A NEW MORPHOLOGY OF SOLAR ACTIVITY AND RECURRENT

GEOMAGNETIC DISTURBANCES: THE LATE-DECLINING PHASE

OF THE SUNSPOT CYCLE

S.-I. AKASOFU1, , H. WATANABE2 and TAKAO SAITO3

1 International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, Alaska, U.S.A.
2 Secondary School, Faculty of Education, The University of Tokyo, Nakano-ku, Tokyo, Japan
3 Taihaku 3-6-29, Taihaku-ku, Sendai 982-0212, Japan

( Author for correspondence; E-mail: sakasofu@iarc.uaf.edu)

(Accepted in final form 24 November 2004)

Abstract. Certain aspects of the Sun and resulting geomagnetic disturbances can be studied better on
the source surface, an imaginary spherical surface of 3.5 solar radii, than on the photospheric surface.
This paper presents evidence that the Sun exhibits one of the most fundamental aspects of activities
most clearly during the late-declining phase of the sunspot cycle. It is the period when 27-day average
values of the solar wind speed and of geomagnetic disturbances tend to be highest during the sunspot
cycle.
Important findings of this study on the late-declining phase of the sunspot cycle are the following:

1. By introducing a new coordinate system, modifying the Carrington coordinates, it is shown that
various solar activity phenomena, solar flares, the brightest coronal regions, and also the lowest
solar wind speed region, tend to concentrate in two quadrants, one around 90 in longitude in the
northern hemisphere (NE) and the other around 270 in longitude in the southern hemisphere
(SW). For this reason, the new coordinate system is referred to as the NESW coordinate system.
2. It is shown that the above results are closely related to the fact that the neutral line exhibits
a single wave (sinusoidal or rectangular) in both the Carrington coordinates and the NESW
coordinate system during the late-declining phase. The shift of the neutral line configuration
during successive solar rotations during the late-declining phase causes longitudinal scatter of
the location of solar flares with respect to the neutral line in a statistical study. The NESW
coordinate system is designed to suppress the shift, so that the single wave location is fixed and
thus a nest of solar flares emerges in the NE and SW quadrants.
3. It is also shown that the single wave is the source of the double peak of the solar wind speed
and two series of recurrent geomagnetic disturbances in each solar rotation, making the 27-day
average solar wind and geomagnetic disturbances highest during the sunspot cycle. The double
peak is a basic feature during the late-declining phase, but is obscured by several complexities
which we identified in this paper; see item 8.
4. The single wave of the neutral line configuration can be approximated by three dipole fields, one
which can be represented by a central dipole (parallel or anti-parallel to the rotation axis) and
two hypothetical dipoles on the photosphere. This configuration is referred to as the triple dipole
model.
5. The location of the two hypothetical photospheric dipoles coincide with the two active regions
(solar flares, the brightest coronal region) and also the lowest solar wind speed region in the
NESW coordinate system; the lowest solar wind regions are the cause of the valleys of the
double peak of the solar wind speed.
Space Science Reviews (2005) 120: 2765
DOI: 10.1007/s11214-005-8052-3 
C Springer 2005
28 S.-I. AKASOFU ET AL.

6. The two hypothetical dipole fields actually do exist at the location of the two active regions in a
coarse magnetic map (5 5 ). The two dipoles follow the HaleNicholson polarity law. Thus,
they are real physical entities.
7. The apparent meridional rotation of the dipolar field on the source surface during the sunspot
cycle results from combined changes of both the central dipole field and of the two photospheric
dipoles, although the central dipole remains axially parallel or anti-parallel. Thus, the Sun has
a general field that can be represented by an axially aligned dipole located at the center of the
Sun throughout the sunspot cycle, except for the sunspot maximum period when the polarization
reversal occurs.
8. The complexity of recurrent geomagnetic disturbances can also be understood by having the
NESW coordinate system for various solar phenomena and the relative location of the earth with
respect to the solar equatorial plane.
9. As the intensity of the two dipoles decreases toward the end of the sunspot cycle, the amplitude
of the single wave decreases, and the neutral line tends to align with the heliographic equator.
10. The neutral line shows a double wave structure during certain epochs of the sunspot cycle. In
such a situation, it can be considered that two NESW coordinate systems are present in one
Carrington coordinate, resulting in four active regions.
11. The so-called classical sector boundary arises when the peaks (top and bottom) of the single
wave reached 90 in latitude in both hemispheres.
12. In summary: A study of the late-declining period of the sunspot cycle is very important compared
with the sunspot maximum period. In the late-declining period, the Sun shows its activities
in the simplest form. It is suggested that some of the basic features of solar activities and
recurrent geomagnetic disturbances that have been studied by many researchers in the past can
be synthesized in a simplest way by introducing the NESW coordinate system and the triple
dipole model. There is a possibility that the basic results we learned during the late phase of the
sunspot cycle can be applicable to the rest of the sunspot cycle.

1. The Neutral Line on the Source Surface

Correlation studies between the photospheric/coronal magnetic fields and the in-
terplanetary magnetic field (IMF) have led to two important concepts, the source
surface and the sector boundaries (cf. Wilcox et al., 1969; Wilcox and Ness, 1985;
Newkirk, 1971; Hoeksema, 1984; Schulz, 1973; Saito, 1975). The source surface
is an imaginary spherical surface of 3.5 solar radii (2.5 solar radii away from the
photosphere). The magnetic equator or more commonly called the neutral line on
the source surface is determined on the basis of the spherical harmonic analysis of
the photospheric field by assuming B = j = 0 in the corona and the magnetic
field lines are perpendicular to the source surface.
Our study of the relationship between solar activity and magnetic storms differs
from many earlier studies of solar activity. We begin our study by focusing on
the neutral line on the source surface, rather than photospheric or chromospheric
features. The reasons for the usefulness of this approach will become apparent in
the following sections; it is sufficient to note here that photospheric features are
often very complex, and some large-scale and fundamental aspects of the Sun may
be more readily noticeable at some points away sufficiently from the photosphere.
This is particularly true when studying geoactive features of the Sun.
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 29

Figure 1. The polarity of the source surface field during the sunspot cycle 21, 22, and 23 (from
Carrington Rotation from 1641 to 2003) The magnetic field points away from the Sun in the white
area and points toward the Sun in the black area (Hoeksema, WSO Homepage).

Figure 1 shows the magnetic equator or the neutral line on the source surface
during sunspot cycles 21, 22, and the first half of 23 (Hoeksema, WSO Homepage).
The neutral line is the dividing line between the region of positive/outward (white)
and negative/inward (black) components of the magnetic field on the source surface.
Unfortunately, accurate data describing the position of the neutral line are available
only for the two and a half cycles in the literature, but the points we attempt to
describe in the paper are clearly evident in Figure 1.
From Figure 1, it can be seen, as expected, that the neutral line is located nearly
parallel to the heliographic equator at both the beginning and end of the sunspot
cycle. This is obviously the simplest case, in which the solar magnetic field on the
source surface resembles an axial dipole, and solar and geomagnetic disturbances
tend to become minimum or absent. During the early phase of the sunspot cycle
21, the field was pointing away from the Sun in the northern hemisphere, while the
30 S.-I. AKASOFU ET AL.

