1

ABSTRACT
Coronal mass ejections (CMEs) are an important factor in coronal and interplanetary dynamics. They can inject large amounts of mass and magnetic fields into the heliosphere, causing major geomagnetic storms and interplanetary shocks, a key source of solar energetic particles. Recent studies using the excellent data sets from the SOHO, Yohkoh, TRACE, Wind, ACE and other spacecraft and ground-based instruments have improved our knowledge of the origins and early development of
CMEs at the Sun and how they affect space weather. I review some key coronal properties of CMEs, their source regions, their manifestations in the solar wind, and their geoeffectiveness. Halo CMEs are of special interest for space weather because they suggest the launch of a geoeffective disturbance toward Earth. However, their correspondence to geomagnetic storms varies over the solar cycle. Although CMEs are involved with the largest storms at all phases of the cycle, recurrent features such as interaction regions and high speed wind streams are also geoeffective.
INTRODUCTION
CMEs consist of large structures containing plasma and magnetic fields that are expelled from the Sun into the heliosphere. Most of the ejected material comes from the low corona, although cooler, denser material probably of chromospheric origin can also be ejected. Much of the plasma observed in a CME is entrained on expanding magnetic field lines, which can have the form of helical field lines with changing pitch angles, i.e., a flux rope.
This paper reviews the well-determined coronal properties of CMEs and what we know about their source regions, and some key signatures of CMEs in the solar wind. I emphasize the recent observations of halo CMEs, which appear as expanding, circular brightenings that completely surround the occulting disk of the SOHO
LASCO coronagraphs, and are important for space weather studies.
Most observations of CMEs have been made by white light coronagraphs, operating in space. These observations are based on the Thomson scattering process, wherein dense plasma structures in the corona become visible via photospheric light scattered off of the free electrons in the plasma. The LASCO coronagraphs
(Brueckner et al., 1995) provide the most recent white light observations on CMEs. These data have been acquired since early 1996, providing fairly continuous coverage of the corona and CMEs from the solar activity minimum through the maximum of cycle 23. The LASCO observations of CMEs are complemented by other SOHO instruments operating at coronal wavelengths, especially
EIT, UVCS and CDS, and the Yohkoh and TRACE spacecraft. Hudson and Cliver (2001) provide a recent summary of non-white light observations of CMEs. The geoeffective aspects of all these CME observations are the focus of this paper.
In the next section I review our understanding of how solar and geomagnetic activity varies over the solar cycle.
In section 3 I discuss the basic properties of CMEs and what we know about their source regions. In Section 4 I review the manifestations of CMEs in the heliosphere and near Earth, their relationship to recurrent solar wind structures, and what structures are most geoeffective. The last section summarizes the important points.
SOLAR AND GEOMAGNETIC ACTIVITY OVER
THE SOLAR CYCLE
It well known that the level of geomagnetic activity tends to follow the solar sunspot cycle. Sunspot counts are a useful measure of the general level of magnetic activity emerging through the photosphere of the Sun, and they rise and fall relatively uniformly over the cycle. Other major classes of solar activity also tend to track the sunspot number during the cycle, including active regions, flares, filaments and their eruptions, and CMEs
(e.g., Webb and Howard, 1994). This activity is transmitted to Earth through the solar corona and its expansion into the heliosphere as the solar wind.
The cyclical variation of the solar magnetic field can be summarized as follows. The sunspot number indicates the scale of the emergence through the solar surface of regions of the strongest flux. The photospheric flux is indicative of the strength of the toroidal flux which in a
Babcock-type dynamo model of solar activity is built up from the basic solar dipole field over the solar cycle. The total photospheric flux varies by about a factor of three over the cycle. A useful technique for estimating the interplanetary flux is to calculate the flux through the source surface above which the field is assumed to be radial (e.g., Luhmann et al., 1998). The source surface flux tracks the lower order, dipole contribution of the solar field over the cycle and is indicative of the variation of the total open flux from the Sun. Although the total interplanetary magnetic field (IMF) flux at 1 AU is stronger at maximum and weaker near minimum than the total open flux (e.g., Wang et al., 2000), it is still strong during the declining phase when it is dominated by the open flux in high speed wind streams from coronal holes.
2

Since solar activity is transmitted to Earth via the solar wind, how closely does the cycle of geoactivity follow that of solar activity? The overall correspondence between the sunspot and geoactivity cycles can be seen in long-term plots comparing various indices of the two kinds of activity. On annual time scales the geoactivity cycle has more structure than the solar cycle. Figure 1 shows the annual number of disturbed days with the geomagnetic index Ap > 50 vs sunspot number over the last 6 solar cycles. Ap is more variable than sunspot number, but does tend to track the sunspot cycles in amplitude.
Figure 1 also illustrates the double-peaked nature of the geoactivity cycle. In general, geomagnetic activity exhibits a peak near sunspot maximum and another during the declining phase of the cycle. These peaks vary in amplitude and timing, and the peak around maximum may itself consist of two peaks (Richardson et al., 2000).
The two main peaks are usually considered to represent the maximum phases of two components of geoactivity that have different solar and heliospheric sources. The first peak is associated with transient solar activity, i.e.,
CMEs, that tracks the solar cycle in amplitude and phase.
The later peak is attributed to recurrent high speed streams from coronal holes, and is often higher than the early peak.
Richardson et al. (2000; 2001) recently studied the relative contributions of different types of solar wind structures to the aa index from 1972-1986. They identified CME-related flows, corotating high-speed streams, and slow flows near the Earth, finding that each type contributed significantly to aa at all phases of the cycle. For example, CMEs contribute ~50% of aa at solar maximum and ~10% outside of maximum, and high speed streams contribute ~70% outside of maximum and
~30% at maximum. Thus, both types of sources, CMEs and coronal holes/high speed streams, contribute to geoactivity all phases of the cycle.
CMEs, however, are responsible for the most geoeffective solar wind disturbances and, therefore, the largest storms.
Enhanced solar wind speeds and southward magnetic fields associated with interplanetary shocks and ejecta are known to be important causes of storms (e.g., Tsurutani et al., 1988; Gosling et al., 1991). The reason that any
CME/magnetic cloud encountering Earth is likely to be geoeffective is that enhanced speeds and, particularly, sustained southward IMF will occur within or ahead of
Figure 1: Annual number of geomagnetic disturbed days with
Ap index > 50, (dashed line and hatched area) vs. annual sunspot number (solid line) for solar cycles 17-23. Courtesy
NOAA National Geophysical Data Center, Boulder, CO.
3 most CMEs traveling within the heliosphere. Strong southward fields often occur either in magnetic clouds or in the preceding post-shock regions, or both. Slower
CMEs not driving shocks are probably associated with many smaller storms. Zhao et al. (1993) found that 78% of all periods with southward IMF

