The Potential Geomagnetic Effectiveness of Coronal
Mass Ejection and Stream Interaction Regions
G. M. Lindsay, C. T. Russell, and J. G. Luhmann
Institute of Geophysics and Planetary Physics
University of California, Los Angeles
Abstract
Previous studies have indicated that the largest
geomagnetic storms are caused by extraordinary increases in
the solar wind velocity and/or southward interplanetary
magnetic field (IMF) produced by coronal mass ejections and
their associated interplanetary shocks. However, much more
frequent small to moderate increases in solar wind velocity
and compressions in the IMF can be caused by either coronal
mass ejections or fast/slow stream interactions. This study
examines the relative statistics of the magnitudes of
disturbances associated with the passage of both
interplanetary coronal mass ejections (CMEs) and stream
interaction regions using an exceptionally continuous
interplanetary data base from the Pioneer Venus Orbiter at
0.7 AU throughout most of solar cycle 21 (1979-1988). It is
found that both stream interactions and CMEs produce magnetic
fields significantly larger than the nominal IMF. Increases
in field magnitude that are up to two and three times higher
than the ambient field are observed for stream interaction
regions and CMEs, respectively. Both stream interactions and
CMEs produce large positive and negative Bz components at 0.7
AU (~14 nT on average), but only CMEs produce Bz magnitudes
greater than 35 nT. CMEs are often associated with sustained
periods of positive or negative Bz, whereas stream
interaction regions are more often associated with
fluctuating Bz. CMEs tend to produce larger solar wind
electric fields than stream interactions. Yet, stream
interactions tend to produce larger dynamic pressures than
CMEs. Dst predictions based on the solar wind duskward
electric field and dynamic pressure indicate that CMEs
produce the largest geomagnetic disturbances while the low
speed portion of stream interaction regions are
least geomagnetically effective. Both stream interaction
regions and CMEs contribute to low and moderate levels of
activity with relative importance determined by their solar-
cycle dependent occurrence rates.
Introduction
Many previous studies have shown that geomagnetic storms
are caused by changes in the solar wind, such as an increase in
the incoming momentum (nV) or a southward turning in
the interplanetary magnetic field (IMF) [Siscoe (1966);
Crookeret al., (1977); Gosling et al., (1987) and (1990)].
Siscoe(1966) derived a relationship between magnetic activity
and changes in the incoming momentum flux. Crooker et al.,
(1977), confirmed that geomagnetic activity (as given by the Ap
index) was indeed well correlated with incoming momentum flux,
but that during solar-cycle 20 geomagnetic activity was best
correlated with the product of the southward component of the
IMF (Bz) and the square or higher power of the solar wind speed
(V). Murayama et al.,(1982) found that magnetospheric
response, as monitored by the Dst index, was best correlated
with a combination of the previously mentioned parameters:
VBz(nV2)1/3. Tsuratani et al., (1990) formulated a direct
interplanetary criteria for geomagnetic storms of intensity Dst
<-100 nT: the duskward electric field (VBz) must be larger than
5 mV/m, and in the Earth's reference frame, these fields just
last longer than three hours.
The interplanetary sources of enhanced
interplanetary parameters are stream interaction regions, CMEs,
and/or interplanetary shocks. The association between high-
speed recurrent streams and geomagnetic activity was pointed
out in Sheeley et al., (1976). Sheeley found an agreement
between the recurrence patterns of coronal holes, high speed
solar wind streams and geomagnetic disturbances showing a
repeatable pattern of enhanced geomagnetic activity with ~27
day period (approximately equal to the period of
solar rotation). These observations support results such as
those in Crooker et al., (1977) who found that
geomagnetic activity is closely related to solar wind velocity
alone. In a follow-on study, Sheeley etal., (1979) examined
the change in recurrent storm patterns with solar cycle. It
was found that unlike the speed enhancements which diminish in
strength with solar cycle (Vmax ~600-700 km/s versus Vmax ~700-800
km/s), the recurrent disturbances maintained their level of
intensity. This observation argues for a dependence of
geomagnetic activity on density (since velocity and number
density are anti-correlated and assuming the polarity of Bz is
independent of V) as proposed by Murayama et al., (1982).