field was pointing toward the Sun in the southern hemisphere. As the sunspot cycle
progressed, the neutral line exhibited irregular structures for a short period. During
the maximum epoch of the cycle, the source surface is divided longitudinally, one
half away from and the other half toward the sun. It is during such a period when
the concept of the so-called classical sector structure is most appropriate (Wilcox
et al., 1969; Svalgaard et al., 1974; Saito, 1972; Smith, 2001). During the declining
phase, the neutral line developed first a double wave structure, then a single wave;
the line reverted to a double wave structure for a few months, and finally returned
to a single wave structure during the late-declining phase. Toward the end of the
cycle, the amplitude of the single wave diminished, and the neutral line almost
aligned with the heliographic equator, but with reversed polarity. A similar trend
occurred during the sunspot cycles 22 and 23, except for the changes of reversal of
the magnetic polarity.
Figure 2 shows the heliospheric current sheet configuration for the Carrington
Rotation 1719 during the cycle 21, computed on the basis of the geometry of the

Figure 2. The source surface field distribution and the current sheet during the Carrington rotation
1719 to a distance of 2 au. The geometry of the current sheet is computed on the basis of the neutral
line on the source surface by assuring the solar wind blows radially away from the Sun beyond the
source surface (Akasofu and Fry, 1986).
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 31

neutral line on the source surface (Akasofu and Fry, 1986). The earths orbit is
also shown. The earth was located below (when the field was pointing toward the
Sun) during approximately one-half of the solar rotation and above (when the field
was pointing away from the Sun) during the rest. The change of the polarity of the
IMF observed at the location of the earth occurred when the earth encountered the
current sheet. Therefore, this situation is not really the classical longitudinal sector
boundary crossing, but is the current sheet crossing from above or below it. Except
for the sunspot maximum period, Figure 2 represents the most common situation
during the sunspot cycle.
Although it is necessary to examine the relationship between geomagnetic distur-
bances and the source surface pattern throughout the sunspot cycle, we pay attention
first to the late-declining phase of the sunspot cycle. The most important point to
be noted here is that this is the period when the neutral line has a relatively stable
and single (sinusoidal or rectangular) wave structure. In the following sections,
we show:

The two high-speed streams and recurrent magnetic storms occur during the late declining
phase of the sunspot cycle, when the neutral line exhibits a single wave structure on the source
surface.

To demonstrate the above statement, we examine first the stable single wave
structure of the neutral line on the source surface during the late-declining phase
of the sunspot cycle.
It is known that there is an arch-like magnetic structure below the neutral line.
Figure 3ac show the magnetic field configuration computed by Hakamada (1987).
Thus, it can be predicted that the corona is bright approximately along the neutral
line, since dense plasma is trapped in the arch-like magnetic field structure. This is
indeed the case. Figure 4 shows the coronal features in 1974, 1984, and 1995 for six
solar rotations during the declining phase for three solar cycles, the coronagraph
data from Mauna Loa in 1974, the coronagraph data from SMM in 1984, and the soft
X-ray data from Yohkoh, 1994, respectively. The neutral line on the source surface
is superimposed on each of the above datasets. This is one way to demonstrate the
presence of the single wave neutral line during the late-declining phase of every
sunspot cycle. The source surface is an imaginary surface, and thus the neutral line
is supposed to be an imaginary line, obtained by assuming the current-free corona,
namely B = j = 0, above the photosphere. In order to show that the neutral
line thus obtained is indeed physically meaningful, Figure 4 shows the neutral line
and coronal features for six successive rotations during the declining period (1974,
1984, and 1994) for the three sunspot cycles; note, however, that the neutral line
in Figure 3a is repeated for 1974, because it is the only one available. It can be
clearly seen that both agree reasonably well. As Figure 3c shows, by the definition
of the source surface (beyond which the field lines are considered to be radial), one
can consider that the top of the arch field lines are broken into two, one pointing
32 S.-I. AKASOFU ET AL.

(a)

(b)

(c)

Figure 3. (a) The source surface field distribution for the average situation in 1974. The thick line
indicates the neutral line. The magnetic field is pointing away from the Sun in the area of the solid
lines and toward the Sun in the area of the dashed lines (Hakamada, 1987). (b) The closed field lines
projected as the photosphere for the same period as Figure 3a (Hakamada, 1987). (c) The 3D view of
the closed field lines for the same period as Figure 3a (Hakamada, 1987).
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 33

Figure 4. The coronal brightness in 1974, 1984, and 1994. The neutral line is superposed; see the
text. From the top, the corona graph data from Mauna Loa in 1974 (cf. Figure 1a of Saito et al., 1988),
the coronal data from SSM in 1984 (Hundhausen, SMM Homepage), and the soft X-ray data from
Yohkoh in 1994 (Shibata, Yohkoh Homepage).

away from and the other pointing toward the Sun. This is substantiated by Zhao
and Webb (2003). However, note that the bright coronal features are expected to be
located well below two solar radii (Figure 3c), so that the neutral line and magnetic
arch are not always co-located.
From these observations, we can tentatively reach two important conclusions:
1. The neutral line inferred from the photospheric magnetic field can be considered
to be the magnetic equator on the source surface and can be considered to be
real.
2. The neutral line tends to exhibit a stable single wave structure during the late-
declining phase for about a year or so.

2. Solar Cycle Variations of the Neutral Line and a Triple Dipole Model

As shown earlier in Figure 1, the magnetic field on the source surface changes fairly
systematically during a sunspot cycle. Based on the spherical harmonic analysis of
the source surface field, Saito et al. (1989a) showed that the equivalent dipole field
on the source surface rotates its axis from 0180 , or from 180360 depending
on the sunspot cycle; see also Figure 3.5 of Saito (1988). This particular feature is
illustrated in Figure 5.
Figure 6a shows an example of the spherical harmonic analysis in our particular
study for Carrington Rotation 1720. When the magnetic field distribution during
the declining phase of the sunspot cycle is analyzed by the spherical harmonic
34 S.-I. AKASOFU ET AL.

Figure 5. The sunspot cycle variation of the neutral line and the equivalent dipole, together with the
sunspot number and the source surface field (Saito et al., 1989a).

method, the first term is a sinusoidal curve. This situation is shown in both the
rectangular and spherical coordinates. The central dipole based on the spherical
harmonic analysis is inclined with respect to the rotation axis, and the magnetic
equator is represented by a sinusoidal wave in the rectangular coordinates, and is an
inclined circle in the spherical coordinates. The magnetic equatorial plane is thus
an inclined flat circular plane.
However, the spherical harmonic analysis is a method of mathematical represen-
tation. Since the solar rotation is an important factor in generating the main dipole
field of the sun, it is likely that there is a magnetic field which may be represented
by a central dipole that is likely to be parallel or anti-parallel with respect to the
rotation axis. Indeed, it is well known that, the polar region of the photosphere
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 35

(a)

(b)

Figure 6. (a) This figure illustrates how the first term of the spherical harmonic analysis (right)
looks when the source surface field (left) is analyzed. Note that the central dipole and the neutral
line are inclined with respect to the ecliptic plane: Carrington rotation 1720. (b) The three-dipole
approximation: it is assumed that the central dipole should be parallel or anti-parallel with respect to
the rotation axis and two hypothetical photospheric dipoles are present. Compare the left-hand side
of Figure 6a (the spherical harmonic analysis) with the right-hand side of Figure 6b (the three-dipole
approximation).
36 S.-I. AKASOFU ET AL.