-10 nT for durations

3 hr were associated with one or more CME signatures.
Compression and draping of fields in the preceding ambient solar wind can also enhance southward IMF
(Gosling and McComas, 1987).
SOLAR CHARACTERISTICS OF CMES
Properties of CMEs
The measured properties of CMEs include their occurrence rates, locations relative to the solar disk, angular widths and speeds (e.g., Kahler, 1992; Webb,
2000; St. Cyr et al., 2000). There is a large range in the basic properties of CMEs. Their speeds, accelerations, masses and energies extend over 2-3 orders of magnitude, and their widths exceed by factors of 3-10 the sizes of flaring active regions.
CMEs can exhibit a variety of forms, some having the classical “three-part” structure and others being more complex with interiors with bright emitting material. The basic structure of the former kind consists of a bright leading arc followed by a darker, low-density cavity and a bright core of denser material (Figure 2). These may represent pre-event structures which erupt: a prominence and its overlying coronal cavity, and the ambient corona
(e.g., streamer) which is compressed as the system rises.
Partly because of their increased sensitivity, field of view and dynamic range, the SOHO LASCO C2 and C3 coronagraphs have observed many different forms of
CMEs, including those with large circular regions resembling flux ropes and halo CMEs.
Figure 2
: Evolution of a “3-part” CME on 2 June 1998. The three features are: 1) bright curved leading edge, followed by 2) darker region, and 3) bright, interior structure, here a prominence. Note circular structures just above the prominence, suggesting a flux rope. From Plunkett et al. (2000).
Halo CMEs appear as expanding, circular brightenings

that completely surround the coronagraph occulting disk
(Howard et al., 1982). Their observation suggests that consecutive C2 images.
these CMEs are moving outward either toward or away from the Earth and are detected after expanding to a size exceeding the diameter of the coronagraph's occulter
(Figure 3). Other observations of associated activity on the solar disk are necessary to distinguish whether a halo
CME was launched from the frontside or backside of the
Sun. Other CMEs which have a larger apparent angular size than typical limb CMEs, but do not appear as complete halos, are called `partial halo' CMEs.
Halo CMEs are important to study for three reasons: 1)
They are known to be the key link between solar eruptions and many space weather phenomena such as major storms and solar energetic particle events; 2) The source regions of frontside halo CMEs are usually located within a few tens of degrees of Sun center, as viewed from Earth (Webb et al., 2001a; Cane et al., 2000). Thus, the source regions of halo events can be studied in greater detail than for most CMEs which are observed near the limb: 3) Frontside halo CMEs must travel approximately along the Sun-Earth line, so their internal material can be sampled in situ by spacecraft near the Earth. Three spacecraft, SOHO, Wind and ACE, now provide solar wind measurements upstream of Earth.
The frequency of occurrence of CMEs observed in white light tends to track the solar cycle in both phase and amplitude, which varies by an order of magnitude over the cycle (Webb and Howard, 1994). LASCO has now
Figure 3 : Illustrations of 2 kinds of halo CMEs: (a)
Symmetrical, gradual CME forming a complete ring around the
C2 occulter on 12 May 1997. Associated with a C1 flare, filament eruption and EUV wave; (b) Asymmetrical, impulsive halo CME on 17 February 2000. The CME was fast and associated with an M1/2N flare. These are difference images of
4 observed from solar minimum in early 1996 through the rise phase and maximum of the current (23rd) solar cycle.
It has detected CMEs at a rate slightly higher than the earlier observations, reaching >4/day at maximum (St Cyr et al., 2000; C. St. Cyr., priv. comm., 2001). Halo CMEs occurred during the rise phase of this cycle at a rate of 1-
6/month, about 10% the rate of all CMEs. Full halo
CMEs were only detected at a rate of ~4% of all CMEs. If
CMEs occurred randomly at all longitudes and LASCO detected all of them, this rate would be about 15%. This suggests that LASCO sees as halos CMEs that are brighter (i.e., denser) than average.
The latitude distribution of the central position angles of
CMEs tends to cluster about the equator at minimum but broaden over all latitudes near solar maximum. This remains true in the LASCO data. Hundhausen (1993) noted that this CME latitude variation more closely parallels that of streamers and prominences than of active regions, flares or sunspots. On the contrary, the angular size distribution of CMEs varies little over the cycle, maintaining an average width of about 45 o (Hundhausen,
1993; Howard et al., 1985). The CME size distribution observed by LASCO is affected by its increased detection of very wide CMEs, especially halos (St. Cyr et al.,
2000). However, although the average width of LASCO
CMEs is 72 o , the median of 50 o is similar to that of the earlier measurements.
Estimates of the apparent speeds of the leading edges of
CMEs range from about 20 to over 2000 km/s. Thus, these speeds range from well below the sound speed in the lower corona to well above the Alfven speed. The annual average speeds of SOLWIND and SMM CMEs varied over the solar cycle from about 150-475 km/s, but their relationship to sunspot number was unclear (Howard et al., 1986; Hundhausen et al., 1994a). The LASCO
CME speed distributions are similar in range to the prior measurements, but do show a tendency to increase with sunspot number in this cycle (St. Cyr et al., 2000; N.
Gopalswamy, priv. comm., 2001). The average CME speeds have remained at their highest from 1999-2001.
The annual average speed of full halo CMEs is 1.5-2 times greater than that of all CMEs, suggesting that

