoto, and S. Kokubun,
McEwen, D. J., E. Yee, B. A. Whalen, and A. W. Yau, Equatorward and poleward expansion of the auroras during
Electron energy measurements in pulsating auroras, Can. J. auroral substorms, J. Geophys. Res., 98, 5743-5759, 1993.
Phys., 59, 1106-1115, 1981.
Nakamura, R., D. N. Baker, T. Yamamoto, R. D. Belian, E. A.
McHenry, M. A., C. R. Clauer, and E. Friis-Christensen, Bering III, J. R. Benbrook, and J. R. Theall, Particle and
Relationship of solar wind parameters to continuous, day- field signatures during pseudobreakup and major expansion
side high-latitude traveling ionospheric convection vortices, onset, J. Geophys. Res., 99, 207-221, 1994.
J. Geophys. Res., 95, 15,007-15,022, 1990.
Newell, P. T., and C.-I. Meng, Mapping the dayside ionosphere
McPherron, R. L., Growth phase of magnetospheric sub- to the magnetosphere according to particle precipitation
storms, J. Geophys. Res., 75, 5592-5599, 1970.
characteristics, Geophys. Res. Lett., 19(6), 609-612, 1992.
McPherron, R. L., Substorm related changes in the geomag- Newell, P. T., S. Wing, C.-I. Meng, and V. Sigillito, The auroral
netic tail: The growth phase, Planet. Space Sci., 20, 1521- oval position, structure, and intensity of precipitation from
1539, 1972.
1984 onward: An automated on-line database, J. Geophys.
McPherron, R. L., Critical problems in establishing the mor- Res., 96, 5877-5882, 1991.
phology of substorms in space, in Magnetospheric Physics, Nielsen, E., J. Bamber, Z.-S. Chen, A. Brekke, A. Egeland,
edited by B. M. McCormac, pp. 335-347, D. Reidel, Nor- J. S. Murphree, D. Venkatesan, and W. I. Axford, Substorm
well, Mass., 1974.
expansion into the polar cap, Ann. Geophys., 6, 559-572,
McPherron, R. L., Magnetospheric substorms, Rev. Geophys., 1988.
17, 657-681, 1979.
Nielsen, E., R. D. Elphinstone, D. J. Hearn, J. S. Murphree,
Meng, C.-I., and R. Lundin, Auroral morphology of the mid- and T. Potemra, Oval intensification event observed by
day oval, J. Geophys. Res., 91, 1572-1584, 1986.
STARE and Viking, J. Geophys. Res., 98, 6163-6171, 1993.
Moldwin, M. B., and W. J. Hughes, Observations of earthward Nishida, A., Excitation of polar substorms by northward interand
tailward propagating flux rope plasmoids: Expanding planetary magnetic field, in Earth's Magnetospheric Prothe
plasmoid model of geomagnetic substorms, J. Geophys. cesses, edited by B. M. McCormac, pp. 400-405, D. Reidel,
Res., 99, 183-198, 1994.
Norwell, Mass., 1972.
Montbriand, L. E., The proton aurora substorm, in The Radi- Nishida, A., and Y. Kamide, Magnetospheric processes preating
Atmosphere, edited by B. M. McCormac, pp. 366-373, ceding the onset of an isolated substorm: A case study of
D. Reidel, Norwell, Mass., 1971.
the March 31, 1978, substorm, J. Geophys. Res., 88, 7005-
Montbriand, L. E., Proton auroras during substorms, in Sub- 7014, 1983.
storms-l, Proceedings of the First International Conference on Nishida, A., S. J. Bame, D. N. Baker, G. Gloeckler, M. Scholer,
Substorms, Eur. Space Agency Spec. Publ., ESA SP-335, E. J. Smith, T. Terasawa, and B. Tsurutani, Assessment of
445-456, 1992.
the boundary layer model of the magnetospheric substorm,
Moshupi, M. C., C. D. Anger, J. S. Murphree, D. D. Wallis, J. Geophys. Res., 93, 5579-5588, 1988.