remains as a unipolar region throughout the sunspot cycle, indicating that the field
represented by the central dipole is parallel or anti-parallel to the rotation axis
throughout the sunspot cycle, except during the sunspot maximum period when
magnetic moment becomes zero and reverses. Thus, we try to represent the source
surface field in terms of the central dipole that is parallel (or anti-parallel, depending
on the sunspot cycle), together with hypothetical dipolar fields on the photosphere
(Saito et al., 1989b, 1991).
Indeed, based on the physical consideration that there is a dipole field that may
be represented by a central dipole, parallel or anti-parallel with respect to the ro-
tation axis, other dipole fields are required to reproduce the configuration of the
neutral line on the source surface. It can be shown that two hypothetical dipolar
fields on the photosphere, together with the central dipole field, can reproduce the
neutral line as a first approximation. Such a model is called the triple dipole model
in this paper. Figure 6b shows the single wave neutral line observed during Car-
rington Rotation 1720 (see Figure 6a) on the left, and the simulated neutral line
on the right, in both rectangular and spherical presentations using the triple dipole
model. It can be seen that the observed and computed neutral lines agree fairly well
when this simple assumption of the triple dipole model is used. The direction of the
two dipolar fields satisfies the HaleNicholson polarity law and thus is not random.
Note that in the rectangular coordinate, the two dipoles appear to be oriented in the
opposite direction, but in the 3D presentation, they are directed in approximately the
same direction. Figure 7 shows, from the top, the number of sunspots, the neutral
line on the source surface, the sunspot cycle variation of the assumed axial dipole,
the two hypothetical dipoles on the photosphere, and the neutral line computed by
assuming the three-dipole approximations for the sunspot cycle 21. It is surpris-
ing that the complex changes of the neutral line can be so well approximated by
the triple dipole model, although a better agreement needs a few more auxiliary
dipoles.
It is important to note that the total magnetic moment MT = M + M|| (where
M|| denotes the axial component and M denotes the combined moment of the two
photospheric dipoles) remains nearly finite throughout the sunspot cycle. In fact,
it is variations of M that make the apparent meridional rotation of the equivalent
dipole axis possible on the source surface (not on the photosphere) during the
sunspot cycle, as shown in Figure 5. During the sunspot maximum phase, the
dipole axis of the source surface field is almost perpendicular to the rotation axis ,
while the axial dipole is very weak.
Further, it can be shown that the two hypothetical dipoles needed to reproduce
the observed neutral line are not imaginary ones. Saito et al. (1989b) demonstrated
that two large-scale dipolar fields do exist on the photosphere at about the locations
where the triple dipole model indicates. First of all, in the upper two diagrams
in Figure 8a (the Carrington Rotation 1750), we show that the axial dipole and
two hypothetical dipoles projected radially onto the photosphere (namely, without
referring to the photospheric magnetic map) can reproduce well the neutral line on
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 37

Figure 7. The sunspot cycle variation of the source surface field based on the triple dipole model,
in which the central dipole (assumed to be parallel or anti-parallel to the rotation axis) and the two
photospheric dipoles are considered (Saito et al., 1989b).

the source surface. It can be seen that one of the two projected dipoles is located
in the northern hemisphere and the other in the southern hemisphere. Then, after
projecting the two dipoles onto the photosphere, we examine the corresponding
magnetic field map in a coarse mesh (55 ). One can clearly see a dipolar structure
at the location of each dipole. This is an example of a blind test we performed. The
polarization of the two dipolar fields should be such that they represent the fields
above the photosphere. The two dipoles do not correspond to individual sunspots.
In general, it is likely that a number of sunspot groups (or their remnants) as a whole
constitute a large-scale dipolar structure. In some Carrington rotations, one of the
photospheric dipolar fields is rather weak, so that the neutral line can be roughly
approximated by a single photospheric dipole. Such a case may be considered to be
the transition from the late-declining phase to the minimum phase of the sunspot
cycle, when the two photospheric dipoles of different intensities are decreasing
and one of them becomes very weak. Such a case occurred during 1974, as can be
seen in Figure 8b. In Section 5, we demonstrate further that the two dipolar regions
38 S.-I. AKASOFU ET AL.

determined by the triple dipole model are active centers and are flare nests during
the late-declining phase of the sunspot cycle.
It is interesting to note that Saito et al. (1993) predicted the shape of the outer
solar corona by extending the neutral (current) sheet from the neutral line on the
source surface (Figure 8c). The locations of the coronal streamers can be identified
as the longitude where the neutral line on the source surface has a large meridional
component and where folds develop (compare the photograph and the insert); the
folded portions of the sheet look brighter than a flat portion. Thus, some coronal
streamers are not independent streamers (like a sea urchin), but are folded portions
of the current sheet. The agreement between the observed and predicted corona is
another evidence that the neutral line can be considered to be real.

(a)

Figure 8. (a) The source surface field (upper left), the neutral line computed by the three dipoles (upper
right), and the 3D representation of the three dipoles (lower right) are shown. The two hypothetical
dipoles are projected onto a coarse mesh (5 5 ) map. Two large-scale dipolar fields are present
where the two hypothetical dipoles are projected. (b) The single wave case in which the neutral line
can be represented by the central dipole and a single photospheric dipole. (c) The solar outer corona
during the 1991 eclipse and the predicted shape of the solar outer corona based on the neutral line
(Saito et al., 1993).
(Continued on next page)
THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 39

(b)

(c)

Figure 8. (Continued)
40 S.-I. AKASOFU ET AL.

3. Recurrent Magnetic Storms

It was Bartels (1932), a geophysicist, who discovered the so-called M-region,

which is now understood to be a source of high-speed, geoactive stream of the solar
wind. It has long been known that during the late-declining phase of the sunspot
cycle, two high-speed streams are present. In this section, we show that the two
streams are the basic feature and are separated by roughly 180 in longitude; one
extends from the northern coronal hole and the other from the southern coronal
hole. Magnetospheric physicists have also learned that specific solar wind quan-
tities, such as the solar wind speed (V), and the IMF carried by the solar wind,
are responsible for magnetospheric disturbances. Magnetospheric storms and sub-
storms are mainly related to a specific combination of solar wind quantities, such as
VBz (Caan et al., 1975) and VB2 sin4 (/2) (Perreault and Akasofu, 1978; Akasofu,
1981), where B and Bz denote the magnitude and the northsouth component of
the IMF, respectively, and the polar angle of the IMF vector. The solar wind-
magnetosphere interaction constitutes a dynamo of power = V B 2 sin4 (/2) 2o
that drives the magnetospheric current system, where o denotes approximately
the radius of the cross-section of the magnetotail. Magnetic fields induced by the
current system can be identified as the magnetic fields of substorms and storms,
while the currents discharge process in the polar upper atmosphere causes the
aurora. Therefore, magnetic storms and the aurora are different manifestations of
enhanced dynamo power. As we shall see later, the latitudinal distribution of the
solar wind speed and the relative location of the earth with respect to the solar eclip-
tic plane also significantly modulate the solarterrestrial relationship that is also a
focus of our interest. This particular phenomenon has been known as the summer
and winter solstice minima and the spring and fall maxima (Chapman and Bartels,
1940).
Based on the degree of magnetic disturbances measured in terms of the mag-
netic index during several 11-year sunspot cycles, it is generally known that
magnetic disturbances tend to peak well after, not during, the sunspot maxi-
mum period, and that the so-called recurrent storms are responsible for it. Fig-
ure 9a, which shows the sunspot number R together with the daily C9 index
(see below) and the three hourly magnetic index Kp (27-day average) for the
period from 19552002, illustrates this trend clearly for several recent sunspot
cycles. Unfortunately, this well-established relationship is often forgotten and
must be learned again during every new sunspot cycle by a newer generation of
researchers.
The C9 index was devised by Bartels (1958) to study long-term trends of the
solar-terrestrial relationship in a semi-graphical/digital way. As we shall see, the C9
index is the index best suited for our particular purpose of studying the relationship
between the Sun and magnetic disturbances; for the relationship between the Kp,
C9, and the other indices, see Akasofu and Chapman (1972). The C9 index has
been available since 1884, but unlike the Kp index, the C9 index has not been used
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 41