LASCO sees as halos CMEs which are faster and, hence, more energetic than the average CME.
MacQueen and Fisher (1983) first described two kinematical classes of CMEs based on their associated surface events: slow, gradually accelerating CMEs associated with prominence eruptions, and fast CMEs with constant speeds associated with flares. Above a height of about 2R
S
the speeds of typical CMEs are relatively constant. Despite its increased field of view, only 17% of all LASCO CMEs exhibit acceleration out to
30 R
S
(St. Cyr et al., 2000). Clearly the acceleration for most CMEs occurs low in the corona (St. Cyr et al.,
1999), especially the fastest ones (Zhang et al., 2001).
Using LASCO data, Sheeley et al. (1999) and Srivastava et al. (1999) find that gradually accelerating CMEs are balloon-like with central cores which accelerate more slowly than the leading edge, whereas fast CMEs move at constant speed even as far out as 30R
S
. However, when viewed well out of the skyplane, gradual CMEs look like smooth halos which accelerate to a limiting value then fade away, while fast CMEs have ragged structure and decelerate (Sheeley et al., 1999). Examples of these classes of halo events are shown in Figure 3.
Finally, the masses and energies of CMEs require difficult instrument calibrations and have large uncertainties. The average excess masses of CMEs derived from the older coronagraph data were a few x
10 15 g (Howard et al., 1985; Hundhausen et al., 1994b).
However, more recent studies using Helios (Webb et al.,
1996) and SOHO LASCO (Howard et al., 1997;
Vourilidas et al., 2000) data indicate that CME masses have been underestimated, probably because mass outflow can continue well after the CME's leading edge leaves the instrument field of view. Average CME kinetic energies from the older data were a few x 10 31 erg, likewise underestimated.
Solar Source Regions of CMEs
Here I briefly summarize our knowledge of the nearsurface features which appear to be involved in the initiation of CMEs. See recent reviews for more details, including Hundhausen (1999), Forbes (2000), Webb
(2000) and Cliver and Hudson (2002). Many CMEs appear to arise from large-scale, closed structures, particularly preexisting coronal streamers (e.g.,
Hundhausen, 1993). Low and Hundhausen (1995) emphasized that the prominence/streamer involves a dual flux system, one part of which is a flux rope with its associated coronal cavity, which may help to drive the eruption once the streamer is disrupted. Many energetic
CMEs actually involve the disruption (“blowout”) of a preexisting streamer, which can increase in brightness and size for days before erupting as a CME (Figure 1; Howard et al., 1985; Hundhausen, 1993; Subramanian et al.,
1999). The streamer usually disappears afterwards, but may eventually reform.
5
Previous statistical association studies indicated that erupting filaments and X-ray events, especially of long duration, were the most common near-surface activity associated with CMEs (e.g., Webb, 1992). Most optical flares occur independently of CMEs and even those accompanying CMEs may be a consequence rather than a cause of CMEs (Gosling, 1993; but see Hudson et al.,
1995). The fastest, most energetic CMEs, however, are usually also associated with surface flares, and my studies show that reported flares are associated with most (~85%) frontside, full halo CMEs. This rate may be higher than reported previously because the sources of halo CMEs can be clearly viewed near Sun center.
Comparisons of soft X-ray data with the white light observations have provided many insights into the source regions of CMEs. Sheeley et al. (1983) showed that the probability of associating a CME with a soft X-ray flare increased linearly with the flare duration, reaching 100% for flare events of duration >6 hours. The SMM CME observations indicated that the estimated departure time of flare-associated CMEs typically preceded the flare onsets. Harrison (1986) found that such CMEs were initiated during weaker soft X-ray bursts that preceded any subsequent main flare by tens of minutes, and that the main flares were often offset to one side of the CME (c.f.,
Webb, 1992). This offset is also shown by Plunkett et al.
(2002) for LASCO CMEs. The latitude distribution of the
CMEs peaks at the equator, but the distribution of EUV activity observed by the EIT associated with these CMEs is bimodal with peaks 30 o north and south of the equator.
This offset is confirmed for the distribution of sources associated with halo CMEs (Figure 4; Webb et al.,
2001a).
This pattern indicates that many CMEs can involve more complex, multiple polarity systems (Webb et al., 1997) such as modeled by Antiochos et al. (1999). This multipolar pattern was evident in the LASCO observations during the recent solar minimum in 1996-
1997, in which the simple, bipolar streamer belt observed in the outer coronagraphs overlay twin arcades flanking the equator as observed by the inner coronagraph
(Schwenn et al., 1997). The eruption of such a streamer as a CME would imply that it originated in a quadrupolar magnetic region.
The most obvious X-ray signatures of CMEs in the low corona are the arcades of bright loops which develop after the CME material has apparently left the surface (Kahler,
1977-Skylab; Hudson and Webb, 1997-Yohkoh). Prior to the eruption, an S-shaped structure, called a sigmoid by
Rust and Kumar (1996) often develops in association with a filament activation. A sigmoid is indicative of a highly sheared, non-potential coronal magnetic field. The sigmoid structure denotes the dominant helicity in a given
Figure 4: Histogram of the great circle distances between the

surface source activity (S) associated with each full halo CME and the sub-Earth point (C). The average of this distribution is
35 o . The average is 28 o when limb sources are excluded. hemisphere, being S-shaped or right-handed in the south and reverse-S or left-handed in the north. This type of structure might be an important precursor of a CME (e.g.,
Canfield et al. 1999), although its geoeffectiveness has
Distance between Sub-Earth Point and CME Source
14
12
10
8
6
C
X
S
4
2
0
0 20 40 60 80
Great Circle Distance (deg.)
100 120 not been established. Eventually an eruptive flare can occur, resulting in the bright, long-duration arcade of loops. Sterling et al. (2000) call this process “sigmoid-toarcade” evolution. These arcades suggest the eruption and subsequent reconnection of strong magnetic field lines associated with the CME system (c.f., Svestka and Cliver,
1992).
Dimming regions observed in X-rays and in the EUV imply that material is evacuated from the low corona (c.f.,
Hudson and Webb, 1997). The dimming regions can be much more extensive than the flaring activity and often map out the apparent base of the white light CME
(Thompson et al., 2000). Thus, the dimming events appear to be one of the earliest and best-defined signatures in soft X-ray and EUV emission of the actual mass ejected from the low corona. Gopalswamy et al.
(1999; 2000) has reported two observations of large-scale coronal enhancement s, not dimmings, that could be the near-surface material that appears later in the white light
CME.
Surveys of solar activity associated with frontside halo
CMEs have been made primarily with images of the low corona from the SOHO EIT and Yohkoh SXT instruments. Hudson et al. (1998) and Webb et al. (2000a) examined halo CMEs observed during the first half of
1997, finding that about half of the halo events were associated with frontside activity as expected. Studies of
CME source regions using EIT and LASCO data have been reported by Thompson et al. (1999a,b; 2000) and
Plunkett et al. (2000). The activity associated with halo
CMEs includes the formation of dimming regions, of long-lived loop arcades, flaring active regions, large-scale
6 coronal waves that propagate outward from the CME source region, and filament eruptions. Optical flares and filament eruptions can also be identified in H

and radio images. Specifically, Webb et al. (2001a) find that 2/3 of the halo CMEs are associated with filament eruptions and with dimmings. Metric radio bursts are also often associated with the CMEs.
The frequent detection of EUV waves is probably the most exciting new aspect of the SOHO EIT observations.
The waves are most clearly seen in movies of differences of consecutive pairs of full-disk 195