The geomagnetic effectiveness of CMEs was investigated by
Gosling et al., (1991). This study concluded that all major
geomagnetic storms (as defined by Kp-max 8- and Kp 6- for
at least three three-hour intervals during a twenty-four hour
period) occurring from 1978-1982 were accompanied by
interplanetary shocks. Approximately, 93% of the
interplanetary shocks producing major storms were found to be
driven by CMEs. Gosling, et al., (1991) also found that major
storms comprised only ~4% of the total number of geomagnetic
storms occurring during 1978-1982. Large storms represented
~8% of the total number of storm sand were found to be
primarily caused by CME-driven shocks. Medium and small storms
comprised the remaining ~88% of the total number of storms.
Yet, most of these medium and small storms (~68%) could not be
explained in terms of either CMEs or shocks. Overall, only 37%
of all geomagnetic storms could be explained in terms of CME
driven shocks or CMEs.
The purpose of this study is to determine whether the ~63%
of geomagnetic storms not related to CME passage can be the
consequence of stream interaction region passage.
This analysis is done by examining the solar wind velocity (V),
southward magnetic field (Bz), and the parameters VBz,and V2Bz
associated with the passage of CMEs and fast/slow stream
interactions as observed by the Pioneer Venus Orbiter(PVO) at
0.7 AU. The results of the analysis at 0.7 AU will be compared
to a similar analysis of CMEs performed at 1.0 AU by Gosling
etal., (1991). Because the study at 1.0 AU provides a definite
association between the characteristics of local solar wind and
the intensity of the resulting geomagnetic storm, inferences
can be made as to the potential geomagnetic effectiveness of
both CMEs and stream interactions observed at 0.7 AU.
Data Analysis
The data used in this study were obtained on
the Pioneer Venus Orbiter (PVO) spacecraft from 1979
to 1988, approximately the duration of solar cycle 21. For
our purpose, the averaged 10-min-resolution University of
California, Los Angeles, magnetometer data [Russell et
al., 1980] and the full 9-min-resolution plasma data
[Intriligator et al., 1980] as derived by the Ames
Research Center investigators (both archived at the
National Space Science and Data Center) were used. The
coordinate system of the data is the Venus Solar Orbital
system (VSO), wherein x lies in the orbital plane and
points toward the Sun, z is directed north from the Venus
orbital plane, and y lies in the orbital plane pointing
opposite the direction of orbital motion. The orbit of
Venus is inclined ~3 from the solar equator in contrast
to Earth's 7.3 . The relationship between the VSO system
and the geocentric solar magnetospheric (GSM) system is
as follows: x lies in the orbital plane (within 3 of
the ecliptic plane) and points toward the Sun in both VSO
and GSM; y points opposite the direction of planetary
motion; however, in GSM, y is perpendicular to the dipole
axis so that it points below and above the ecliptic plane
as the dipole axis rotates about the Earth's spin axis,
whereas in VSO, y always lies in the Venus orbital plane;
in both GSM and VSO, z completes the right-handed system,
but in VSO, z points northward and is always
perpendicular to the orbital plane, whereas in GSM, z
points in the same sense as the northern magnetic pole but
is not necessarily perpendicular to the ecliptic. In
the following statistical study, we directly compare
properties in the two different coordinate systems,
assuming that since 10 years of data are considered,
the differing directions of their polar axes do not
alter the basic distribution of Bz values. However, for
any particular storm, the precise direction of the
magnetic field in GSM coordinates will depend on the time
of arrival of the field at the Earth.
Figure 1. This figure shows a typical example of a
CME driving an interplanetary shock as seen in the
PVO magnetometer and plasma data. The top three panels show
the ion temperature, density, and speed. The next two panels
show total magnetic field, Bt, and north/south component of
the IMF, Bz. The bottom two panels display the related time
series of V2Bz and VBz. Each panel spans a two day interval.
The shock, marked by the dashed line, is identified as
a discontinuous increase in ion temperature, density,
velocity, and magnetic field magnitude. In the
sheath region, the interval between the shock and the CME
boundary, the velocity increases, and the magnetic field is
enhanced and rotates from southward to northward. The CME is
characterized by a below-ambient ion temperature,
monotonically decreasing velocity, and smoothly varying total
magnetic field. The smooth, large scale rotation in Bz, from
northward to southward, suggests that this CME is a magnetic
cloud.
Figure 2. This figure shows a typical example of a
stream interaction region as seen in the PVO magnetometer
and plasma data. The panels span a two day time period and
are organized in the same manner as in Figure 1. The
stream interaction is identified by a sudden increase in
ion temperature and velocity accompanied by a decrease in
ion density. The stream interface occurs where the change
in velocity is steepest. It is marked by the solid vertical
line, is the thin region which separates the slow, dense
plasma from fast, thin plasma.