J. H. Whitteker, and L. H. Brace, Characteristics of trough Nordenskiold, A. E., Vegas Fard Kdng Asien och Europe, F. &
region auroral patches and detached arcs observed by ISIS G. Beijer, Stockholm, 1880-1881.
2, J. Geophys. Res., 84, 1333-1346, 1979.
Oguti, T., TV observations of auroral arcs, in Physics of Auroral
Murphree, J. S., and L. L. Cogger, Observations of substorm Arc Formation, Geophys. Monogr. Ser., vol. 25, edited by
onset, in Substorms-l, Proceedings of the First International S.-I. Akasofu and J. R. Kan, pp. 31-41, AGU, Washington,
Conference on Substorms, Eur. Space Agency Spec. Publ., D.C., 1981.
ESA SP-335, 207-211, 1992.
Oguti, T., Modes of pulsating auroras and related geomagnetic
Murphree, J. S., and R. D. Elphinstone, Correlative studies pulsations, in Proceedings of the Fourth Symposium on Cousing
the Viking imagery, Adv. Space Res., 8, 9-19, 1988. ordinated Observations of the Ionosphere and the Magneto-
Murphree, J. S., L. L. Cogger, C. D. Anger, D. D. Wallis, and sphere in the Polar Regions, Mem. Natl. Inst. Polar Res. Spec.
G. G. Shepherd, Oval intensification associated with polar Issue Jpn., 22, 94-103, 1982.
arcs, Geophys. Res. Lett., 4(4), 403-406, 1987.
Oguti, T., Pulsating aurorae, indicators of substorm recovery
Murphree, J. S., L. L. Cogger, and R. D. Elphinstone, Obser- processes, in Substorms-I, Proceedings of the First Internavations
of distortions of optical features in the UV auroral tional Conference on Substorms, Eur. Space Agency Spec.
distribution, IEEE Trans. Plasma Sci., 17, 109-115, 1989. Publ., ESA SP-335, 433-444, 1992.
Murphree, J. S., R. D. Elphinstone, D. J. Hearn, and L. L. Oguti, T., and K. Hayashi, Multiple correlation between au-
Cogget, Viking optical substorm signatures, in Magneto- roral and magnetic pulsations, 2, Determination of electric
spheric Substorms, Geophys. Monogr. Ser., vol. 64, edited by currents and electric fields around a pulsating auroral
J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima, pp. patch, J. Geophys. Res., 89, 7476-7481, 1984.
241-256, AGU, Washington, D.C., 1991.
Oguti, T., and K. Hayashi, Polarization and wave form of
Murphree, J. S., R. D. Elphinstone, L. L. Cogger, and D. J. magnetic pulsations below pulsating auroras: Magnetic ef-
Hearn, Interpretation of optical substorm onset observa- fects of electric currents induced in an ionization tail of a
tions, J. Atmos. Terr. Phys., 55, 1159-1170, 1993.
moving auroral patch, J. Geomagn. Geoelectr., 37, 65-91,
Murphree, J. S., R. A. King, T. Payne, K. Smith, D. Reid, J. 1985.
Adema, B. Gordon, and R. Wlochowicz, The Freja ultravi- Oguti, T., and T. Watanabe, Quasi-periodic poleward propaolet
imager, Space Sci. Rev., 70, 421-446, 1994a.
gation of on-off switching aurora and associated geomag-
Murphree, J. S., J. B. Austin, D. J. Hearn, L. L. Cogger, R. D. netic pulsations in the dawn, J. Atmos. Terr. Phys., 38,
Elphinstone, and J. Woch, Satellite observations of polar 543-551, 1976.
arcs, J. Atmos. Terr. Phys., 56, 265-284, 1994b.
Oguti, T., S. Kokubun, K. Hayashi, K. Tsuruda, S. Machida,

230 ß Elphinstone et al.' MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
T. Kitamura, O. Saka, and T. Watanabe, An auroral torch vol. 64, edited by J. R. Kan, T. A. Potemra, S. Kokubun, and
structure as an activity center of pulsating auroras, Can. J. T. Iijima, pp. 17-27, AGU, Washington, D.C., 1991.
Phys., 59, 1056-1062, 1981a.
Pudovkin, M. I., V. S. Semenov, G. V. Starkov, and T. A.