images, and appear as expanding wave fronts moving outward from flaring active regions. The EIT has now observed many waves in association with CMEs, but only about half of the halo
CMEs are associated with EIT waves (Webb et al.,
2001a). Thompson et al. (1999) interpreted them as fast mode MHD waves, similar to the Moreton waves observed in chromospheric lines.
CMES IN THE HELIOSPHERE AND THEIR
GEOEFFECTIVENESS
Solar Wind Signatures of CMEs
CMEs carry into the heliosphere large amounts of coronal magnetic fields and plasma, which can be detected by remote sensing and in-situ spacecraft observations. The passage of this material past a single spacecraft is marked by certain distinctive signatures, but with a great degree of variation from event to event (e.g., Gosling, 1993).
These signatures include transient interplanetary shocks, depressed proton temperatures, cosmic ray depressions, flows with enhanced helium abundances, unusual compositions of ions and elements, and magnetic field structures consistent with looplike topologies.
A widely used single-parameter signature of ejecta is the occurrence of counterstreaming suprathermal electrons.
Since suprathermal electrons carry electron heat flux away from the Sun along magnetic field lines, when found streaming in both directions along the field they are interpreted as signatures of closed field lines and, thus, as a good proxy for CMEs in the solar wind (e.g., Gosling,
1993). An important multiple signature of an interplanetary (I)CME, is a magnetic cloud (e.g., Burlaga,
1991), defined as several-hour flows with large-scale rotations of unusually strong magnetic field accompanied by low ion temperatures. The magnetic field data from clouds often provide good fits to flux rope models, e,g,,
Marubashi (1986, 1997); Lepping et al. (1990); Russell
(2001).
Another class of ejecta plasma signatures are the abundances and charge state compositions of elements and ions which are systematically different in ICMEs compared with other kinds of solar wind (Galvin, 1997).
As the corona expands outward, the electron density decreases so rapidly that the plasma becomes collisionless

and the relative ionization states become constant, thus reflecting the conditions of origin in the corona. The charge states of minor ions (Z>2) in CME flows usually suggest slightly hotter than normal coronal conditions
(i.e., >2 MK) at this “freezing in” location. In addition, transient flows often exhibit element and ion abundances that are enhanced relative to the typical solar wind.
Unusually low ionization states of He and minor ions have also been detected in CME flows. Although rarely observed before, enhanced He + flows have been detected in the flows from several recent halo CMEs with more sensitive instruments on the SOHO, Wind and ACE spacecraft. In each of these events an erupting filamenthalo CME could be associated with either a dense and compact `plug' or an extended flow of cool plasma in the trailing edge of a magnetic cloud. This is likely material from the filament itself, consistent with near-Sun observations showing that erupting filaments lag well behind the leading edge of their associated CMEs. In a recent study of Wind magnetic clouds, Lepping and
Berdichevsky (2000) find that half show a significant increase in density toward the rear of the cloud. In a preliminary study from 1995-2000, I have found evidence of either dense or cool material within ~60 ICMEs, most associated with halo CMEs.
The Geoeffectiveness of CMEs
Since the launch of SOHO, halo CMEs are being used to study the influence of Earth-directed CMEs on geoactivity. Analyses of the relation between halo CMEs and geomagnetic storms have been carried out by
Brueckner et al. (1998), Webb et al. (2000), Cane et al.
(2000) and St. Cyr et al. (2000) and indicate a high degree of correlation near solar minimum and a decreased association near the current maximum. Webb et al. (2000) studied the geoeffectiveness of halo CMEs in early 1997 by comparing the onset times of the frontside halo events with storms at Earth 2-5 days later. Defining storms as having a peak depression in Dst

-50nT, they found that all 6 frontside halos with surface sources within 0.5 R
S
of
Sun center were associated with magnetic clouds, or cloud-like structures at 1 AU, and with moderate-level storms. Brueckner et al. (1998) also found a strong relationship between storms and LASCO CMEs from
1996 to mid-1997, and Cane et al. (1998; 1999) found good associations between frontside halo CMEs and ejecta signatures at 1 AU. St. Cyr et al. (2000) found similar, but weaker associations between halo CMEs and storms. They studied LASCO halo CMEs from 1996-June
1998, and concluded that 83% (15 of 18) of intense storms, were preceded by frontside halo CMEs. However,
25 of the frontside halo CMEs did not produce such large storms and were, therefore, false alarms.
All these studies included partial halo CMEs. LASCO has now observed a sufficient number of CMEs to permit statistical studies of only full (360 o ) halo events, those most likely to be directed along the Sun-Earth line. In a
7 study of 89 frontside full halos observed from 1996-2000,
Webb et al., (2001a) found that about 70% of the halos were associated with shocks and/or counterstreaming electrons or other ejecta signatures at 1AU. Magnetic clouds or cloud-like structures were involved with ~60% of the halos, although this rate rises to ~70% if the peculiar year 1999 is excluded. Somewhat surprisingly, when averaged over this entire period, only half of the frontside halo CMEs could be well associated with moderate-level or greater storms. The average travel time from the onsets of the halo CMEs to the onsets of the storms at Earth was 3.3 days.
Table 1 shows how the association degree between halo
CMEs and storms, the associated storm peaks, and the travel times vary over the solar cycle from activity minimum to maximum. Also shown is whether the parameter increased or decreased and by what factor. We see that the average CME rate and speeds increased with the cycle and the travel time decreased, as expected for more energetic events. However, although the storm intensities steadily increased, the degree of association between the halo CMEs and storms decreased monotonically to yield the overall 50% average.
TABLE 1 Variation of Halo CMEs with Storms
Halos with Storms (%)
1997 1998 1999 2000
92 54 39 35
Assoc. Storm Peak (-Dst) 80
Travel Time (days) 3.7
96
3.45
Solar Cycle Variation (Minimum to Maximum)
Halo CME Rate
Halo CME Speeds
Halos & Storms
Assoc. Storm Intensity
Travel Time
Inc.
Inc.
Dec.
Inc.
Dec.
129 149
3.2 x10 x1.5 x3 x2 x1.2
3.0
However, confirming previous studies, I find that most
(~70%; 15 of 21) of the most intense storms ( Dst