Figure 3. This plot shows the distribution of solar
wind velocity associated with stream interactions and
coronal mass ejections observed at 0.7 AU. Percent
occurrence, on a linear scale, is plotted versus velocity (in
50 km/s bins). For comparison purposes, the distribution of
all solar wind data at 1.0 AU (given in Gosling et al.,
1991) is also shown. It is immediately obvious that the bulk
of the distributions from the CMEs (thick line) and
the stream interactions (thin line) are similar. The CMEs
have a median velocity of ~364 km/s, while the
stream interactions have a median velocity of ~347 km/s.
The differences in the two distributions arise in their high
velocity tails. The steam interaction region velocities
extend to ~800 km/s, but the CME distribution exhibits a high
speed tail to velocities in excess of ~1000 km/s. The solar
wind distribution has a median velocity of ~355 km/s, roughly
equivalent to that of both the CMEs and stream interactions.
The solar wind distribution ends at velocities ~100 km/s
higher than the stream interaction distribution and ~200 km/s
less than the CME distribution.
Figure 4. Shown here is the distribution of Bz magnitudes
observed in association with the passage of CMEs and
stream interactions at 0.7 AU. The distribution of Bz for
all solar wind data at 1.0AU, scaled to 0.7 AU, is also
shown. As is expected, the medians of all three are ~0,
indicating that the north/south field fluctuates as
often positive as negative. Here, it is seen that the
stream interaction distribution and the solar wind
distribution have a similar range, ~ 24 nT. However, the
CME distribution shows a large high field tail for both
positive and negative Bz.
Figure 5. This figure shows the related distribution of
density. Percent occurrence is plotted on a linear scale
versus density. This plot shows that both
stream interaction regions and CMEs have
characteristically higher densities associated with their
passage than does the ambient solar wind. Further,
stream interaction regions (median density = 8.2 cm-3) tend
to have higher associated densities than CMEs
(median density = 7.3 cm-3). This observation is a natural
consequence of the fact that the solar wind is always
compressed in a stream interaction region. On the other
hand, the average CME travels near the solarwind speed
and thus usually does not produce significant compression
in the solar wind. Moreover, the bodies of CMEs
typically have densities similar to the density of the
surrounding solar wind. The enhanced densities in the CME
density distribution result from a combination of the few
fast CMEs compressing the ambient medium ahead and the few
CMEs which are themselves denser than average.
Figure 6. This figure shows the variation of
|VBz| distributions for regions in the CME sheath and within
the CME (top panel), and prior to and following stream
interface (bottom panel). Percent occurrence is plotted in a
log scale vs |VBz| in mV/m. Solid lines represent the pre-CME
or pre-SI values while dashed lines represent the post-CME or
post-SI values. In both panels, the dashed line with the cross
symbols represents the distribution of the quiet solar wind.
The point where interplanetary conditions are conducive to
major geomagnetic storms (as defined by Dst <-100 nT, (Tsuratani
et al., 1990)) is marked based on the criteria |VBz|>5 mV/m.
What is striking here is that the pre-CME and post-CME
distributions appear very similar. However, the pre-CME median
is ~0.58 mV/m, and the post-CME median is ~0.85 mV/m. This
difference arises due to the small percentage (~1.2%) of the
post-CME distribution which has |VBz| as large as ~30mV/m.
Approximately 9.5% of both distributions fall at values
greater than ~5.0 mV/m in the range associated with Dst <-100
nT. There is a greater difference between the distributions of
pre-SI and post-SI periods, especially between ~4.00 -10.0 mV/m.
The pre-SI median is ~0.36 mV/m. The post-SI median is ~0.56
mV/m. Only ~1.51% of the pre-SI distribution is greater than
~5.0 mV/m, whereas ~3.47% of the post-SI distribution is
greater than that value. Figure 6 shows that both pre-CME
regions and post-CME regions have characteristics much more
conducive to geomagnetic storms than either pre-SI or post-SI
regions, especially for major storms.
Figure 7. This figure shows distributions of
dynamic pressure in the same format as Figure 6.
Percent occurrence is plotted on a log scale versus
dynamic pressure in nanopascals. In the top panel the
pre-CME and post-CME cases are very similar and
demonstrate that equal magnitude dynamic pressures are
likely to occur in both the pre-CME and post-CME regions.