Oguti, T., S. Kokubun, K. Hayashi, K. Tsuruda, S. Machida, Kornilova, On separation of the potential and vortex parts
T. Kitamura, O. Saka, and T. Watanabe, Statistics of pul- of the magnetotail electric field, Planet. Space Sci., 39,
sating auroras on the basis of all-sky TV data from five 563-568, 1991.
stations, 1, Occurrence frequency, Can. J. Phys., 59, 1150- Pulkkinen, T. I., et al., Modeling the growth phase of a sub-
1157, 1981b.
storm using the Tsyganenko model and multispacecraft
Oguti, T., J. H. Meek, and K. Hayashi, Multiple correlation observations: CDAW-9, Geophys. Res. Lett., 18(11), 1963between
auroral and magnetic pulsations, J. Geophys. Res., 1966, 1991a.
89, 2295-2303, 1984.
Pulkkinen, T. I., R. J. Pellinen, H. E. J. Koskinen, H. J.
Oguti, T., K. Hayashi, T. Yamamoto, J. Ishida, T. Higuchi, and Opgenoorth, J. S. Murphree, V. Petrov, A. Zaitzev, and
N. Nishitani, Absence of hydromagnetic waves in the mag- E. Friis-Christensen, Auroral signatures of substorm recovnetospheric
equatorial region conjugate with pulsating au- ery phase: A case study, in Magnetospheric Substorms, Geororas,
J. Geophys. Res., 91, 13,711-13,715, 1986.
phys. Monogr. Ser., vol. 64, edited by J. R. Kan, T. A.
Oguti, T., et al., Fast auroral evolution and related magnetic Potemra, S. Kokubun, and T. Iijima, pp. 333-341, AGU,
field changes on the ground and at conjugate satellites, J. Washington, D.C., 1991b.
Geomagn. Geoelectr., 40, 505-536, 1988.
Pulkkinen, T. I., D. N. Baker, R. J. Pellinen, J. Buchner, H. E.
Oguti, T., E. Kaneda, M. Ejiri, S. Sasaki, T. Yamamoto, K. J. Koskinen, R. E. Lopez, R. L. Dyson, and L. A. Frank,
Hayashi, R. Fujii, and K. Makita, Studies of auroral dynam- Particle scattering and current sheet stability in the geomagics
by aurora-TV on the Akebono (EXOS-D) satellite, J. netic tail during the substorm growth phase, J. Geophys.
Geomagn. Geoelectr., 42, 555-564, 1990.
Res., 97, 19,283-19,297, 1992.
Oguti, T., N. Nishitani, and R. Nakamura, Auroral activity and Raeder, J., J. Berchem, and M. Ashour-Abdalla, The imporits
connection with magnetospheric processes--Magnetic tance of small scale processes in global MHD simulations:
field conjugacy between the polar ionosphere and the mag- Some numerical experiments, in Physics of Space Plasmas,
netosphere, J. Geomagn. Geoelectr., 43, 353-368, 1991. No. 14, edited by T. Chang and J. R. Jasperse, Mass. Inst.
Ohtani, S., and T. Tamao, Does the ballooning instability Technol. Cent. for Theor. Geo/Cosmo Plasma Phys., Camtrigger
substorms in the near-Earth magnetotail, J. Geo- bridge, in press 1996.
phys. Res., 98, 19,369-19,379, 1993.
Rees, M. H., and D. Lummerzheim, in Auroral Physics, edited
Ohtani, S., et al., A multisatellite study of a pseudosubstorm by C.-I. Meng, M. J. Rycroft, and L. A. Frank, pp. 29-35,
onset in the near-Earth magnetotail, J. Geophys. Res., 98, Cambridge Univ. Press, New York, 1991.
19,355-19,367, 1993.
Rogers, E. H., D. F. Nelson, and R. C. Savage, Auroral
Ohtani, S., et al., Four large-scale field-aligned current systems photography from a satellite, Science, 183, 951-952, 1974.
in the dayside high-latitude region, J. Geophys. Res., 100, Rossberg, L., The effect of the solar wind velocity on substorm
137-153, 1995.
activity, J. Geophys. Res., 94, 13,571-13,574, 1989.
