Large-scale structure of the fast solar wind

Abstract

1] We present the results of a comprehensive study of the fast solar wind near solar minimum conditions using interplanetary scintillation (IPS) data taken with the EISCAT system in northern Scandinavia, and a recent extremely long baseline observation using both EISCAT and MERLIN systems. The results from IPS observations suggest that the fast wind inside 100 solar radii (R) can be represented by a two-mode model in some cases but this distinction is much less clear by in situ distances beyond 1 astronomical unit (215 R). Two distinct fast streams are seen in the extremely long baseline IPS observation; comparison of the IPS line of sight with a synoptic map of white light indicates the faster mode overlies the polar crown and the slower fast mode overlies an equatorial extension of the polar coronal hole.

Full text

Large-scale structure of the fast solar wind
M. M. Bisi,
1,3
R. A. Fallows,
1
A. R. Breen,
1
S. Rifai Habbal,
2
and R. A. Jones
1
Received 10 November 2006; revised 10 July 2006; accepted 13 March 2007; published 2 June 2007.
[1]We present the results of a comprehensive study of the fast solar wind near solar
minimum conditions using interplanetary scintillation (IPS) data taken with the EISCAT
system in northern Scandinavia, and a recent extremely long baseline observation using
both EISCAT and MERLIN systems. The results from IPS observations suggest that the
fast wind inside 100 solar radii (R

) can be represented by a two-mode model in some
cases but this distinction is much less clear by in situ distances beyond 1 astronomical unit
(215 R

). Two distinct fast streams are seen in the extremely long baseline IPS
observation; comparison of the IPS line of sight with a synoptic map of white light
indicates the faster mode overlies the polar crown and the slower fast mode overlies an
equatorial extension of the polar coronal hole.
Citation: Bisi, M. M., R. A. Fallows, A. R. Breen, S. Rifai Habbal, and R. A. Jones (2007), Large-scale structure of the fast solar
wind, J. Geophys. Res.,112 , A06101, doi:10.1029/2006JA012166.
1. Introduction
[2] Interplanetary scintillation (IPS) is the rapid variation
in the signal received by radio antennas at Earth from a
compact radio source arising from scattering by small-scale
(100 km) density variations in the solar wind. Measure-
ments of IPS allow the solar wind velocity to be inferred
over all heliographic latitudes and a wide range of helio-
centric distances [e.g., Dennison and Hewish, 1967]. When
two radio telescopes are used and the separation of the
raypaths in the plane of the sky from source to each
telescope is close to the radial direction centered at the
Sun, a high degree of correlation between the scintillation
patterns recorded at the two telescopes may be observed
[e.g., Armstrong and Coles, 1972; Coles, 1996]. The time
lag for maximum cross correlation can be used to estimate
the outflow speed of the irregularities producing the scin-
tillation. More sophisticated methods involving fitting the
results of a scattering model to the observed auto- and cross-
spectra may also be adopted [e.g., Coles et al., 1995; Coles,
1996; Klinglesmith, 1997].
[3] In weak scattering, the observed scintillation pattern
can be treated as arising from the sum of all the scattering
events along the raypath. If the raypath passes through two
or more streams of solar wind, their contributions are
independent and can be modelled separately provided that
their location along the raypath can be established. Studies
using the European Incoherent Scatter (EISCAT) IPS data
set have confirmed that the best results of fitting a weak
scattering model to the data [Coles, 1996; Klinglesmith,
1997] are obtained if regions of the raypath overlying dark
regions in white light observations are assumed to be
immersed in fast wind and slow streams are assumed to
overlie high intensity white light regions identified with
streamers [Coles, 1996]. Throughout these discussions we
define ‘‘polar crown flow’’ as the central region of the polar
fast streams (as seen in interplanetary space above the large
polar coronal holes outside of solar maximum) and ‘‘flank
flow’’ as the region of fast flow between this crown outflow
and the equatorward boundary of the fast stream.
[4] Ulysses observations from the first polar pass (1994 –
1996) revealed the existence of a latitudinal gradient in the
fast wind speed across the polar regions, with the highest
velocities measured at the highest latitudes [Phillips et al.,
1995; Goldstein et al., 1996; Woch et al., 1997; McComas et
al., 2000; Habbal and Woo, 2001]. The boundary between
the fast and slow outflow measured by Ulysses was located
at about ±20°around the heliographic equator [Woch et al.,
1997].
[5] The most commonly accepted view of the fast wind
places its origin within polar coronal holes, at the centre of
the supergranular cells [Dupree et al., 1996] or the bound-
aries of supergranular cells [Hassler et al., 1999]. This view
implies a subsequent superradial expansion of the bound-
aries of the polar outflow so that the fast wind can occupy a
significant volume of the heliosphere, as observed by
Ulysses at solar minimum. However, using white light,
radio ranging and ultraviolet spectroscopic observations in
the inner corona, Woo and Habbal [1997] and Habbal and
Woo [2001] suggested that the fast wind originates in the
quiet Sun as well as in the polar coronal holes, and that
the outflow is predominantly radial. In the latter scenario,
the fast solar wind from the quiet Sun wind would be
expected to be marginally slower and slightly denser than
the outflow from the polar coronal hole.
[6] In this paper we use IPS data from EISCAT [Rishbeth
and Williams, 1985; Wannberg et al., 2002] taken during the
last solar minimum (1994 to 1997) supplemented by in situ
measurements from the solar wind observations over the
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, A06101, doi:10.1029/2006JA012166, 2007
1
Institute of Mathematical and Physical Sciences, University of Wales,
Aberystwyth, GB, UK.
2
Institute for Astronomy, University of Hawaii, USA.
3
Now at Center for Astrophysics and Space Sciences, University of
California, San Diego, La Jolla, California, USA.
Copyright 2007 by the American Geophysical Union.
0148-0227/07/2006JA012166$09.00
A06101 1of9