-
150nT) of this cycle were associated with one or more full and/or partial halo CMEs. (The other 6 had possible associations.) For an intense magnetic storm to occur, there must be a sustained period of strong southward IMF to provide an efficient transfer of energy and momentum from the solar wind to the magnetosphere. For example, in the relatively simple 1997 events on January 6-10 and
May 12-15, the magnetosphere's response to the magnetic cloud in each event closely followed the southward field.
Dst rose in response to the shock compression, then decreased in proportional response to the southward IMF in both the compressed region before the cloud and in the front part of the cloud. Thus, a key reason why the halo
CME studies have found that the observation of a frontside halo CME is not a sufficient condition for a

large storm may be that not all CMEs contain or create sufficiently strong, sustained southward field.
The January 1997 event also is a nice example of the geoeffectiveness of a halo CME having only weak observable solar surface manifestations (e.g., Webb et al.,
1998). Despite its occurrence near the winter solstice (see next section), the CME caused a large magnetic storm leading to the demise of the Telstar 4 satellite. Without observation of the halo CME, the appearance of the magnetic cloud and storm at Earth would not have been predicted, and the storm would have become another
“problem” storm, one with no identifiable solar source.
Thus, we can conclude that halo CMEs with associated surface features, especially if near Sun center, usually presage at least moderate geomagnetic storms, but that at least half as many such storms occur without halo-CME forewarning. Therefore, the simple observation of the occurrence of a probable frontside halolike CME has already significantly increased our ability to forecast the occurrence of storms of moderate or greater levels at
Earth. Moderate storms not associated with CMEs are usually caused by Earth passage through the heliospheric current sheet (HCS) and related corotating interaction regions (CIRs).
CMES IN THE CONTEXT OF THE GLOBAL SOLAR
AND HELIOSPHERIC MAGNETIC FIELD
We now have a new understanding of how the magnetic topology of the Sun, CMEs, and the heliosphere are interrelated (see Crooker, 2000, and Webb et al., 2001b, for recent discussions). Globally the slow solar wind is confined to the vicinity of the heliospheric extension of the streamer belt, which is the locus of the heliospheric current sheet at the Sun. This belt now appears to be the source of most CMEs. Thus, CMEs tend to carry with them the imprint of the lower harmonics of the Sun's magnetic field and their magnetic structure tends to merge with that of the heliosphere. This paradigm has important implications for understanding the geoeffectiveness of
CMEs.
Figure 5 is a simplified schematic view illustrating how the solar magnetic topology fits into the larger heliospheric context. This view is for the simplest solar magnetic configuration at solar minimum, when the Sun's magnetic field can be approximated as a dipole whose axis is tilted slightly to the axis of rotation. At other
Figure 5: Schematic diagram illustrating the imprint of the solar magnetic field on two CME flux ropes forming under the helmet streamer belt, one in the northern hemisphere (N) with left-handed (L.H.) chirality, and one in the southern hemisphere
(S.), with right-handed (R.H.) chirality. After Crooker (2000). phases of the cycle, the dipole tilt increases and higher harmonics distort the field, but the basic arguments here
8
N
S probably still apply.
Separating the opposite-polarity, polar coronal holes in
Figure 5 is a wide band forming the base of the streamer belt. Since CMEs can arise from and disrupt streamers, they are intimately tied to the streamer belt. The heavy curve is the projection of the HCS onto the solar surface as the heliomagnetic equator. Dipolar field lines (the N–S curved arrows) form an arcade from whose apex arises the current sheet. The high speed wind flowing from the polar holes constricts the oppositely-directed fields to a narrow current sheet which appears like a ballerina's skirt.
The rotating, warped HCS appears as a sector boundary crossing at Earth; during one solar rotation Earth will be immersed in two large sectors of opposite polarity, each with the relatively high speed wind of its parent hole.
The short straight arrows mark the locations of two flux ropes under the helmet streamer belt associated with
CMEs. The arrows represent the direction and orientation of the projected axes of the flux ropes, roughly parallel to the heliomagnetic equator. The directions of the two arrows determine the chirality (sense of twist or helicity) of the resulting flux ropes, which tend to be left-handed,
‘L.H.’, (right-handed, ‘R.H.’) in the northern, ‘N.’,
(southern, ‘S.’) hemisphere. These match a hemispherically-dependent magnetic field pattern associated with filaments low in the corona (Martin et al., 1994).
Thus, Figure 5 shows how a CME and its flux rope follows the Sun's magnetic topology.
Marubashi (1986, 1997)), Rust (1994), Bothmer and Rust
(1997) and others associated sets of erupting filaments with magnetic clouds. For the clouds that were fit by a cylindrical flux rope model, the orientations of the rope axes correlated with those of the filaments, and the cloud helicities matched those predicted from the filament chiralities. Zhao and Hoeksema (1997, 1998) found correlations between the orientations of cloud axes fit with flux ropes, the orientations of their filaments, and the duration and maximum intensity of southward IMF in those clouds.
In our paradigm, the direction of the leading field in a
CME should have the same orientation as the solar dipole

field. Since the dipolar field changes polarity at every solar maximum, the direction of the leading field in
CMEs and clouds should also change at maximum. This was confirmed by Bothmer and Rust (1997), Bothmer and
Schwenn (1998) and Mulligan et al. (1998). Mulligan et al. also found a solar cycle variation in the orientation of cloud axes, with low inclinations dominating during solar minimum and the ascending phase and high inclinations during maximum and the declining phase. Thus, in our example in Figure 5 for solar minimum, the flux rope/CMEs would have leading southward and trailing northward fields, whereas at maximum a flux rope with a northward axis parallel to a highly inclined heliomagnetic equator would have little southward field. These relationships provide a predictive capability for the strength and duration of southward IMF in a magnetic cloud based on observations of the associated filament pattern.
High density can also impact the strength of a storm. For southward IMF, the magnetosphere responds to a change in solar wind density by an increase in plasma sheet density. Thomsen et al. (1998) noted that the January
1997 cloud would have been much more geoeffective had the extremely high density at its trailing edge occurred when the IMF was southward. In the solar wind an ICME can be associated both with high-density features of solar origin in the slow flow itself (i.e., filament material) and of interplanetary origin, at compression regions ahead of the CME and sometimes trailing the CME if being overtaken by a CIR.
So the interaction of a CME with the surrounding heliospheric patterns can affect its geoeffectiveness over the cycle. During the declining phase of the solar cycle. the 27-day recurrence pattern imposed by the corotation of coronal holes and their high-speed solar wind streams dominates the solar wind. Then, the peak strength of storms coincides with passage of the leading edge of a high-speed stream, not with the subsequent high-speed flows. This is due to compression of pre-existing southward IMF at the leading edge of CIRs and their associated sector boundaries.(e.g., Crooker and Cliver,
1994; Tsurutani et al., 1995).
CIRs separate regions of high and low speed flows, identified with coronal holes and streamers, respectively, at the Sun. Thus, the solar sources of the recurrent activity include the boundaries between coronal holes and streamers. Crooker and Cliver (1994) showed that the largest, usually recurrent storms associated with highspeed streams are often due to CMEs associated with the
CIRs. Since CMEs arise from closed fields in the streamer belt and the HCS base, they are part of the slow flows which border fast flows from coronal holes. Since the streamer belt and HCS is tilted with respect to the ecliptic, spacecraft there, and the Earth, will observe a
CME in the slow flow immediately preceding the leading
9 edge of a high-speed stream. Since the HCS acts as a channel for CMEs which can effectively carry the sector boundary (Crooker et al., 1998), the probability of encountering a CME in the ecliptic plane increases near sector boundaries (Crooker et al., 1993).
The fast flow of a high-speed stream pushes into the slow flow ahead and compresses it, creating the CIR. ICMEs in the slow flow also can contain southward fields which can be compressed in the CIR. The CIR immediately follows the ICME so that southward IMF both in the ICME and in the CIR contribute to storm strength. Fenrich and
Luhmann (1998) noted that the geoeffectiveness of CIR compression of the trailing part of an ICME will have a solar cycle variation; north-leading magnetic clouds that are favorable for this effect occur between those solar maxima, such as cycle 21 to 22, when the Sun's dipolar field is opposite that shown in Figure 5.
The recent solar and heliospheric spacecraft observations of CMEs help to clarify why individual CMEs seem less well associated with storms near solar maximum. The occurrence rate of CMEs increases over minimum by an order of magnitude, leading to multiple CMEs per day over the Sun or even from a single region. (These can confuse halo CME identifications.) Gopalswamy et al.
(2001; 2002) have found since early 1998 several tens of
LASCO CMEs wherein a faster CME overtakes a slower one within 30 R
S
of the Sun, producing an interaction.
Reconnection or sandwiching of each CMEs’s field lines are likely in such cases. The combination of sequential eruptions of CMEs and their subsequent interactions can produce complex ejecta at 1 AU. Such ejecta often consist of high speed flows with shocks and other ICME signatures, but poorly defined magnetic structures with what Burlaga et al. (2001) call ‘tangled fields’.
In addition, the rate at which CMEs actually encounter
Earth near maximum is modified by their broadening latitude distribution. Thus, although the CME rate is considerably higher at maximum, proportionally fewer
CMEs are ejected near the ecliptic because of the highly tilted streamer belt. Because the flux ropes associated with this tilted belt tend to be north-south oriented, the individual CMEs that do reach the ecliptic will contain little southward field. Finally, the “background” solar wind into which the CMEs are injected is itself much more complex near maximum. This creates more frequent and complicated interactions of ejecta with the existing structure leading to distortions and compressions which are difficult to simulate and predict (e.g., Odstrcil and
Pizzo, 1999).
The Seasonal Variation
Although not directly related to solar activity, another periodic variation in geoactivity has important consequences for geoeffectiveness. This is the semiannual or seasonal variation characterized by a higher average