(a)

(b)

Figure 9. (a) From the top, the sunspot number, the C9 index, and the Kp index from 1955 to 2003.
(b) The sunspot number and the 27-day average solar wind speed from 1965 to 2003.

much by solar physicists and geophysicists; Figure 9a illustrates and also assures
that the C9 and the familiar Kp indices are similar.
Figure 9b shows the sunspot number and the 27-day average solar wind speed
during 19652002 (OMNI data), including the sunspot cycles 2023. In contrast to
42 S.-I. AKASOFU ET AL.

the general belief that the solar wind speed is highest during the sunspot maximum
period, Figure 9b clearly illustrates that the 27-day average solar wind speed tends
to become higher toward the late phase of the sunspot cycle, well after the sunspot
maximum period. The solar wind speed during sunspot maximum years is relatively
low, even at a minimum; this was clearly seen in 1980. Indeed, the 27-day average
solar wind speed is almost out of phase with the sunspot variations. This trend is
also evident for other kinds of average speed. Using the C9 index, we examine the
implications of Figure 9b in more detail in the next section. In summarizing the
above facts, we have:

1. Geomagnetic disturbances measured in terms of 27-day average, tend to peak

well after, not during, the sunspot maximum period.
2. The 27-day average solar wind speed tends to peak toward the end of a sunspot
cycle.
3. The solar wind speed during sunspot maximum years is relatively low.

These important facts have not necessarily received full attention from magne-
tospheric physicists and solar physicists, who tend to emphasize the importance of
the sunspot maximum years. As early as 1955, Pecker and Roberts (1955) discussed
the tendency of sunspots to suppress the solar wind. They noted this trend based
on geomagnetic disturbances and proposed the concept of cone of avoidance, but
their findings received little attention from most solar physicists. This point will be
discussed further in later sections.
The occurrence of high-speed solar wind streams, together with highest levels
of geomagnetic disturbances, is why we first address the late-declining phase of
the sunspot cycle. We plan to demonstrate that the Sun shows its fundamental and
yet simple aspects of solar activity which can be best studied during the declining
phase.
One purpose of our study is to identify the reasons why the 27-day average
solar wind speed becomes highest and why the magnetosphere is most disturbed
during the late-declining period of the sunspot cycle. Although it is widely known
that the coronal hole is somehow responsible for those features, the specifics of the
relationships are not generally understood. One of the purposes of this paper is to
show that such an inquiry can lead us to understand some of the most basic aspect
of solar activity.
Effects of the high-speed solar wind streams on geomagnetic disturbances dur-
ing the late-declining phase of the sunspot cycle can be effectively studied using the
magnetic index C9. Figure 10a shows the C9 index for the declining period, cor-
responding to Figure 9a and b, namely 19731975, 19831985, and 19931995.
The first column shows sunspot activities in terms of numbers from 0 to 9, the
second column shows the year and the Bartels rotation number, the third column
the beginning date of the Bartels rotation number, the fourth column the C9 index
(09), and the last column shows the first part of the C9 index during the next
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 43

(a)

(b)
Figure 10. (a) The C9 index and the sunspot number during the declining phase of the sunspot
cycle, 19731975, 19831985, and 19931995. (b) The C9 index and the sunspot number during the
maximum phase of the sunspot cycle, 19481949, 19571958, and 19791980.

Bartels rotation; note that higher C9 numbers are shown by a thicker bold and
larger character.
Bartels devised this particular presentation to study a long-term solar-terrestrial
relationship. The daily C9 index for 27-days, one solar rotation period observed
from the earth, is given in one row. Therefore, if large C9 index numbers appear
vertically for several times, it means that geomagnetic disturbances recur approx-
imately every 27 days, for several solar rotations, indicating that there is some
fixed structure on the sun, which causes magnetic disturbances. If small C9 index
44 S.-I. AKASOFU ET AL.

numbers appear vertically for several solar rotations, it means that there is some
fixed structure on the sun, which prevents a high-speed solar wind to blow, and thus
geomagnetic disturbances on the earth are weak. In Section 5, we identify the two
lowest solar wind regions.
First, it can be seen clearly in Figure 10a that there were only a few sunspots
during highly disturbed periods. Sunspots in those years may be compared with
those in 19481949, 19571958, and 19791980 when the sunspot number was
very high (Figure 10b). In those years, magnetic disturbances were not necessarily
high and even relatively low, except for occasional high values of the C9 index;
compare Figure 10a and b, particularly 1974 and 1979.
Second, the high-speed years (19731974, 19831984, 19941995 in Figure 9b)
corresponded to the period of a series of magnetic storms called recurrent storms
which recurred for at least several solar rotations, each of which tended to last
for 710 days, instead of a few days for solar flare-associated storms. Third, it is
most important to note that those years of recurrent storms occurred well after the
sunspot maximum years (1968 for cycle 20, 1979 for cycle 21, 1989 for cycle 22);
this can be seen by simply comparing Figure 10a and b.
It may be pointed out here that in contrast to a common myth, the number of
sunspots is not directly related to the degree of geomagnetic disturbances (Watanabe
et al., 1996). This can easily be seen in Figure 10b. Years of high sunspot numbers
are not necessarily years of high magnetic disturbances. C9 = 8 or C9 = 9 storms
were rare in 19481949 or 19571958 and absent in 1979-1980, in spite of the fact
that the sunspot number was very high. Sunspots are not generally considered to be
geoactive (Mustel, 1964) unless flares and CMEs occur in their vicinity. Sunspots
without such activities may even suppress (Pecker and Roberts, 1955), divert
(Saemundsson, 1962), or deflect (Saito, 1965) high-speed solar wind, because
the magnetic field of sunspots forms an arch-like structure (Figure 3b and c). In
Section 5, we identify such regions.
Another important feature illustrated by Figure 10a is that there was a pair of re-
current storms during each solar rotation in 1974 that occurred during sunspot cycle
20 (19731975). Further, these two recurrent storms were separated by roughly two
weeks, indicating that the responsible high-speed streams were separated roughly
by 180 in longitude. Further, a pair of recurrent storms is separated by a pair of
very low C9 values.
On the other hand, in 1994 (during sunspot cycle 22 in Figure 10a) there was only
one recurrent storm in each solar rotation. In order to understand about the occur-
rence of only one recurrent storm, note that one recurrent storm was present in the
spring equinoctial months and the other in the fall equinoctial months, with a min-
imum of geomagnetic disturbances in June/July and December/January. Sunspot
cycle 21 was intermediate between the two sunspot cycles 20 and 22. Figure 11a
and b show that these particular trends are clearly seen during other sunspot cycles
as well. The double recurrent storms are clearly seen in 1922, 1930, 1942, and 1974
(Figure 11a), while single recurrent storms were observed in 1943, 1953, 1973, and
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 45

(a)

(b)

Figure 11. (a) The C9 index during the period when two recurrent magnetic storm groups were
prominent. (b) The C9 index during the period when one recurrent storm group was prominent.
46 S.-I. AKASOFU ET AL.