poles of the sun (SWOOPS) [Bame et al., 1992] instrument
on Ulysses, observations from Large Angle Spectroscopic
Coronagraph (LASCO) [Brueckner et al., 1995] and Extreme
Ultraviolet Imaging Telescope (EIT) [Delaboudiniere et al.,
1995] instruments on Solar and Heliospheric Observatory
(SOHO), as well as images from the Mauna Loa Mk III
coronagraph [Fisher et al., 1981], and the Soft X-Ray
Telescope (SXT) [Acton et al., 1989] instrument on Yohkoh
[Ogawara, 1987; Ogawara et al., 1991] to investigate
whether there is evidence of a two-mode structure within
the fast solar wind, as originally proposed by Woo and
Habbal [1997]. In carrying out this investigation, two main
questions were considered: First, is the solar wind best
represented by a two-mode fast wind; and second, is the
boundary between two modes of fast wind related to the
coronal hole boundaries seen in EUV and X ray. We first
describe the data (section 2), and the data analysis tools
used to infer flow speeds (section 3). The findings are
discussed in section 4, followed by concluding remarks in
section 5.
2. Observations
[7]Woo and Habbal [1997] suggested that a small step in
plasma number density seen in the Ulysses data indicated
the presence of two modes of fast wind, with the fastest (and
least dense) emerging radially from the polar coronal hole
and a second, slightly slower and denser stream emerging
from the quiet Sun. An EISCAT observation of the radio
source J0741+271 above the north polar coronal hole on
11 July 1995 appeared to be better represented not by fast
and slow streams but by a two-mode fast wind, prompting
a study into this question. In this study it was important to
consider only low-noise, high-latitude data from fast-wind
dominated observations of simple, isolated radio sources.
To reduce ground noise in the antenna sidelobes, only data
from observations at elevation angles of around 10°or
more were considered. The scattering model assumes that
all the scintillation comes from a single point source.
Structured or multiple sources in the beam will give rise
to superpositions of scintillation patterns which, at present,
cannot be fitted in a manner which is not misleading. The
sources used are all primary calibrators at 21-cm wave-
length for the ‘‘A’’ configuration of the VLA. Thus they
have no decorrelation over 35 km but they could still have
a small effect on the IPS, for which the Fresnel Radius at
an observing wavelength of 32 cm is approximately 90 km.
[8] For the scintillation to be strong enough for the signal
spectrum to be clearly resolved but not so strong as to
approach the strong scattering regime in the fast solar wind,
only observations for which the point of closest approach of
the raypath to the Sun lay between 15 R