Opgenoorth, H., J. Oksman, K. U. Kaila, E. Nielsen, and
Rostoker, G., Ground based magnetic signatures of the phases
W. Baumjohann, On the characteristics of eastward drifting
of magnetospheric substorms, in Magnetospheric Physics,
omega bands in the morning sector, J. Geophys. Res., 88,
edited by B. M. McCormac, pp. 325-333, D. Reidel, Norwell,
Mass., 1974.
9171-9185, 1983.
Rostoker, G., and T. Eastman, A boundary layer model for
Opgenoorth, H. J., et al., Coordinated observations with
magnetospheric substorms, J. Geophys. Res., 92, 12,187-
EISCAT and the Viking satellite: The decay of a westward
12,201, 1987.
travelling surge, Ann. Geophys., 7, 479-500, 1989.
Rostoker, G., and J. C. Samson, Can substorm expansion
Pedersen, B. M., R. Pottelette, L. Eliasson, J. S. Murphree,
phase effects and low-frequency Pc magnetic pulsations be
R. D. Elphinstone, A. Bahnsen, and M. Jespersen, Auroral
attributed to the same source mechanism?, Geophys. Res.
kilometric radiation from transpolar arcs, J. Geophys. Res.,
Lett., (3), 271-274, 1984.
97, 10,567-10,573, 1992.
Rostoker, G., S.-I. Akasofu, J. Forster, R. A. Greenwald,
Pellinen, R. J., and W. J. Heikkila, Observations of auroral Y. Kamide, K. Kawasaki, A. T. Y. Lui, R. L. McPherron,
fading before breakup, J. Geophys. Res., 83, 4207-4217, and C. T. Russell, Magnetospheric substorms: Definition
1978.
and signatures, J. Geophys. Res., 85, 1663-1668, 1980.
Pellinen, R. J., H. E. J. Koskinen, T. I. Pulkkinen, J. S. Roux, A., S. Perraut, P. Robert, A. Morane, A. Pedersen,
Murphree, G. Rostoker, and H. J. Opgenoorth, Satellite A. Korth, G. Kremser, B. Aparicio, D. Rodgers, and
and ground-based observations of a fading transpolar arc, J. R. Pellinen, Plasmasheet instability related to the westward
Geophys. Res., 95, 5817-5824, 1990.
traveling surge, J. Geophys. Res., 96, 17,697-17,714, 1991a.
Pellinen, R. J., T. I. Pulkkinen, A. Huuskonen, and K.-H. Roux, A., S. Perraut, A. Morane, P. Robert, A. Korth,
Glassmeier, On the dynamical development of the down- G. Kremser, A. Pederson, R. Pellinen, and Z. Y. Pu, Role
ward field-aligned current in the substorm current wedge, J. of the near-Earth plasmasheet at substorms, in Magneto-
Geophys. Res., 100, 14,863-14,873, 1995.
spheric Substorms, Geophys. Monogr. Ser., vol. 64, edited by
Perona, G. E., Theory on the precipitation of magnetospheric J. R. Kan, T. A. Potemra, S. Kokubun, and T. Iijima, pp.
eleo* .................................................... ..... , ,ho ,;,,. ,,, a ,,aa ............. * Z r2_.... - - , , hng ton, "' ta. w., ' 1991b.
phys. Res., 77, 101-111, 1972.
Samson, J. C., L. R. Lyons, P. T. Newell, F. Creutzberg, and B.
Persson, M. A. L., A. T. Aikio, and H. J. Opgenoorth, Satellite- Xu, Proton aurora and substorm intensifications, Geophys.
ground-based coordination: Late growth and early expan- Res. Lett., 19(21), 2167-2170, 1992.
sion phase of a substorm, in International Conference on Sandahl, I., L. Eliasson, and R. Lundin, Rocket observations
Substorms-2, edited by J. R. Kan, J. D. Craven, and S.-I. over a pulsating aurora, Geophys. Res. Lett., 7(5), 309-312,
Akasofu, pp. 157-164, Geophys. Inst., Univ. of Alaska, 1980.