1994 (Figure 11b); note also that the sunspot numbers were low in the years in
Figure 11a and b. This asymmetry will be explained in Section 4.
Figure 12a and b show the daily average solar wind speed in 19811985 and
19901994. It is interesting to examine several of their features in light of the points
mentioned above. The year 1981 is typical of solar wind speed variations during
high sunspot numbers. There occurred intermittent impulsive changes throughout

(a)

Figure 12. (a) The daily variations of the solar wind speed from 1981 to 1985. (b) The daily variations
of the solar wind speed from 1990 to 1994.
(Continued on next page)
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 47

(b)
Figure 12. (Continued)

the year. The year 1981 may be contrasted with the years 1983 and 1984, which were
dominated by the two high-speed peaks during about a 1-month period, actually
every 27/2 days and thus the two-stream situation. The sunspot cycle 22 was some-
what different and complex compared with the sunspot cycle 21. In 1990, 1991,
and 1992 the two-stream situation occurred intermittingly. However, prominent
and large modulations became clear only in 1993 and 1994. Figure 12a and b also
show clearly that the double stream, not a single stream, is a fundamental aspect of
the solar wind during the late-declining phase. Similar two-stream situations were
observed by the Ulysses spacecraft (Smith et al., 2001), so that they are intrinsic
to the solar wind and are not caused by any of earths effects. Another important
feature to be noted is that many high-speed stream pairs do not necessarily have
the same intensity.
48 S.-I. AKASOFU ET AL.

From the above study of the C9 index and the solar wind, we may conclude:
1. The late-declining phase of the sunspot cycle coincides with the period when
the high-speed stream, well-developed recurrent storm groups occur. This phase
coincides with the period when a clear single wave structure of the neutral line
occurs.
In addition, the solar wind speed as a function of magnetic latitude on the source
surface has been studied by many workers (cf. Kojima et al., 1987; McComas et al.,
2003); see Section 5.
2. The solar wind speed is minimum at the neutral line (the magnetic equator on
the source surface) and increases toward higher latitudes.
Although it could be said that these facts are generally understood in terms of
the so-called spring and fall of geomagnetic disturbances, complexities associated
with the development of coronal holes and the tilt angle of the solar rotational axis
with respect to the ecliptic plane tend to obscure these facts, particularly those cases

(a)

Figure 13. (a) Upper: The assumed distribution of the solar wind speed distribution. Lower: The solar
wind speed variation depicted at a point (fixed in space) on the ecliptic place on the source surface as
the Sun rotates once. (b) From the top: The computed solar wind speed, density, the IMF magnitude,
the assumed variation of the IMF THETA angle, the completed IMF PHI angle, and the estimated ,
AE and Dst indices. For the THETA changes, see the text. (c) From the top: The solar wind speed,
the IMF magnitude, THETA and PHI angles, the computed , and the observed AE and Dst indices
in July and August 1974.
(Continued on next page)
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 49

(b)
Figure 13. (Continued)

shown in Figure 11b (spring maximum, but no fall maximum, or fall maximum,
but no spring maximum).

4. Simulation of Recurrent Geomagnetic Storms

Hakamada and Akasofu (1982) modeled the solar wind flow pattern by assuming
(1) the neutral line has a simple sinusoidal structure, (2) the solar wind speed is
minimum (300 km/s) at the neutral line, and (3) the solar wind speed increases
toward higher latitudes (Figure 13a) as confirmed in later studies. Obviously, this
represents a very simple situation, perhaps best represented during 1973 and 1974,
when coronal holes were well developed. At a fixed point in space in the ecliptic
plane at a distance of 3.5 solar radii from the center of the Sun, the solar wind speed
will have a double-wave structure, as the Sun rotates once in 25 days, 27 days seen
from the earth. This is equivalent to scanning the solar wind speed along the equator
50 S.-I. AKASOFU ET AL.

(c)
Figure 13. (Continued)

near the winter and summer solstices, and at 7.5 near the spring equinox and
+7.5 near the fall equinox. The magnetic equator is simulated by a sinusoidal wave
as an approximation of the single wave structure, so that the source regions of the
high-speed stream are simulated by the peak and valley regimes of the sinusoidal
neutral line. The sinusoidal curve is thus a very crude approximation of the single
wave neutral line. The basic assumption in the model is that during the declining
phase of the sunspot cycle, the polar coronal holes from both hemispheres extend
toward lower latitudes and often cross the equator, as is the case of the period in
1973 and 1974.
The spring and fall maxima of geomagnetic disturbances can also be understood
from Figure 13a. In this simple and idealized situation, during the winter and
summer solstice months, a fixed equatorial point at a radial distance of 3.5 solar
radii from the center of the Sun (not rotating with the Sun) scans the solar wind
along heliographic latitude 0 ; the scanned wind speed per solar rotation is shown
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 51

in the lower half of Figure 13a. The point depicts two peaks of speed (750 km/s)
per solar rotation. During the spring months, the point scans the solar wind along
the latitude 7.5 line. As a result, the peak speed exceeds 750 km/s, depicting
the high-speed stream from the southern coronal hole. On the other hand, the point
depicts the peak speed of 600 km/s from the northern coronal hole. This situation
is reversed during the fall months. These differences contribute to the winter and
summer solstice minima and the spring and fall maxima during each revolution of
the earth around the Sun; the reader may refer to Saito (1972) for more detail. The
differences of the geomagnetic disturbances in the spring and fall period can also
be understood in this way, although the difference of the development of coronal
holes in the two hemispheres contribute greatly to the difference in the spring and
fall. In spite of such a complexity, the two streams are the basic structure during
the late-declining period of the sunspot cycle.
The solar wind speed and IMF quantities at a distance of 1 au, computed on
the basis of the model, are developed by Hakamada and Akasofu (1982) and are
shown in Figure 13b. It is assumed that the IMF THETA (latitudinal) angle fluctu-
ates sinusoidally, simulating turbulent conditions of the solar wind; random noise
produces a similar result. The resulting function and the two geomagnetic indices
AE and Dst are also estimated. These simulated quantities may be compared with
Figure 13c, which shows the corresponding observed solar wind quantities and the
observed , AE, and Dst in 1974; note that the PHI angle is reversed in the model.
The agreement between the simulated and observed 27-day variations of the solar
wind quantities and two geomagnetic disturbances is quite reasonable. The assumed
velocity distribution as a function of distance from the neutral line (namely, the he-
liomagnetic latitude) is also in good agreement with the IPS observation by Kojima
et al. (2001).
The computed IMF field lines are shown in Figure 14. One can see the familiar
Parker spiral and the well-developed co-rotating IMF structure; the front of the

Figure 14. The computed IMF configuration from the Sun to a distance of 2 au.
52 S.-I. AKASOFU ET AL.

corotating structure is the shock front, which is formed as a faster wind advances
into a slower wind, although the shock does not develop fully at a distance of
1 au. This simple situation is quite different from flare-induced magnetic storms, in
which we have to consider effects of so-called magnetic clouds (Burlaga, 1988).
As is well known, the comparison between the observed 27-day variations
(Figure 13c) and their simulations (Figure 13b) indicates that recurrent magnetic
storms begin at the time of the arrival of the shock wave/sector boundary (the co-
rotating interplanetary structure) crossing, which causes a high IMF magnitude B
associated with the interplanetary co-rotating structure (Figure 13c) (note B2 in the
function) and turbulence of the solar wind. Each sinusoidal fluctuation generates
a substorm. Intense substorms cause a magnetic storm as their non-linear conse-
quences. This simulation study suggests that no other basic ingredient is needed
to cause recurrent magnetic storms, except for the changes of the relative location
of the earth with respect to the heliographic latitude and changes of the solar wind
speed distribution on the source surface.
When the neutral line has a single wave structure, it is expected from Figure 13a
that the two streams are separated by approximately 180 in longitude. In reality,
it can also be expected that the source regions of the two streams, which are also
separated by 180 in longitude, are not identical in terms of latitude or intensity.
Hakamada (1987) and many others inferred the solar wind speed on the source
surface based on the interplanetary scintillation (IPS). His figure (included here as
Figure 15) may be compared with Figure 13a. Thus, the solar wind speed, observed
at the earth or by spacecraft in interplanetary space, depends on its distribution