and 120 R

were
considered. In the past, this has proven to be the best range
when considering mid- to high-latitude fast solar wind
observations at this frequency. This eliminates possible
distortion of the results by the effects of strong scattering.
[9] In order to detect small variations in the fast solar
wind structure, it is necessary to eliminate observations
containing a significant amount of slow wind. This was
done using a two-stage selection process. In the first stage
any observation in which less than 30°of the raypath
around the point of closest approach to the Sun lay above
the coronal hole was eliminated (A detailed description of
how sections of the raypath are identified as lying above a
coronal hole is given in the next section). This provided a
quick way of eliminating observations containing a large
proportion of slow solar wind. The list of observations was
then refined by calculating the proportion of scintillation
being generated in the region of the raypath which overlay
the coronal hole. Only observations in which more than
85% of the scintillation came from above the coronal hole
were considered for the main study. The slow wind was
assumed to scintillate 3.5 times more than the fast wind in
this calculation: this is likely to be an overestimate, as
EISCAT observations suggest that the density variations in
the slow wind are normally 2 –3 times greater than those in
the fast wind over the distance range considered in this
study [Fallows et al., 2002], thus giving us considerable
confidence that we were indeed eliminating observations
which contained a significant proportion of slow wind.
[10] The majority of the IPS observations discussed in
this paper were made between 1994 and 1997 using the
EISCAT radio telescopes. The receiving frequency was on a
10 MHz bandwidth centered on 933.5 MHz from 1994 to
1995 and an 8 MHz bandwidth centered on 931.5 MHz
from 1996 to 1997. A single extremely long-baseline
observation combining data taken at 1420 MHz by EISCAT
and MERLIN telescopes in northern Scandinavia and the
UK, respectively [Bisi et al., 2005; Breen et al., 2006] is
discussed in section 4.3.
[11] The Ulysses data used in this paper are hourly
averaged SWOOPS ions radial velocity data taken during
the first polar pass of the Sun. The data were from the mid-
to high-latitude southern and northern polar passes where
the solar wind velocity did not dip below 600 km s
1
for
each hourly averaged data-point used, thus eliminating
interaction regions and periods of large variations in velocity
seen at the slow to fast wind transition in the Ulysses velocity
data.
3. Data Analysis
[12] A weak scattering interplanetary scintillation model
allowing for the presence of two solar wind streams in the
line of sight has been used for the last decade to analyze
EISCAT IPS observations [e.g., Coles, 1996; Klinglesmith,
1997; Canals, 2002]. The current analysis routines perform
all fitting to the auto- and cross-power spectra, with the
resulting correlation functions displayed for a more conve-
nient visualization.
[13] The power spectrum of the density variations in the
solar wind giving rise to IPS is described by a power law
with exponent a, cut off by an ‘‘inner scale’’ q
i
corresponding
to the dissipation scale of the turbulence. Previous studies
[e.g., Yamauchi et al., 1996] have found ato vary between
2 and 4 and the inner scale in kilometers to be roughly equal
to the distance of the observation in solar radii, although it is
not very well determined. We adjust these parameters to
match the high-frequency tail of the autopower spectrum,
within these boundaries. The density variations are assumed
to be anisotropic and normally extended in the radial
direction. The model assumes that a fast stream will be
present in the central portion of the line of sight, flanked by
a slow stream at either end. The slow stream is given an
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
2of9
A06101

extra weighting to account for its greater density and hence
greater scattering power. Following the work of Little and
Ekers [1971], each stream can be modelled as having a mean
speed with random velocity components in the radial (dV
k
:
indicated by a skewed cross-correlation function) and
transverse (dV
?
: indicated by a reduced area under the
cross-correlation function) directions.
[14] This work required the model to be extended to
allow for a third stream in the line of sight. This had to
be done in such a way that the total number of free
parameters was not substantially increased. To achieve this,
the velocity of a second fast stream was modelled in place
of the dV
k
of the main fast stream. The second fast stream
was assumed to have the same remaining properties (dV
?
and axial ratio) as the main fast stream and to occupy a band
above the mainstreamer belt (Figure 1). The second fast
stream was given a small additional weighting of 1.5 to
account for the step in Ulysses’ density values noted by Wo o
and Habbal [1997]. This results in a model which can either
assume a single fast stream with a random radial component
or assume two discrete fast streams with the second
occupying a variable band above the streamer belt: an
increase of only one free parameter. The model will also
allow the width of the second fast stream to be different on
each side of the line of sight but this is only for the purpose
of fixing the fast/faster stream boundary to the coronal hole
boundary seen in EUV and X ray for a direct comparison.
No fit is performed involving both widths independently.
[15] The region of the IPS raypath overlying the polar
coronal hole, and hence assumed to be immersed in the fast
stream, is estimated by ballistically projecting the raypath
down onto a white light map constructed from either Mauna
Loa MkIII data or, when available, from LASCO observa-
tions. Determination of the region of the raypath overlying
the X-ray coronal hole is carried out in similar fashion using
synoptic maps constructed from SXT and EIT measure-
ments. In the analysis routines the portion of the raypath
above the coronal hole is described in terms of two Sun-
centered angles along the raypath from the point of closest
approach, q
in
(Earth side of the line of sight) and q
out
(source side of the line of sight). These two angles are
illustrated in Figure 1. The assumption of a fast stream
flanked by slow flow is normally a valid one since the
geometry of the line of sight between Earth and the radio
source usually ensures that the tail ends lie above the
equatorial streamer belt. Furthermore, since the level of
scattering giving rise to IPS falls as 1/R
4
, a fast stream in the
tail of the line of sight will not have any effect on the
measurements in comparison to the much denser slow
stream. However, slow flow in the tails of the line of sight
will still have an effect.
[16] Figure 2 shows a model fit to one of the observa-
tions, radio source J1120+143 on 8 September 1995, used
in this paper. The cross-correlation function shows a skewed
main peak with a kink at longer time lags indicating the
definite presence of a second stream. Mauna Loa MkIII
observations indicate that 10–20°at the tail ends of the line
of sight lay above the streamer belt and so is assumed to be
in slow flow (Figure 3). The power law exponent awas set
to 3.0 (matching the high-frequency tail of the autopower
spectrum), the inner scale to 35 km (equal to the distance of
the point of closest approach) and the axial ratios of the fast
and slow streams to 2.3 and 1.5, respectively. Model fitting,
leaving the slow stream velocity set at 350 km s
1
, gave the
dominant fast stream a velocity of 872 km s
1
with a spread
of 193 km s
1
. However, if the slow velocity was also
fitted, it was found to rise to approximately 700 km s
1
,
fitting the kink in the cross-correlation function, but clearly
well above the velocity expected for solar wind flow above
the streamer belt. Hence for this fit, it was left fixed at 350
km s
1
, typical of slow wind speeds seen in IPS observa-
tions and by Ulysses.
[17] A fit using the new model is given for the 8 September
1995 observation in Figure 4. The model used the same
spectral parameters as before, but with two discrete fast
streams and a second stream width in place of the uniform
spread of velocities about the mean single fast stream.
Faster and fast velocities of 882 km s
1
and 621 km s
1
Figure 1. Diagram demonstrating the geometry of the new
model including two fast streams. The line of sight is
displayed as though it were mapped back down to the
corona. It should be noted that, although the second fast
stream width is fixed for both sides of the line of sight, each
tail end above the streamer belt may have a variable width.
Figure 2. The EISCAT IPS observation of J1120+143 on
8 September 1995, fitted with the two-stream weak scat-
tering model (dotted line). Following the method of analysis
described in the text, the power law exponent awas set to
3.0, the inner scale to 35 km and the axial ratios of the fast
and slow streams to 2.3 and 1.5, respectively. The dominant
fast stream has a velocity of 872 km s
1
with a random
radial component of 193 km s
1
. The slow stream has a
fixed velocity of 350 km s
1
.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
3of9
A06101