Fairbanks, 1994.
Sandholt, P. E., M. Lockwood, T. Oguti, S. W. H. Cowley,
Pudovkin, M. I., Physics of magnetospheric substorms: A re- K. S.C. Freeman, B. Lybekk, A. Egeland, and D. M. Willis,
view, in Magnetospheric Substorms, Geophys. Monogr. Ser., Midday auroral breakup events and related energy and

34, 2 / REVIEWS OF GEOPHYSICS Elphinstone et al.: MODULAR AURORAL ACTIVITY ß 231
momentum transfer from the magnetosheath, J. Geophys. Earth's magnetotail, J. Geophys. Res., 98, 15,425-15,446,
Res., 95, 1039-1060, 1990.
1993.
Sauvaud, J. A., J. Crasnier, K. Mouala, R. A. Kovrazhkin, and Stagg, J. M., British Polar Year Expedition, Fort Rae, N. W.,
N. V. Jorjio, Morning sector ion precipitation following Canada, 1932-33, vol. 1, Discussion of Results, R. Soc.,
substorm injections, J. Geophys. Res., 86, 3430-3438, 1981. Burlington House, London, 1937.
Sauvaud, J. A., et al., Large-scale response of the magneto- Starkov, G. V., Y. I. Feldstein, and F. Shevnina, Auroras on
sphere to a southward turning of the interplanetary mag- the day side of the oval during substorms, Geomagn. Aeron.,
netic field, J. Geophys. Res., 92, 2365-2376, 1987.
11, 72-75, 1971.
Scholer, M., G. Gloeckler, D. Hovestadt, F. M. Ipavich, Stenbaek-Nielsen, H. C., T. N. Davis, and N. W. Glass, Rela-
B. Ilecker, D. N. Baker, W. Baumjohann, B. T. Tsurutani, tive motion of auroral conjugate points during substorms, J.
and R. D. Zwickl, Simultaneous observation of the plasma Geophys. Res., 77, 1844-1858, 1972.
sheet, Geophys. Res. Lett., 11(10), 1034-1037, 1984. Takahashi, K., S.-I. Ohtani, and K. Yumoto, AMPTE CCE
Sergeev, V. A., The discrete activations of the magnetosphere observations of Pi 2 pulsations in the inner magnetosphere,
during the substorm explosive phase, paper presented at Geophys. Res. Lett., 19(14), 1447-1450, 1992.
Solar Terrestrial Physics Symposium, Sao Paulo, Brazil, Trefall, H., S. Ullaland, J. Stadnes, I. Signstad, T. Pytte,
1974.
K. Bronstad, J. Bjordal, R. H. Karas, R. R. Brown, and
Sergeev, V. A., High time-resolution correlation between the J. Munch, Morphology and fine time structure of an early
magnetic field behavior at 37 Rœ distance in the magneto- morning electron precipitation event, J. Atmos. Terr. Phys.,
tail plasma sheet and ground phenomena during substorm 37, 83-105, 1975.
expansive phase, J. Geophys., 49, 176-185, 1981.
Troshichev, O. A., N. P. Dmitrieva, and B. M. Kuznetsov,
Sergeev, V. A., and A. G. Yahnin, The features of auroral Polar cap magnetic activity as a signature of substorm
bulge expansion, Planet. Space Sci., 27, 1429-1440, 1979. development, Planet. Space Sci., 27, 217-221, 1979.
Sergeev, V., E. M. Sazhina, N. A. Tsyganenko, J. A. Lundblad, Tsurutani, B. T., T. Gould, B. E. Goldstein, W. D. Gonzalez,
and F. Soraas, Pitch-angle scattering of energetic protons in and M. Sugiura, Interplanetary Alfven waves and auroral
the magnetotail current sheet as the dominant source of (substorm) activity: IMP 8, J. Geophys. Res., 95, 2241-2252,
their isotropic precipitation into the nightside ionosphere, 1990.
Planet. Space Sci., 31, 1147-1155, 1983.