Figure 15. The solar wind speed distribution inferred from the interplanetary scintillation. The
distribution is projected onto the photosphere (Hakamada, 1987).
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 53

Figure 16. The schematic illustration of the two high-speed streams during the late-declining phase.

of the solar wind speed on source surface. We modeled only a simplest situa-
tion of the sinusoidal distribution. A great variety of situations can occur and the
prediction of geomagnetic disturbances depends greatly on many other secondary
factors.
The double stream situation is schematically shown in Figure 16. It shows
two cross-sections of the velocity pattern, one at +7.5 and the other 7.5 in
heliographic latitude. In this model, the width of the streams becomes larger at
higher latitudes in this simple model. It takes about 710 days for each stream to
pass by (or overtake) the earth and its magnetosphere. Thus, the magnetosphere is
engulfed by both streams for 1420 days out of 27 days (one solar rotation observed
by the earth). As we observed in our discussion of Figure 9a and b, it is for this
reason that the 27-day average speed of the solar wind becomes so high during the
late-declining phase of the sunspot cycle.
The above situation describes the simplest case. Single recurrent storm cases in
1953 and 1994 can be explained when the earth is located above the heliographic
equator in the fall months and the southern solar wind is weak or when the earth
is located below the heliographic equator in the spring months and the northern
solar wind is weak. In this way, one can explain the asymmetric C9 patterns in
Figure 11b.
Further, the solar wind distribution on the source surface is not as simple as
that shown in Figure 13a. There can exist some limited areas, from which higher
or slower solar wind can emanate. The actual 3-D structure of the streams may be
between the simulated case and a case like water flow from several hoses. During the
period of sunspot maximum, a large number of sunspots are expected to suppress
or deflect the solar wind flow, making the situation complicated.
The suggested relationship between the coarse photospheric magnetic field
structures and the source surface field in 1984 and 1994 is schematically shown in
54 S.-I. AKASOFU ET AL.

(a)

(b)

Figure 17. (a) The inferred vertical configuration of the magnetic field lines at latitude +10 to 10
based on the photospheric field distribution (in 1884). The field lines of individual sunspots or sunspot
pairs are not shown; for NESW longitude, see Section 5. (b) Ibid, except for 1994.

Figure 17a and b. These figures show both the photospheric field between +10
and 90 in latitude and the vertical field structure at +10 and 10 in latitude,
as well as the field on the source surface. These schematic sketchs are based on
a number of observed facts and the magnetic configurations that were discussed
earlier and are thus consistent with earlier literature.
It is expected that the closed magnetic field lines, at the location of the two dipolar
fields, stand high to the height of 2.5 solar radii and reach the source surface. It
is expected that the solar wind speed becomes low several days after the central
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 55

meridian passage of the two dipolar fields on the photosphere. As we show later in
Figure 20, this is indeed the case. The magnetic field lines from individual sunspots
and sunspot pairs lie low and closed. The other field lines reach the source surface,
becoming open field lines. It is likely that the solar wind originates along open
photospheric field lines that constitute a large bundle in the outer corona above the
source surface. It is obvious that this situation can occur where the polarity of the
root of these open-field lines and the polarity of the source surface are the same.
However, although all the field lines on the source surface are supposed to be open
field lines, the whole source surface is not necessarily the coronal hole. Coronal
holes reside within the open field regions, but do not necessarily occupy the entire
source surface; it should be noted that the observed coronal hole represents the
open field region below 1.5 solar radii, where a good contract of the brightness is
present.

5. Solar Phenomena in Heliomagnetic Longitude

After successfully demonstrating the importance of the neutral line on the source
surface, it can be suggested that the neutral line the magnetic equator can be a
new frame of reference in examining solar activities and geomagnetic disturbances.
However, one problem we face in this attempt is that the neutral line shifts gradually
or randomly in longitude in the Carrington coordinates during successive rotations.
The shift can be easily seen in data from 1984, 1994, 2002, and other years as
presented in Figure 1. Thus, we attempt to suppress the shift by introducing a new
heliomagnetic longitude for the neutral line based on the spherical harmonic
analysis of the source surface magnetic field by

11 
g
= arctan +
h 11
where g11 and h11 are the spherical harmonic coefficients, and;
= 0 for an even cycle, or
= 180 for an odd cycle.
As an example, Figure 18a shows the neutral line during 1993 and 1994 (sunspot
cycle 22) in the new coordinate system; the new longitude does not exactly coincide
with the Carrington longitude; this is why the Carrington rotation number used for
the new coordinate system has a decimal number. In the new coordinate system
thus defined, one can see that the sinusoidal wave (more like a rectangular wave)
is centered around 180 in longitude. Hereafter, the new coordinates are referred to
as the NESW coordinate system. The reason for this will become clear later in this
section. Figure 18b shows the typical neutral line during three sunspot cycles 20,
21, and 22. There are great similarities among them. The configuration of the cycle
20 (the only available data in the cycle) is a little different for those for the cycles
56 S.-I. AKASOFU ET AL.

21 and 22; as explained in Section 2, it shows the configuration toward the end of
the late-declining phase when one of the photospheric dipoles became weak.
Figure 18a shows also the location of frequent occurrence of solar flares; for
details, see Saito et al. (2003). One can easily see that one group of flares tends to
appear around 90 in longitude in the northern hemisphere (NE), while the other
group appears around 270 in longitude in the southern hemisphere (SW). The
exception is the Rotation 1875.54, in which a cluster of flares occurred around
180 in longitude. This case may be explained in terms of the double wave case that
is described in Section 6. Therefore, 99% of the total 78 flares were statistically
limited in either the NE quadrant in the northern hemisphere or the SW quadrant in
the southern hemisphere. Further, most of them occurred near the neutral line. In
order to confirm the above facts, we reproduce in Figure 18c the results obtained

(a)
Figure 18. (a) Neutral Line during a successive solar rotation during the late-declining phase in the
NESW coordinate system. (b) Comparison of the neutral lines during the late-declining phase of the
three sunspot cycles 20, 21, and 22 in the NESW coordinate system. (c) Locations of solar flares
during the declining phase of two sunspot cycles 21 and 22 in the NESW coordinate system. All the
flares in 1984 are compiled from the results obtained by Matsuura (1997), while those in 1994 are
superposed from Figure 18a.
(Continued on next page)
THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 57

(b)

(c)

Figure 18. (Continued)

58 S.-I. AKASOFU ET AL.

Figure 19. Coronal distribution during sunspot cycles 20, 21, and 22 in the NESW coordinate system.
Data from different instruments are used.