respectively fitted best with a fast stream width of 21°. The
model clearly fits the kink in the cross-correlation function.
The quality of fit to the data using each model is best judged
by comparing the reduced-chi-square (c
n
2
) parameter, as
described in Reiff [1983], for the fits. In this example the
original two-stream model (two free parameters) gave c
n
2
=
1.69 and the new three-stream model (three free parameters)
gave c
n
2
= 0.99, indicatinga significantimprovementin the fit.
[18] The IPS data presented in this paper are therefore
analyzed in the following way. Only the mean fast stream
velocity and associated dV
k
are fitted; all other parameters,
including those of the slow stream, are chosen such that
they best represent the data (in terms of the overall shape of
the correlation functions) but are not fitted. The results of
these analyses were taken to represent the null-hypothesis
result, in which the fast solar wind was best represented by a
single stream with some finite variation in flow speed. Next,
in order to determine whether the observations could be
better represented using a two-mode fast wind, we used the
three-stream model to fit two discrete fast velocities and a
width in the line of sight for the second stream. All other
parameters were left as in the two-stream fit. The quality of
fit, as given by the c
n
2
, combined with the values obtained in
each fit are used to judge whether the observations are
best described by two discrete fast streams. The width
obtained for the second fast stream is then used to compare
the boundaries of both fast streams with coronal hole
boundaries seen in SXT and EIT synoptic maps. However,
since this fit restricts the width of the second fast stream to
be equal on each side of the line of sight, a further analysis
is performed in which the width is set according to the EUV
and X-ray coronal hole boundary and only the faster and
fast stream velocities are fitted.
4. Results
[19] Both EISCAT IPS observations and SWOOPS data
were analyzed as part of this investigation into the structure
of the fast solar wind.
4.1. EISCAT IPS Data
[20] The main case results are summarized in Table 1. Of
the eight observations that met all of the criteria for a clear
fast dominated flow, five of them (with c
n
2
values high-
lighted in bold in the table) were better fitted using the dual-
fast wind model. The faster of the two streams was the
stream at the higher latitude of the two streams. In all the
remaining cases, neither of the model fits could be preferred
over any other.
[21] Only one case showed a best fit when the boundary
between fast streams agreed with the coronal hole boundary
seen in X-ray/EUV. In all the remaining cases, the model
fitting indicated a boundary much lower in latitude than that
seen in X-ray/EUV. However, these results do not indicate
any strong or systematic relationship between the fast
Figure 3. A white light map of the east limb of the Sun, Carrington rotations 1889 – 1900 centered on
8 September 1995, constructed using data taken from the Mauna Loa MkIII Coronagraph at 1.7 R

with
the IPS raypath mapped onto it and both SXT-5°latitude and MkIII coronal hole boundaries circled as
well as the P-point, the point of closest approach of the IPS raypath to the Sun in the plane of the sky.
Figure 4. The EISCAT IPS observation of J1120+143 on
8 September 1995, fitted with the three-stream weak
scattering model (dotted line). The spectral and slow stream
parameters are set as in Figure 2. The faster and fast
velocities are 895 and 661 km s
1
, respectively, with a fast
stream width of 18°.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
4of9
A06101