Tsyganenko, N. A., Global quantitative models of the geomag-
Sergeev, V. A., V. S. Semenov, and M. V. Sidneva, Impulsive netic field in the cislunar magnetosphere for different disreconnection
in the magnetotail during substorm expan- turbance levels, Planet. Space Sci., 35, 1347-1358, 1987.
sion, Planet. Space Sci., 35, 1199-1212, 1987.
Tsyganenko, N. A., A magnetospheric magnetic field model
Sergeev, V. A., P. Tanskanen, K. Mursula, A. Korth, and R. C. with a warped tail current sheet, Planet. Space Sci., 37, 5-20,
Elphic, Current sheet thickness in the near-Earth plasma 1989.
sheet during substorm growth phase, J. Geophys. Res., 95, Ullaland, S., G. Kremser, P. Tanskanen, A. Korth, A. Roux,
3819-3827, 1990.
K. Torkar, L. P. Block, and I. B. Iversen, On the develop-
Sergeev, V. A., B. Aparicio, S. Perraut, M. V. Malkov, and ment of a magnetospheric substorm influenced by a
R. J. Pellinen, Structure of the inner plasma sheet at mid- storm sudden commencement: Ground, balloon, and satnight
during steady convection, Planet. Space Sci., 39, 1083- ellite observations, J. Geophys. Res., 98, 15,381-15,401,
1096, 1991.
1993.
Sergeev, V., M. Malkov, and K. Mursula, Testing the isotropic Vallance Jones, A., Optical emissions from aurora, in Aurora,
boundary algorithm method to evaluate the magnetic field Geophys. and Astrophys. Monogr., vol. 9, pp. 80-128,
configuration in the tail, J. Geophys. Res., 98, 7609-7620, D. Reidel, Norwell, Mass., 1974.
1993.
Vallance Jones, A., Overview of auroral spectroscopy, in
Sergeev, V., L. Vagina, R. D. Elphinstone, J. S. Murphree, Auroral Physics, edited by C.-I. Meng, M. J. Rycroft, and L.
D. J. Hearn, L. L. Cogger, M. Johnson, Comparison of UV A. Frank, pp. 15-28, Cambridge Univ. Press, New York,
optical signatures with the substorm current wedge as pre- 1991.
dicted by an inversion algorithm, J. Geophys. Res., 101, Vallance Jones, A., F. Creutzberg, R. L. Gattinger, and F. R.
2615-2627, 1996.
Harris, Auroral studies with a chain of meridian-scanning
Shepherd, G. G., J. S. Murphree, and C. D. Anger, Observa- photometers, 1, Observations of proton and electron aurora
tions of discrete arcs with the Viking ultraviolet imager, in magnetospheric substorms, J. Geophys. Res., 87, 4489-
Ann. Geophys., 6, 153-157, 1988.
4503, 1982.
Shumilov, O. I., O. M. Raspopov, E. A. Kasatkina, R. D. Vassiliadis, D., A. J. Klimas, D. N. Baker, and D. A. Roberts,
Elphinstone, and R. Kreutsberg, The dynamics of particle A description of the solar wind-magnetospheric coupling
precipitation in the polar cap during an SC: Examination of based on nonlinear filters, J. Geophys. Res., 100, 3495-3512,
a specific event, Geomagn. Aeron., 31, 477-484, 1991.
1995.
Singer, H. J., W. J. Hughes, P. F. Fougere, and D. J. Knight, Vetoulis, G., and L. Chen, Global structures of Alfven-bal-
The localization of Pi 2 pulsations: Ground-satellite obser- looning modes in magnetospheric plasmas, Geophys. Res.
vations, J. Geophys. Res., 88, 7029-7036, 1983.
Lett., 21(19), 2091-2094, 1994.
Slavin, J. A., E. J. Smith, B. T. Tsurutani, D. G. Sibeck, H. J. Vifias, A. F., and T. R. Madden, Shear flow-ballooning insta-
Singer, D. N. Baker, J. T. Gosling, E. W. Hones Jr., and bility as a possible mechanism for hydromagnetic fluctua-
F. L. Scarf, Substorm-associated traveling compression re- tions, J. Geophys. Res., 91, 1519-1528, 1986.
gions in the distant tail: ISEE 3 Geotail observations, Geo- Vorobjev, V. G., G. V. Starkov, and Y. I. Feldstein, The
phys. Res. Lett., 11(7), 657-660, 1984.
auroral oval during the substorm development, Planet.