by Matsuura (1997), who plotted flares that occurred during 1984 in sunspot cycle
21. During the sunspot cycle 21, most flares occurred almost directly under the
neutral line, in spite of the fact that flares and the neutral line are not located on the
same spherical surface. During the sunspot cycle 22, the flares in the NE quadrant
is shifted a little away from the neutral line. This shift may be due to the fact that
the center of the magnetic arch or archade (Figure 3c) deviated from the radial line
from the center of the sun (Saito et al., 2000); see also Figure 17a and b. Here, we
deal with the phenomenon between the photosphere and the source surface, rather
than solar flares which occur in the photosphere/chromosphere.
Figure 19 shows the coronal distribution for the late-declining phase during the
sunspot cycle 20, 21, and 22. Note that different datasets were used for the different
cycles, depending on their availability; the satellite data were not available for the
cycle 20. The top figure is compiled from Figure 4 (Hakamada, 1987), so that
it represents the same period as that for Figure 8b, when one of the photospheric
dipoles became very weak. In the NESW coordinate system, a bright coronal region
was located in the SW quadrant. The middle figure shows the SMM coronagraph
(Hundhausen, HAO Homepage) in the NESW coordinate system. Although this
coronagraph has a low resolution in longitude, one can note that the brightness
distribution is related to the neutral line (see Figure 4). The bottom figure superposes
Yohkoh data for Carrington Rotations 18731886 (Shibata, Yohkoh Homepage).
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 59

Figure 20. Distribution of the speed of the solar wind in the NESW coordinate system. (Kojima et al.,
Homepage).

One can clearly see that the brightest regions are located in the NE quadrant and
the SW quadrant. The relative location of these regions with the neutral line can
be clearly seen (Watanabe et al., 2003). It is also clear that the northern coronal
hole is always located in the vicinity of 0 in longitude, while the southern one
is at about 180 in this new NESW longitude, suggesting the 180 separation (or
27/2-day separation) of a pair of recurrent storms (see Figures 10a, 13ac and 16).
Next, in Figure 20, we examine the distribution of the solar wind speed observed
by the interplanetary scintillation method (Kojima, IPS Homepage). This feature is
located above the source surface and the speed is projected on it. One can see that
two lowest speed regions (in red) occurred, one in the NE quadrant and the other
in the SW quadrant.
Thus, faster solar wind observed should be located between the two lowest speed
regions. A typical double peak case for both the solar wind speed and geomagnetic
disturbances is expected in 1984; in fact, it is clearly seen in Figures 10a and 12b.
In Section 2, we showed that the triple dipole model can reproduce reason-
ably well the configuration of the neutral line. It was shown there that the two
60 S.-I. AKASOFU ET AL.

photospheric dipoles are present in the magnetic field map in a coarse mesh (5 5 ).
These studies were made in the late 1980s and early 1990s, by several authors,
including the present authors. Unfortunately, the earlier studies did not have rele-
vant data at that time as we have now in this paper. In Section 5, by introducing
the NESW coordinate system, it was shown that the locations of solar flares, the
coronal distribution and the solar wind distribution can be fairly systematically
presented.
Thus, let us go back to the triple dipole model. Figure 21 shows the location
of the two hypothetical photospheric dipoles that can approximately reproduce
the configuration of the neutral line on the source surface. The location of the
two hypothetical dipoles on the photosphere is determined purely on the basis of
attempting to best reproduce the neutral line. The top figure in Figure 21 shows
the neutral line for the sunspot cycle 20. Unfortunately, this is the only neutral line
available for this cycle in the literature, which is the same as the neutral line in
Figures 3a, 4, and 8b. This occurred near the end of the late-declining phase, when
one of the two photospheric dipoles became very weak (the weak one was omitted
in Figure 8b).

Figure 21. The location of two hypothetical photospheric dipoles that can reproduce the neutral line
on the source surface (the NESW coordinate system).
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 61

Figure 22. Schematic illustration of the equatorial plane for the single wave and double wave cases.

Figure 23. A large amplitude double wave case. From the top, (a) the magnetic field on the source
surface, (b) soft X-ray corona (Yohkoh), (c) FeXIV coronagraph (SAC Peak) at the computed neutral
line based on the photospheric fields in the NESW coordinate system.

In the cycles 21 and 22, the hypothetical two dipoles are located approximately
where various solar activities are presented (Figure 19) and also where the lowest
solar wind speed regions are present (Figure 20). Therefore, it is reasonable to
conclude that two hypothetical dipoles are unmistakably realistic physical entities.
As mentioned earlier in Section 2, the two dipoles satisfy the HaleNicholson
polarity law, adding credibility of the presence of the two dipoles.
62 S.-I. AKASOFU ET AL.

Figure 24. Summary of Section 5.

6. Double Wave Case

The double wave of the neutral line can occur when the magnetic equatorial plane
warps in a particular way as shown in Figure 22. In Figure 23, the top diagram
shows a large amplitude double-wave situation which occurred in November, 1998
(see Figure 1). It is one of the most complicated configurations of the neutral line
during the sunspot cycle. The second and third diagrams show that solar activities
are not uniform and are located approximately where we can expect by considering
that there are two single waves in one NESW or Carrington coordinates, as shown
in the bottom diagram. There are four photospheric dipoles in the double wave
case, although these are not of equal intensity (Saito et al., 2002). They follow the
NESW rule and the HaleNicholson polarity law, so that they are not a random
feature.

7. Concluding Remarks

1. In this paper, we demonstrated that certain aspects of solar activities and their
consequences on the magnetosphere can be studied and understood better in
terms of the magnetic field distribution on the source surface than complicated
photospheric features. We first chose the late-declining phase during which
the neutral line has a single sinusoidal (or rectangular) structure, and the solar
wind speed and geomagnetic disturbances are highest during the sunspot cycle.
2. One of the important points we showed in this paper is that the configuration
of the neutral line on the source surface can tell us some important large-scale
and perhaps some fundamental features of the photosphere, which may not be
so obvious on the photosphere because of complexity of a variety of intense
small-scale features there.
3. Demanding that there must be a field represented by an axial dipole at the
center of the Sun (because the unipolar regions remain in both the northern
and southern polar regions throughout the sunspot cycle), we showed that
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 63

the deviation of the neutral line from the straight line along the heliographic
equator in the Carrington coordinates is caused by two hypothetical dipole
fields on the photosphere. As a first approximation, two large-scale dipoles are
needed to distort to the straight neutral line to become a sinusoidal wave-like
curve.
4. By introducing a new coordinate (NESW) system, it was shown that the loca-
tion of the two hypothetical dipoles coincide approximately with the nest of
the solar flares, the brightest region of corona and also the source of the lowest
solar wind speed. Thus, the two hypothetical photospheric dipoles are realistic
physical entities. Figure 24 summarizes all the results in Section 5.
5. The stable sinusoidal (or rectangular) wave configuration of the neutral line
provides the basis of the double peak of the solar wind speed and of geomag-
netic disturbances during the late-declining epoch of the sunspot cycle.
6. The double peak structure in the solar wind speed and recurrent geomagnetic
disturbances is a basic feature, but it is often obscured by various complexities
on the sun and the relative location of the earth with respect to the heliographic
equatorial plane.
7. Recurrent geomagnetic disturbances are related to the stable and sinusoidal (a
rectangular) neutral line structure. This is why recurrent geomagnetic distur-
bances tend to occur during the late-declining phase of the sunspot cycle and
why geomagnetic disturbances tend to occur at that time of the sunspot cycle.
8. The fact that the double high-speed solar wind stream is sometimes not appar-
ent may be due to the fact that a complex distribution of sunspots during the
period of sunspot maximum suppresses or deflects the high-speed streams.
9. As the intensity of the two dipoles becomes weak, the neutral line tends to
align with the heliographic equator. Since the intensity of the two dipoles is, in
general, not equal, one photospheric dipole, together with the central dipole,
may be enough to represent the neutral line.
10. The results of our study based on the single wave during the late-declining
phase of the sunspot cycle may be applicable to the double wave situation
and also the rest of the sunspot cycle. The classical sector structure may be
considered as the case when the amplitude of the single wave reaches 90 in
latitude.
11. Our results are based on the assumption that the Sun has a magnetic field which
can be represented by a central, axial dipole. Since we can explain a number
of phenomena based on this assumption, such a magnetic field is expected to
exist, except that it becomes very weak during the sunspot maximum phase,
represented by a central dipole.
12. However, the central dipole field does not make the meridional rotation during
the sunspot cycle. The apparent rotation on the source surface results from the
changes of the two photospheric dipoles.
13. It is hoped that the present study presents a new hint in understanding the
evolution of solar activities as a magnetic variable star during the sunspot
64 S.-I. AKASOFU ET AL.

cycle. The fact that two large-scale dipolar field regions are nests of solar
flares seems to suggest that flares are the process to reduce the imbalance of
the positive and negative fluxes (McIntosh, 1981).