stream boundaries inferred from the IPS analyses and those
obtained from EUV/X-ray Carrington maps. We therefore
consider it important to compare these results with the one
and only available set of high latitude in situ solar wind
data recorded, the SWOOPS measurements made by the
Ulysses spacecraft while making its first polar pass at solar
minimum.
4.2. Ulysses Swoops in Situ Data
[22] The Ulysses SWOOPS ions hourly averaged data
were ‘‘binned’’ (averaged further) for approximately every
100 data points. In order to restrict the study to fast wind, a
low-latitude cut-off point was established at the lowest
latitude above which the solar wind velocity never dipped
below 600 km s
1
in any of the original hourly averaged
data points. The data were split into four different sections
corresponding to the increasing and decreasing latitude
parts of the northern and southern polar passes. Each section
was then fitted with polynomials of orders 1 to 3 via a least
squares fitting method. A bimodal linear fit (i.e., straight
line with a change in gradient) was also performed with the
breakpoint between the two linear fits set to a range of
latitudes from 75°N/S to 10°from the low-latitude cut-off.
[23] Using the c
n
2
value as a measure of the goodness of
fit, the data appeared to be best represented by the bilinear
fit in three out of the four segments and a linear fit in one
out of the four segments as shown in Tables 2 and 3, and in
Figure 5. It should be noted that in those cases where a
bilinear fit gave the lowest c
n
2
value closest to 1 in Table 2,
the improvement of fit gained was generally very small. It
should also be noted that in the increasing latitude northern
pole case, the c
n
2
value is always overfitted which is why the
straight-line linear fit is taken as the most appropriate in this
case. Table 3 gives details of the best least squares fit in
each case, along with the latitudes of the polar coronal holes
as seen in SXT and EIT.
[24] There was no clear relationship between the latitude
of the inflexion point of the bilinear fits and the latitude of
the coronal hole boundary seen in X-ray or EUV emission
during the same Carrington rotation. The only clearly
significant result indicated a single fast wind with higher
velocities at higher latitudes, which is in good agreement
with the results of McComas et al. [2000].
[25] Figure 6 shows a rough comparison of the two-mode
fast IPS best-fitted data and ion SWOOPS data, bearing in
mind that the observations were taken at different times and
so cannot be taken as a direct comparison. However, they
can be used to get the general picture of the two different
data sets. To save overcrowding on the plot, only the most
significantly fitted Ulysses data from each hemisphere are
used. The fast velocities and their associated spread (dV
k
)
from the traditional two-stream model fits are used as an
illustration of the range of velocities given by the IPS
modeling. dV
k
was not modelled in the two-mode fast wind
model fitting and so no spread in velocity can be inferred
from these results. Instead, they are used to illustrate the
range of latitudes covered by each fast stream, as
determined by the fast/slow wind boundaries and the fast/
faster wind boundaries.
4.3. Extremely Long Baseline Observation
[26] Recent developments in the IPS observing technique
have included the use of much more widely separated
antennas with separations up to 2000 km in the radial
baseline, carried out with EISCAT and the Multi-Element
Radio Linked Interferometer Network (MERLIN) based in
Table 1. Summary of IPS Results
Date Source V
fast
dV
k
V
mid
Wdth
in
Wdth
out
rmsVV
slow
q
in
q
out
wtslow AR aq
i
c
n2
14 July 1995 J0741+312 754 444 80 300 65°70 8 2.0 3.0 36 2.96
842 443 31°31°1.72
886 456 35°40°2.07
6 September 1995 J1120+143 811 211 0 450 70°60°3 2.3 3.0 35 5.83
844 548 23°23°3.57
931 755 50°40°6.03
7 September 1995 J1120+143 872 210 0 450 70°60°2 2.0 3.0 35 3.70
895 708 34°34°2.90
929 794 45°35°3.15
8 September 1995 J1120+143 872 193 0 350 70°60°2 2.3 3.0 35 1.69
882 621 21°21°0.99
951 831 50°35°1.70
29 May 1996 J0336+323 827 146 0 350 75°75°8 1.5 3.4 57 1.68
842 738 42°42°1.47
901 827 65°45°1.64
8 June 1996 J0521+166 820 126 0 350 65°80°8 1.6 3.2 26 6.59
867 770 51°51°6.22
856 826 55°30°6.49
9 June 1996 J0521+166 774 215 0 350 70°80°4 1.5 3.1 25 3.90
875 708 54°54°3.05
878 706 55°55°2.99
17 August 1996 J0954+177 827 408 0 450 60°60°10 3.5 2.5 8 1.14
934 609 26°26°1.04
968 644 35°35°1.15
Table 2. Summary of SWOOPS Ions Analysis
a
Type/Order of fit
South
increasing
lat.
South
decreasing
lat.
North
increasing
lat.
North
decreasing
lat.
Straight line 2.77866 1.00748 0.48211 1.13374
Two-stage gradient 2.36277 1.00580 0.46569 1.06449
X
2
parabola 2.42265 1.02292 0.47697 1.10716
X
3
parabola 2.39792 1.05332 0.49470 1.11038
a
The lowest c
n2
value closest to 1 are in bold.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
5of9
A06101