In contrast, in the bottom panel the pre-stream interface
curve has a median of 6.5 nPa and extends beyond the
cutoff of 20nPa, whereas the post-stream interface curve
has a range of 1 - 20 nPa and a median of 5.2 nPa. The
median dynamic pressure for both the pre- and post-stream
interface regions is higher than those of the pre- and
post-CME regions. Increases in the horizontal component
of the Earth's magnetic field (H) are related to the
square root of the solar wind dynamic pressure[Siscoe et
al., 1968]. The observations shown in Figure 7 imply that
the largest increases in H will result from CME passage,
but more generally, an increase in H will result from
stream interaction region passage. Further, the largest
increases in H are likely to result from the passage of
the pre-stream interface portion of the
interaction region.
Figure 8. The time integrated distribution of |VBz|
is displayed in here, configured similarly to Figures 6 and 7.
The top panel shows that the influence imposed by the pre-CME
and post-CME regions is even more alike than Figures 6 and 7
would indicate. The distributions are nearly coincident, as
are the medians (~68 and ~69 Wb/m). The stream interaction-
associated distributions are quite different and show two
distinct peaks. This difference is evident in the medians:
the bulk of the pre-SI distribution is narrow and has a median
of ~41 Wb/m; the post-SI distribution is broader and has a
median of ~59Wb/m. The major storm criterion of Tsuratani et
al, (1992),|VBz| > 5 mV/m for at least 3 hours, implies an
integrated value of ~54 Wb/m. ~56% of both the pre-CME and
post-CME distributions and 48% of the post-SI distribution meet
or exceed this criteria. Only ~32% of the pre-SI distribution
meets or exceeds this level.
Figure 9. This figure shows the distributions of
predicted Dst, calculated from the formula of Burton et al.
[1975] for the CME and stream interaction region data
scaled to 1.0 AU. Percent occurrence is plotted on a log
scale versus Dst in units of nanoteslas. In the top
panel, the pre- and post-CME distributions peak at ~-10
nT. For Dst values greater than zero, both distributions
are similar, ending at ~50 nT. For Dst values less than
zero, the distributions are quite different. The post-
CME distribution shows a greater likelihood of large
negative values and, in fact, extends to ~-400 nT. The
pre-CME distribution ends at ~-200 nT. These differences
are reflected in the medians: ~-6.1 nT for the pre-CME
distribution and ~-21.6 nT for the post-CME distribution.
The pre- and post-stream interface distributions, shown in
the bottom panel, are notably different from the CME
distribution. The pre-stream interface distribution has
a range of ~-100 nT to ~60 nT and a peak in the
distribution at ~0 nT. The post-stream interface
distribution has a range of ~-100 nT to 40 nT and peaks at
~-15 nT. Although the two distributions have similar
ranges, the offset of the peaks is seen in the medians: ~-
2.8 nT for the pre-stream interface distribution and ~-
14.6 nT for the post-stream interface distribution.
These distributions suggest that both the post-CME
and post-stream interface regions will produce
larger geomagnetic storms than the corresponding pre-CME
and pre-stream interface regions. According to the
most likely Dst values, the order of decreasing
geoeffectiveness is post-CME, post-stream interface, pre-
CME, and pre-stream interface.
Conclusion
By surveying the characteristics of stream interactions,
coronal mass ejections, and the ambient solarwind throughout
the last solar cycle, the potential geomagnetic effectiveness
of the disturbed regions associated with stream interaction and
CME passage has been determined. It has been found that both
stream interactions and CMEs can possess the high velocities
and southward magnetic fields conducive to geomagnetic storm
production. However, CMEs exhibited the strongest southward
fields and highest velocities, and therefore, are most likely
to cause the largest geomagnetic storms. On the other hand,
there also exists a population of stream interactions with
VBz characteristics associated with severe, major, and
minor storms. The existence of this population may explain
the source of the storms identified in Gosling et al.,
(1990), that were not associated with either a CME or a CME-
driven shock. Upon surveying the characteristics of VBz both
before and after stream interface passage and before the CME
leading edge and within the CME itself, it is found that the
post-stream interface regions are more likely to be
geomagnetically effective than the pre-stream interface region.
Pre-stream interface regions and quiet solar wind have similar
characteristics and are not very likely to be geomagnetically
effective. Both the pre-CME and post-CME regions have similar
characteristics and are very likely to be a source
of geomagnetic storms.
Back to the GEM online poster page
Last modified: July 03, 1995