Slavin, J. A., et al., CDAW 8 observations of plasmoid signa- Space Sci., 24, 955-965, 1976.
tures in the geomagnetic tail: An assessment, J. Geophys. Walker, R. J., K. N. Erikson, R. L. Swanson, and J. R. Winck-
Res., 94, 15,153-15,175, 1989.
ler, Substorm-associated particle boundary motion at syn-
Slavin, J. A., M. F. Smith, E. L. Mazur, D. N. Baker, E. W. chronous orbit, J. Geophys. Res., 81, 5541-5550, 1976.
Hones Jr., T. Iyemori, and E. W. Greenstadt, ISEE 3 Wallis, D. D., J. R. Burrows, M. C. Moshupi, C. D. Anger, and
observations of traveling compression regions in the J. S. Murphree, Observation of particles precipitating into

232 ß Elphinstone et al.: MODULAR AURORAL ACTIVITY 34, 2 / REVIEWS OF GEOPHYSICS
detached arcs and patches equatorward of the auroral oval,
J. Geophys. Res., 84, 1347-1360, 1979.
Walt, M., Institute summary, in Magnetospheric Physics, edited
Convective generation of "giant" undulations on the
evening diffuse auroral boundary, J. Geophys. Res., 99,
19,499-19,512, 1994.
by B. M. McCormac, pp. 3-19, D. Reidel, Norwell, Mass., Yau, A. W., B. A. Whalen, and D. J. McEwen, Rocket-borne
1974.
Wargentin, P., Vetenskapernas Historia, Om Nordskenet,
measurements of particle pulsation in pulsating aurora, J.
Geophys. Res., 86, 5673-5681, 1981.
Proc. R. Swed. Acad. Sci., 161-171, 1752.
Wilkinson, D.C., An archive of space weather effects on
spacecraft, supplement to Eos Trans. AGU, Fall Meet.
Suppl., 435, 1995.
Yeoman, T. K., M.P. Freeman, G. D. Reeves, M. Lester, and
D. Orr, A comparison of midlatitude Pi 2 pulsations and
geostationary orbit particle injections as substorm indicators,
J. Geophys. Res., 99, 4085-4093, 1994.
Yahnin, A. G., V. A. Sergeev, R. J. Pellinen, W. Baumjohann, Yumoto, K., K. Takahashi, T. Sakurai, P. R. Sutcliffe,
K. U. Kaila, H. Ranta, J. Kangas, and O. M. Raspopov, S. Kokubun, H. Luhr, T. Saito, M. Kuwashima, and N. Sato,
Substorm time sequence and microstructure on 11 November
1976, J. Geophys., 53, 182-197, 1983.
Multiple ground-based and satellite observations of global Pi 2
magnetic pulsations, J. Geophys. Res., 95, 15,175-15,184, 1990.
Yahnin, A., et al., Features of steady magnetospheric convec- Zelenyi, L. M., R. A. Kovrazkhin, and J. M. Bosqued, Velocitytion,
J. Geophys. Res., 99, 4039-4051, 1994.
dispersed ion beams in the nightside auroral zone: Aureol 3
Yamamoto, T., and T. Oguti, Recurrent fast motions of pulsating
auroral patches, 1, A case study on optical and
quantitative characteristics during a slightly active period, J.
observations, J. Geophys. Res., 95, 12,119-12,139, 1990.
Geophys. Res., 87, 7603-7614, 1982.
Yamamoto, T., K. Makita, and C.-I. Meng, A particle simulation
of the westward traveling surge, J. Geophys. Res., 98,
L. L. Cogger, R. D. Elphinstone, and J. S. Murphree, Department
of Physics and Astronomy, University of Calgary,
13,653-13,675, 1993.
2500 University Drive, NW, Calgary, Alberta T2N 1 N4, Can-
Yamamoto, T., M. Ozaki, S. Inoue, K. Makita, and C.-I. Meng, ada. (e-mail: rob@hobbes.phys.ucalgary.ca)