level of geoactivity at the equinoxes than at the solstices.
This effect is quite apparent in statistical analyses using common indices of geoactivity, including Dst (Cliver and
Crooker, 1993; Cliver et al., 2000). Surprisingly a pronounced seasonal effect is also seen in the occurrences of intense storms (Crooker et al., 1992).
Historically, three hypotheses have been put forth to explain the seasonal variation. These are the Russell-
McPherron (1973) effect, the equinoctial hypothesis, and the axial hypothesis (see Cliver et al., 2000). The Russell-
McPherron (RM) mechanism is a projection effect which depends upon season. The IMF, ordered in the Sun's heliographic coordinate system, projects a southward component in Earth's dipole-ordered coordinate system when it points toward the Sun during spring and away from the Sun during fall. In the equinoctial hypothesis geoactivity is maximum when the angle between the Sun-
Earth line (i.e., the solar wind flow direction) and Earth's dipole axis is 90 o , and less at other times. This effect works by reducing the coupling efficiency of the magnetosphere at the solstices. Finally, in the axial hypothesis, geoactivity peaks when Earth is at its highest heliographic latitudes, or best aligned with any radially propagating transient ICME flows and recurrent highspeed streams from mid-latitude coronal holes.
Although all these mechanisms may play some role in the seasonal variation, the RM effect has generally been the accepted mechanism because of its apparent success in explaining geoactivity data sets (e.g., Crooker and Cliver,
1994). However, Cliver et al. (2000) find that the equinoctial hypothesis better accounts for most of the semi-annual modulation, ~65%,whereas the other two mechanisms provide only 15-20% each.
CONCLUSIONS
The distribution of geomagnetic disturbances over the solar activity cycle has traditionally been considered to have two peaks representing two major components of geoactivity with different solar and heliospheric sources: one associated with transient solar activity that peaks with the sunspot cycle, and the other associated with recurrent
CIRs, CMEs and high speed streams from coronal holes during the declining phase. In terms of space weather,
CMEs are the most important kind of transient activity because they link the activity at the Sun and its propagation through the heliosphere to the Earth.
However, we now know that both CMEs and CIRs-high speed stream ensembles can be geoeffective at all phases of the cycle. As we have shown, the coronal streamer belt and its extension as the heliospheric current sheet appears to play a major role in both kinds of geoactivity.
What makes solar and heliospheric disturbances geoeffective, in the sense that they cause storms, can be summarized as southward IMF and compression .
10
Southward IMF is important because it allows merging of the IMF and Earth's magnetic field, which transfers solar wind energy and mass into the magnetosphere.
Compression is important because it strengthens existing southward IMF and, to a lesser extent, increases density.
CMEs, the most geoeffective structures, usually contain long duration flows of southward IMF and fast CMEs compress any southward IMF at their leading edges and behind shocks created by the speed difference. In addition, CMEs themselves can carry high-density structures, such as solar filaments. High-speed streams are geoeffective if they compress any southward IMF in
CIRs. This compression can be enhanced when CIRs interact with CMEs erupting through the HCS.
The recent observations of CMEs provide new insight into the problem of why halo CMEs seem less well associated with storms around solar maximum. Although the CME rate is much higher at maximum, because of the highly inclined HCS there are fewer injected into the ecliptic and their inclined flux ropes tend to contain little southward field. CMEs near maximum do tend to be more energetic and faster with shorter travel times in the heliosphere, and to be associated with more intense storms. However, much of this increased geoactivity is due not just to single (halo) CMEs, but also to complex ejecta caused by sequential and/or interacting CMEs, and their interaction with more complicated “background” solar wind structures. Future analyses of the rich data sets now in hand will improve our understanding of the role
CMEs play in space weather throughout the solar cycle.
Finally, recent studies have demonstrated that CMEs carry the imprint of the solar magnetic field out into the heliosphere. These findings imply that many characteristics of ejecta from CMEs can be predicted, at least statistically, from characteristics of the solar field and of associated solar source features, such as filament orientation and chirality and HCS orientation.
ACKNOWLEDGMENT
I thank the Organizing Committee of the SOHO 11
Symposium for inviting my presentation at the meeting. I benefitted from data from the SOHO mission, which is an international collaboration between NASA and ESA, and also from the SOHO/ LASCO CME catalog, generated and maintained by the Center for Solar Physics and Space
Weather, The Catholic University of America in cooperation with the Naval Research Laboratory and
NASA. I am grateful to C. St. Cyr and N. Gopalswamy for use of their preliminary results, and to E. Cliver for helpful comments. This work was supported at Boston
College by Air Force contract AF19628-00-K-0073,
AFOSR grant AF49620-98-1-0062, and NASA grant
NAG5-10833.
REFERENCES