the United Kingdom (this experiment is described further by
Breen et al. [2006]). The difference between time lags for
solar wind speed increases as the antenna separation gets
larger. An observation of J0319+415 from 12 May 2004
produced a correlation function that showed two clear fast
wind peaks [Bisi et al., 2005; Breen et al., 2006].
[27] This observation strongly suggested the presence of
two distinct fast components which could be interpreted in
two ways:
[28] 1. As a region of faster flow from above the polar
coronal hole at the crown and fast flow above the quiet Sun; or
Table 3. Summary of SWOOPS Ions Best Fit Analysis
Variable South increasing lat. South decreasing lat. North increasing lat. North decreasing lat.
Latitude of change in gradient (52.95 +2.58/1.50)°(72.48 +0.48/0.93)°Straight line (+31.94 +1.95/1.04)°
Date of change in gradient 16 February 1994, 46.60 30 October 1994, 302.23 N/A 11 July 1996, 192.48
Carrington rotation during
change in gradient
1879 1888 N/A 1911
Days of year for the
Carrington rotation
1994-37.96-65.29 1994-283.24-310.53 N/A 1996-180.70-207.90
Latitude of SXT coronal hole 60 ± 10°50 ± 10°50 ± 10°50 ± 10°
Latitude of EIT coronal hole Not available Not available Not available 60 ± 10°
c
n2
of fit 2.362770 1.005800 0.482111 1.064490
Nu 94 26 27 89
Significance Level 0.001 0.450 >0.990 0.300
Summary No clear fit Weak evidence for a
bi-modal fast wind
Strong evidence for a
single fast wind
Weak evidence for a
bi-modal fast wind
Figure 5. Plots of averaged Ulysses ions radial velocity for the first polar pass. The two lines represent
the ‘‘best fit’’ determined by the c
n2
fit to the binned data showing that a change in gradient of velocity
with latitude fits the data best. The arrow indicates the inflexion point between the two lines. The latitude
and c
n2
fit values can be seen in Table 3. (a) South polar pass increasing in latitude; July 1993 to
September 1994. (b) South polar pass decreasing in latitude; September 1994 to January 1995. (c) North
polar pass increasing in latitude; March 1995 to July 1995. (d) North polar pass decreasing in latitude;
August 1995 to August 1996.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
6of9
A06101

[29] 2. As faster flow from above the whole of the white
light coronal hole and fast flow above the equatorward
extension of the polar coronal hole.
[30] The outflow from equatorward extensions of coronal
holes is generally accepted to be slower than that from large
polar holes [e.g., Kojima et al., 2004]. Figure 7 shows the
line of sight projected back along the Parker spiral (assum-
ing radial outflow) and down to the solar corona at 2.5 R

and overlaying a synoptic map of white light. The geometry
of the observation does suggest that the Earthward side of
Figure 6. Figure showing the IPS fitted velocity values as a rough comparison to the values of the two
most significantly fitted Ulysses SWOOPS ions radial velocity binned data with latitude (south
decreasing latitude and north increasing latitude). Two-stream modelled velocities (blue) are included
simply as a measure of the range of velocities given by the IPS modeling, using half dV
k
as the error; the
latitude given is that of the point of closest approach. The two-mode fast wind modelled fast (red) and
faster (green) velocities used are from the best fit from each observation using errors quoted from the
least squares fitting routine. The latitude ranges given are derived from the average latitudes of the fast/
faster wind boundaries and the fast/slow wind boundaries.
Figure 7. Synoptic map of LASCO east limb white light data from Carrington rotation 2016, overlayed
by the IPS line of sight to J0319+415. The arrows indicate the positions of the fast/slow wind boundaries
(as inferred from this image) and the fast/faster wind boundary as fitted using the two-mode fast wind
model.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
7of9
A06101