Acknowledgments

The work represented here was supported in part by the National Institute of Polar
Research, Japan, and the Solar-Terrestrial Environment Laboratory, Nagoya Uni-
versity, Japan. A large number of colleagues contributed to this long-term project,
which began in the late 1980s, including N. Sato, H. Misawa, M. Kojima, T. Watan-
abe, K. Shibata, K. Tohmura, Y. Matsuura, and V.R. Dere.

References

Akasofu, S.-I.: 1981, Space Sci. Rev. 28, 121.

Akasofu, S.-I., and Chapman, S.: 1972, Solar Terrestrial Physics, Oxford at Clarendon Press, pp.
1901.
Akasofu, S.-I., and Fry, C. D.: 1986, J. Geophys. Res. 91, 13679.
Bartels, J.: 1932, Terr. Mag. 37, 1.
Bartels, J.: 1958, Planetarische Erdmagnetische Aktivitaet in Graphischer Darstellung, Vandenhoeck
& Ruprecher, Goettingen.
Burlaga, L. F.: 1988, J. Geophys. Res. 93, 7217.
Caan, M. N., McPherron, L., and Russell, C. T.: 1975, J. Geophys. Res. 80, 191.
Chapman, S., and Bartels, J.: 1940, Geomagnetism, Vol. I and II, 1-1049, Oxford at Clarendon Press.
Hakamada, K.: 1987, J. Geophys. Res. 92, 4339.
Hakamada, K., and Akasofu, S.-I.: 1982, Space Sci. Rev. 31, 3.
Hoeksema, J. T.: 1984, Structure and Evolution of the Large Scale Solar and Heliospheric Magnetic
Fields, Ph.D. Thesis, CSSA Rept. 7.
Hoeksema, J. T., WSO homepage: http://quake.stanford.edu/wso/coronal.html
Hundhausen, A. J.: 1993, J. Geophys. Res. 98, 13,177.
Hundhausen, A. J.: SMM homepage: http://hao.ucar.edu/d5/events/smm/syn/
Kojima, M.: IPS homepage : http://gedas.stelab.nagoya-u.ac.jp/sramp/jap/
Kojima, M., Fujiki, K., Ohmi, T., Tokumaru, M., and Yokobe, A.: 2001, J. Geophys. Res. 106(A8),
15,667.
Kojima, M., and Kakinuma, T.: 1987, J. Geophys. Res. 92, 7296.
Matsuura, Y.: 1997, Relationships Among Solar Active Regions, Coronal Holes And Recurrent-Type
Geomagnetic Storms, Master Thesis, Miyagi University of Education, pp. 160.
McComas, D. J., Elliott, H. A., Schwardon, N. A., Gosling, J. T., Skong, R. M., and Goldstein, B. E.:
2003, Geophys. Res. Lett. 30, 10010.
McIntosh, P. S.: 1981, The birth and evolution of sunspots: Observations, in L. E. Crain and Thomas,
J. H., (ed.), The Physics of Sunspots, Sacrament Peak Observatory, Sunspot, New Mexico, pp.
754.
Mustel, E. R.: 1964, Space Sci. Rev. 3(2), 139.
Newkirk, G., Jr.: 1971, Coronal Magnetic Fields, Physics of the Solar Corona, in C. J. Macris, (ed.),
D. Reidel, Dordrecht, Holland.
Pecker J., and Roberts, W. O.: 1955, J. Geophys. Res. 60, 33.
 THE LATE-DECLINING PHASE OF THE SUNSPOT CYCLE 65

Perreault, P., and Akasofu, S.-I.: 1978, J. Roy. Astron. Soc. 54, 547.
Saemundsson, Th.: 1962, Mon. Not. Roy. Astronom. Soc. 123, 299.
Saito, T.: 1965, Rep. Ionos. Space Res. Jpn. 19, 3046.
Saito, T.: 1972, Rep. Ionos. Space Res. Jpn. 26, 245266.
Saito, T.: 1975, Sci. Rep. Tohoku Univ. Ser. 5, Geophys. 23, 37.
Saito, T.: 1988, Solar Cycle Variation of Solar, Interplanetary, and Terrestrial Phenomena, Laboratory
and Space Plasmas, H. Kikuchi, (ed.), Springer-Verlag, New York, p. 473.
Saito, T., Akasofu, S.-I., Kozuka, Y., Takahashi, T., and Numazawa, S.: 1993, J. Geophys. Res. 98,
5639.
Saito, T., Kozuka, Y., Oki, T., and Akasofu, S.-I.: 1991, J. Geophys. Res. 96, 3807.
Saito, T., Oki, T., Akasofu, S.-I., and Olmstead, C.: 1989a, J. Geophys. Res. 94, 5453.
Saito, T., Oki, T., Olmstead, C., and Akasofu, S.-I.: 1989b, J. Geophys. Res. 94(A11), 14,993.
Saito, J., Sawa, M., Yoshimura, K., Masuda, S., and Kosugi, T.: 2003, Space Astronom. Sci. 1389.
Saito, T., Shibata, K., Dere, K. P., and Numazawa, S.: 2000, Adv. Space Res. 26(5), 807.
Saito, T., Watanabe, Ta, and Shibata, K.: 2002, Adv. Space Res. 29(10), 1533.
Schulz, M.: 1973, Astrophys. Space Sci. 24, 371383.
Smith, E. J.: 2001, J. Geophys. Res. 106, 15819.
Smith, E. J.: 2001, Geophys. Res. Lett. 28, 4159.
Shibata, K.: Yohkoh Homepage: http://www.solar.isas.ac.jp/
Svargaard, L., Wilcox, J. M., and Duvall, T. L.: 1974, Solar Physics 37, 157.
Watanabe, H., Akasofu, S.-I., and Saito, T.: 2003, Abstract Week A, A327.
Watanabe, H., Kojima, M., and Kozuka, Y.: 1996, Low speed solar winds from active regions, in
Y. Uchida et al., (eds.), Magnetodynamic Phenomena in the Solar Atmosphere Prototypes of
Stellar Magnetic Activity, Kluwer Academic Publishers, Netherlands, p. 501.
Wilcox, J. M., and Ness, N. F.: 1965, J. Geophys. Res. 70, 5793.
Wilcox, J. M., Severny, A., and Colburn, D. C.: 1969, Nature 224, 353.
Zhao, X. P. and Webb, D. F.: 2003, J. Geophys. Res. 108(A6), 1234.