Antiochos S.K., C.R. DeVore and J.A. Klimchuk, Astrophys. J.,
510, 485-493, 1999.
Bothmer, V., and R. Schwenn, Ann. Geophys., 16, 1, 1998.
Bothmer, V. and D. Rust, in Coronal Mass Ejections, N.
Crooker et al. (Eds.) GM 99 Washington, D.C., AGU, 139,
1997.
Brueckner, G.E. et al., Solar Phys., 162, 357-402, 1995.
Brueckner, G.E. et al., Geophys. Res. Lett., 25, 3019, 1998.
Burlaga, L.F., in Physics of the Inner Heliosphere, Vol. 2, eds.
R. Schwenn and E. Marsch, p. 1, Springer-Verlag, New York,
1991
Burlaga L.F., R.M. Skoug, C.W. Smith, D.F. Webb, T.H.
Zurbuchen and A. Reinard, J. Geophys. Res., 106, 20,957-
20,977, 2001.
Cane, H.V., I.G. Richardson, and O.C. St. Cyr, Geophys. Res.
Lett., 25, 2517, 1998.
Cane, H.V., I.G. Richardson, and O.C. St. Cyr, Geophys. Res.
Lett., 26, 2149, 1999.
Cane H.V., I.G. Richardson and O.C. St. Cyr, Geophys. Res.
Lett., 27, 3591-3594, 2000.
Canfield, R.F., H.S. Hudson, and D.E. McKenzie, Geophys.
Res. Lett., 26, 627, 1999.
Cliver, E.W., and N.U. Crooker, Sol. Phys., 145, 347, 1993.
Cliver, E.W., Y. Kamide, and A.G. Ling, J. Geophys. Res., 105,
2413, 2000.
Cliver, E.W. and H.S. Hudson, J. Atmos. Sol. Terr. Phys., 64,
231-252, 2002.
Crooker, N. U., J. Atmos. Sol. Terr. Phys., 62, 1071, 2000.
Crooker, N.U. and E.W. Cliver, J. Geophys. Res., 99, 23,383-
23,390, 1994.
Crooker, N.U., E.W. Cliver, and B.T. Tsurutani, Geophys. Res.
Lett., 19, 429, 1992.
Crooker, N.U., G.L. Siscoe, S. Shodan, D.F. Webb, J.T.
Gosling, and E.J. Smith, J. Geophys. Res., 98, 9371 1993.
Crooker, N.U., J.T. Gosling, and S.W. Kahler, J. Geophys. Res.,
103, 301, 1998.
Fenrich, F. and J. Luhmann, Geophys, Res. Lett.,25, 2999, 1998.
Forbes, T.G., J. Geophys. Res., 105, 23,153, 2000.
Galvin, A.B., in: Coronal Mass Ejections, N. Crooker et al.
(Eds.) GM 99 Washington, D.C., AGU, p. 253--260, 1997.
Gopalswamy, N., N. Nitta, P.K. Manoharan, A. Raoult, and M.
Pick, Astron. Astrophys., 347, 684-695, 1999.
11
Gopalswamy, N. et al., Geophys. Res. Lett., 27, 1427, 2000.
Gopalswamy, N., M.L. Kaiser, R.A. Howard, and J.L. Bougeret,
Astrophys. J., 548, L91-94, 2001.
Gopalswamy, N., S. Yashiro, G. Michalek, M.L. Kaiser, R.A.
Howard, D.V. Reames, R. Leske, and T. Von Rosenvinge,,
Astrophys. J. Lett., submitted, 2002.
Gosling, J.T., J. Geophys. Res., 98, 18,937-18,949, 1993.
Gosling, J.T. and D.J. McComas, Geophys. Res. Lett., 14, 355,
1987.
Gosling, J.T., D.J. McComas, JL. Phillips, and S.J. Bame, J.
Geophys. Res., 96, 7831-7839, 1991.
Harrison, R.A., Astron. Astrophys., 162, 283-291, 1986.
Howard, R.A., D.J. Michels, N.R. Sheeley, Jr., and M.J.
Koomen, Astrophys. J., 263, L101-L104, 1982.
Howard, R.A., Sheeley, N.R. Jr., Koomen, M.J., and Michels,
D.J., J. Geophys. Res., 90, 8173-8191, 1985.
Howard, R.A., Sheeley, N.R. Jr., Michels, D.J. and Koomen,
M.J., in The Sun and the Heliosphere in Three Dimensions, R.
G. Marsden (Ed.), 107-111, 1986.
Howard, R.A. et al., in Coronal Mass Ejections, N. Crooker et al. (eds.), GM 99, Washington, D.C., AGU, 17-26, 1997.
Hudson, H.S. and E.W. Cliver, J. Geophys. Res., 106, 25,199-
25,213, 2001.
Hudson, H.S. and D. F. Webb, in Coronal Mass Ejections, N.
Crooker et al. (Eds.), GM 99, Washington, D.C., AGU, 27,
1997.
Hudson, H., Haisch, B. and Strong, K.T., J. Geophys. Res., 100,
3473-3477, 1995.
Hudson, H.S., J.R. Lemen, O.C. St. Cyr, A.C. Sterling, and D.
F. Webb, Geophys. Res. Lett., 25, 2481-2484, 1998.
Hundhausen, A.J., J. Geophys. Res., 98, 13,177- 13,200, 1993.
Hundhausen, A.J., in The Many Faces of the Sun, K. Strong et al. (Eds.), New York, Springer-Verlag, 143-200, 1999.
Hundhausen, A.J., Burkepile, J.T. and St. Cyr, O.C., J.
Geophys. Res., 99, 6543-6552, 1994a.
Hundhausen, A.J., Stanger, A.L., and Serbicki, S.A., in Solar
Dynamic Phenomena and Solar Wind Consequences, ESA SP-
373, ESTEC, Noordwijk, The Netherlands, 409-412, 1994b.
Kahler, S.W., Astrophys. J., 214, 891-897, 1977.
Kahler, S.W., Annu. Rev. Astron. Astrophys., 30, 113, 1992.
Lepping, R.P. and D. Berdichevsky, Recent Res. Devel.
Geophys., 3, 77-96, 2000.