the line of sight lies above an equatorward extension of the
polar coronal hole. The transition between flow above a
large polar hole and that above an equatorward extension
would be fairly abrupt and lead to the appearance of two
discrete fast components of flow.
[31] To account for this scenario, a further modification
was made to the analysis program to allow the width of the
fast stream to be fitted on only one side of the line of sight.
The IPS data were analyzed using the two-mode fast wind
model (Figure 8) and the results are unambiguous: It is only
possible to reproduce the shape of the cross-correlation
function if the fast stream occupies part of the line of sight
over the equatorward extension of the coronal hole. No
other scenario would reproduce the correct shape. The
stream boundaries in the IPS line of sight, as found in this
analysis, are also marked in Figure 7.
[32] This result provides convincing evidence of a greater
degree of nonuniformity in the fast solar wind than is seen
in the Ulysses|SWOOPS data.
5. Discussion
[33] The results from IPS and in situ measurements
reported in this paper provide evidence for a change in
character of the solar wind between the polar crown and the
equatorward flanks of the fast wind, but IPS results suggest
a two-mode fast solar wind more strongly than do the in situ
data.
[34] The polar fast wind is faster and appears to vary little
with latitude whereas the wind above the lower latitude
regions of the white light coronal hole shows a more
marked decrease in velocity with decreasing latitude. These
results are generally consistent with the Woo and Habbal
proposals. However, the Ulysses results suggest a more
continuous distribution of solar wind speeds rather than two
distinct modes.
[35] Although the majority of the IPS results indicate that
a two-mode fast wind model is a better fit to the measure-
ments, the c
n
2
values are high in many cases indicating that
the model still does not adequately describe the data. A
possible explanation for this is the steady variation in
velocity with latitude of the flanking fast stream, as seen
in the Ulysses data at lower latitudes, which has not been
included in the two-mode fast wind IPS model. The fact that
some observations do not fit significantly better with the
two-mode fast wind model also suggests this explanation.
[36] The boundary between fast and faster streams in the
IPS lines of sight modelled in the IPS analysis showed no
correlation with the coronal hole and quiet Sun boundary
seen in EUV and X-ray measurements. This suggests that
either there is no distinction between the fast wind emanat-
ing from the quiet Sun and the EUV/X-ray coronal hole, or
that the individual fast wind components have interacted to
a large enough extent for the boundary to be significantly
blurred by IPS distances.
[37] The extremely long baseline observation of 12 May
2004 does suggest two distinct modes of fast wind. How-
ever, in this case, the raypath extended from above the polar
hole into a region above an equatorward extension of the
northern hole. Analysis unambiguously indicated that the
slower fast stream lay above the equatorward extension of
the coronal hole. We therefore propose that the observation
is dominated primarily by high-latitude, low-density flow
from across the raypath near the point of closest approach,
and secondly, by slightly slower higher density flow above
the equatorward extension of the coronal hole.
6. Conclusions
[38] The results presented in this paper provide a useful
test of the alternative models of fast wind internal structure.
They reveal a difference in character between the flow
found in interplanetary space above the polar crown and
above the flanks of the white light coronal hole, though
probably not a sudden transition from one type of flow to
another. Within the fast solar wind the latitudinal gradient in
velocity in the polar crown stream is considerably shallower
than that seen in the equatorward flanks of the fast stream.
However, the Ulysses data do not provide significant
evidence of a bimodal fast wind, though it could be the
case that there is an inhomogeneity in the turbulence that
damps out between IPS distances (typically 50 R

) and
the typical distance of Ulysses (over 10 times that). It can
only suggest that there is a difference in character between
the polar stream and the flanks of the fast wind.
[39] We suggest that the flow observed above the centers
of the polar coronal holes should be regarded as the
archetypal fast wind. The change in latitudinal gradient in
velocity observed in interplanetary space at latitudes near
to those of the boundaries of the X-ray coronal holes may
indicate an internal boundary within the fast wind or may
simply be the signature of a more gradual change in solar
wind properties. The presence of two distinct fast velocities
in the EISCAT-MERLIN result strongly suggests a differ-
ence in character between the high-latitude fast wind
(found above the polar coronal hole) and the flow above
the equatorward extension of the polar hole. It is also likely
that this difference applies to flow above the centers of
polar coronal holes and regions near their equatorward
boundaries.
Figure 8. The EISCAT/MERLIN observation of J0319+415
on 12 May 2004, fitted with the two-mode fast wind
weak scattering model. Parameters are: V
faster
766 km s
1
;
V
fast
640 km s
1
; axial ratio 2.0; q
in
55°;q
out
55°;V
slow
350 km s
1
; weighting 5.3; a3.0; inner scale 85.0; fast
stream width 40°(Earth end of the line of sight only). The
c
n2
is 0.51.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
8of9
A06101