Lepping, R.P., J.A. Jones, and L.F. Burlaga, J. Geophys. Res.,
95, 11,957-11,965, 1990.
Low, B.C. and J.R. Hundhausen, Astrophys. J, 443, 818, 1995.
Luhmann, J.G., J.T. Gosling, J.T. Hoeksema, and X. Zhao, J.
Geophys. Res., 103, 6585, 1998.
MacQueen R.M and Fisher, R.R., Solar Phys., 89, 89-102, 1983.
Martin, S.F., R. Bilimoria, and P.W. Tracadas, n Solar Surface
Magnetism, eds. Rutten, R. J., and C. J. Schrijver, Kluwer
Academic, Holland, 1994.
Marubashi, K., Adv. Space Res. 6, 335-338, 1986.
Marubashi, K., in Coronal Mass Ejections, Geophys. Monogr.
Ser., vol. 99, edited by N. Crooker, et al., p. 147, AGU, 1997.
Mulligan, T., C.T. Russell, and J.G. Luhmann, Geophys. Res.
Lett., 25, 2959-2963, 1998.
Odstrcil, D., and V. J. Pizzo, J. Geophys. Res., 104, 28,225-
28,239, 1999.
Plunkett, S.P. et al., in Proceedings of First ISCS Workshop,
Adv. Space Res., in press, 2002.
Plunkett, S.P. et al., Solar Phys., 194, 371-391, 2000.
Richardson, I.G., E.W. Cliver, and H.V. Cane, J. Geophys. Res.,
105, 18,203, 2000.
Richardson, I.G., E.W. Cliver, and H.V. Cane, Geophys. Res.
Lett., 28, 2569-2572, 2001.
Rust, D.M., Geophys. Res. Lett., 21, 241-244, 1994.
Russell, C.T., in Space Weather, P. Song et al. (Eds.), p 123-
141, Geophys. Monograph 125, AGU, Washington, D.C, 2001.
Russell, C. and R..McPherron, J. Geophys. Res., 78, 92, 1973.
Rust, D. and Kumar, A., Astrophys. J., 464, L199-L202, 1996.
Schwenn, R. et al., Solar Phys., 175, 667-684, 1997.
Sheeley, N.R., Jr., Howard, R.A., Koomen, M.J., and Michels,
D.M., Astrophys. J., 272, 349-354, 1983.
Sheeley, N.R., Jr., Walters, J.H., Wang, Y.-M., and Howard,
R.A., J.Geophys.Res.,104, 24,739- 24,767, 1999.
Srivastava, N., Schwenn, R., Inhester, B., Stenborg, G., and
Podlipnik, B., in Solar Wind Nine, S.R. Habbal et al. (Eds.), AIP
Conf. Proc. 471, AIP, Woodbury, NY, 115-118, 1999.
Sterling, A.C. and Hudson, H.S., Thompson, B.J. and Zarro,
D.M., Astrophys. J., 532, 628-647, 2000.
St. Cyr, O.C., Burkepile, J.T., Hundhausen, A.J., and Lecinski,
A.R., J. Geophys. Res., 104, 12,493- 12,506, 1999.
St. Cyr, O.C. et al., J. Geophys. Res., 105, 18,169-18,185, 2000.
12
Subramanian, P., Dere, K.P., Rich, N.B., and Howard, R.A., J.
Geophys. Res., 104, 22,321- 22,330, 1999.
Svestka, Z. and E.W. Cliver, in Eruptive Solar Flares, Z.
Svestka et al. (Eds.), New York, Springer-Verlag, 1-11, 1992.
Thompson, B.J. et al., Astrophys. J., 517, L151- L154, 1999a.
Thompson, B.J., et al., in Sun-Earth Plasma Connections, J. L.
Burch et al. (Eds.), 31-46, Geophys. Monograph 109, AGU,
Washington, D.C, 1999b.
Thompson, B.J., E.W. Cliver, N. Nitta, C. Delannee and J.P.
Delaboudiniere, Geophys. Res. Lett., 27, 1431-1434, 2000.
Thomsen, M.F., J.E. Borovsky, D.J. McComas, R.C. Elphic, and
S. Maurice, Geophys. Res. Lett., 25, 2545-2548, 1998.
Tsurutani, B.T., W.D. Gonzalez, F. Tang, S.I. Akasofu, and E.J.
Smith J. Geophys. Res., 93, 8519-8531, 1988.
Tsurutani, B.T., W.D. Gonzalez, A.L.C. Gonzalez, F. Tang, J.K.
Arballo, and M. Okada, J. Geophys.Res., 100, 21,717, 1995.
Vourilidas, A., P. Subramanian, K.P. Dere, and R.A. Howard,
Astrophys. J., 534, 456-467, 2000.
Wang, Y.-M., J. Lean, and N.R. Sheeley, Jr., Geophys. Res.
Lett., 27, 505, 2000.
Webb, D.F., in Eruptive Solar Flares, Z. Svestka, B.V. Jackson, and M.E. Machado (Eds.), 234, Springer-Verlag, Berlin, 1992.
Webb, D.F., IEEE Trans. Plasma Sci., 28, 1795-1806, 2000.
Webb, D.F. and R.A. Howard, J. Geophys. Res., 99, 4201, 1994.
Webb, D.F., R.A. Howard, and B.V. Jackson, in Solar Wind
Eight, D. Winterhalter et al. (Eds.), AIP Conf. Proc. 382.
Woodbury, N.Y., Amer. Inst. of Phys., 540-543, 1996.
Webb, D., S. Kahler, P.McIntosh, and J. Klimchuk, J. Geophys.
Res., 102, 24,161-24,174, 1997.
Webb, D.F., E.W. Cliver, N. Gopalswamy, H.S. Hudson, and
O.C. St. Cyr, Geophys. Res. Lett., 25, 2469-2472, 1998.
Webb, D.F., E.W. Cliver, N.U. Crooker, O.C. St. Cyr and B.J.
Thompson, J. Geophys. Res., 105, 7491-7508, 2000.
Webb, D.F., M. Tokumaru, B.V. Jackson and P.P. Hick, EOS,
82, F1001, 2001a.
Webb, D.F., N.U. Crooker, S.P. Plunkett and O.C. St. Cyr, in
Space Weather, P. Song et al. (Eds.), p 123-141, Geophys.
Monograph 125, AGU, Washington, D.C, 2001b.
Zhang, J., K.P. Dere, R.A. Howard, M.R. Kundu and S.M.
White, Astrophys. J., 559, 452-462, 2001.
Zhao, X.-P., and J.T. Hoeksema, Geophys. Res. Lett., 24, 2965-
2968, 1997.
Zhao, X.-P., and J.T. Hoeksema, J. Geophys. Res., 103, 2077-

2083, 1998.
Zhao, X.P., J.T. Hoeksema, J.T. Gosling, and J.L. Phillips, in
Solar- Terrestrial Predictions - IV, Vol. 2, edited. by J. Hruska et al., p. 712, NOAA, Boulder, Colo., 1993.
13