[40]Acknowledgments. We would like to thank the directors and
staff of EISCAT and MERLIN for the IPS data used in this paper and for the
supply of equipment while using these facilities. EISCAT is supported by
the scientific research councils of Finland, France, Germany, Japan, Nor-
way, Sweden, and the United Kingdom. The MERLIN system is supported
by PPARC, as were several of the authors of this paper (RAF, MMB, and
RAJ) during the time this work was carried out. The programs used to
analyze the IPS data are based on routines developed by W. A. Coles and
B. J. Rickett at the University of California, San Diego and we are
extremely grateful not only for the programs but also for much valuable
advice. LASCO data are used courtesy of the LASCO consortium and were
made available by the EIT consortium. We would also like to thank D. J.
McComas and the SWOOPS team for the solar wind plasma data,
A. Lecinski for the MkIII coronagraph white light data, and the Yohkoh
consortium for the SXT data.
[41]Amitava Bhattacharjee thanks W. A. Coles and another reviewer
for their assistance in evaluating this paper.
References
Acton, L. W., M. Bruner, W. Brown, J. Lemen, T. Hirayama, S. Tsuneta,
T. Watanabe, and Y. Ogawara (1989), The solar-A soft X-ray telescope
experiment, Adv. Space Res.,8(11), 93 – 99.
Armstrong, J. W., and W. A. Coles (1972), Analysis of three-station inter-
planetary scintillation data, J. Geophys. Res.,77, 4602 – 4610.
Bame, S. J., D. J. McComas, B. L. Barraclough, J. L. Phillips, K. J. Sofaly,
J. C. Chavez, B. E. Goldstein, and R. K. Sakurai (1992), The Ulysses
solar wind plasma experiment, Astron. Astrophys. Suppl.,92(2), 237.
Bisi, M. M., A. R. Breen, R. A. Fallows, P. Thomasson, R. A. Jones, and
G. Wannberg (2005), Combined EISCAT/ESR/MERLIN interplanetary
scintillation observations of the solar wind, in Solar Wind 11/SOHO 16,
Connecting Sun and Heliosphere, edited by B. Fleck, T. H. Zurbuchen,
and H. Lacoste, Eur. Space Agency, ESA SP-592, 593 – 596.
Breen, A. R., R. A. Fallows, M. M. Bisi, P. Thomasson, C. A. Jordan,
G. Wannberg, and R. A. Jones (2006), Extremely long-baseline interpla-
netary scintillation measurements of solar wind velocity, J. Geophys.
Res.,111, A08104, doi:10.1029/2005JA011485.
Brueckner, G. E., et al. (1995), The large angle spectroscopic coronagraph
(LASCO), Sol. Phys.,162, 357 – 402.
Canals, A. (2002), Interplanetary scintillation studies of the solar wind
during the rising phase of the solar cycle, Ph.D. thesis, The University
of Wales, Aberystwyth.
Coles, W. A. (1996), A bimodal model of the solar wind speed, Astrophys.
Space Sci.,243(1), 87 – 96.
Coles, W. A., R. R. Grall, T. Klinglesmith, and G. Bourgois (1995), Solar
cycle changes in the level of compressive mircoturbulance near the sun,
Geophys. Res., 100.
Delaboudiniere, J. P., et al. (1995), EIT: Extreme-Ultraviolet Imaging Tele-
scope for the SOHO Mission, Sol. Phys.,162, 291 – 312.
Dennison, P. A., and A. Hewish (1967), The solar wind outside the plane of
the ecliptic, Nature,213, 343 – 346.
Dupree, A. K., M. J. Penn, and H. P. Jones (1996), He i 10830 Angstrom
wing asymmetry in polar coronal holes: Evidence for radial outflows,
Astrophys. J.,467, L121+.
Fallows, R. A., P. J. S. Williams, and A. R. Breen (2002), EISCAT mea-
surements of solar wind velocity and the associated level of interplanetary
scintillation, Ann. Geophys.,20, 1279 – 1289.
Fisher, R. R., R. H. Lee, R. M. MacQueen, and A. I. Poland (1981), New
Mauna Loa coronagraph systems, Appl. Opt.,20, 1094 – 1101.
Goldstein, B. E., M. Neugebauer, J. L. Phillips, S. Bame, J. T. Gosling,
D. McComas, Y.-M. Wang, N. R. Sheeley, and S. T. Suess (1996),
ULYSSES plasma parameters: Latitudinal, radial, and temporal varia-
tions, Astron. Astrophys.,316, 296 – 303.
Habbal, S. R., and R. Woo (2001), Connecting the Sun and the solar wind:
Comparison of the latitudinal profiles of coronal and Ulysses measure-
ments of the fast wind, Astrophys. J.,549, L253 – L256.
Hassler, D. M., I. E. Dammasch, P. Lemaire, P. Brekke, W. Curdt, H. E.
Mason, J. C. Vial, and K. Wilhelm (1999), Solar wind outflow and the
chromospheric magnetic network, Science,283, 810.
Klinglesmith, M. (1997), The polar solar wind from 2.5 to 40 solar radii:
Results of intensity scintillation measurements, Ph.D. thesis, University
of California, San Diego.
Kojima, M., A. R. Breen, K. Fujiki, K. Hayashi, T. Ohmi, and M. Tokumaru
(2004), Fast solar wind after the rapid acceleration, J. Geophys. Res.,
109(A18), A04103, doi:10.1029/2003JA010247.
Little, L. T., and R. D. Ekers (1971), A method for analysing drifting
random patterns in astronomy and geophysics, Astron. Astrophys.,10,
306 – 309.
McComas, D. J., B. L. Barraclough, H. O. Funsten, J. T. Gosling,
E. Santiago-Munoz, R. M. Skoug, B. E. Goldstein, M. Neugebauer,
P. Riley, and A. Balogh (2000), Solar wind observations over Ulysses’
first polar orbit, J. Geophys. Res.,105(A5), 10,419 – 10,433.
Ogawara, Y. (1987), The SOLAR-A mission, Sol. Phys.,113 , 361 – 370.
Ogawara, Y., T. Takano, T. Kato, T. Kosugi, S. Tsunta, T. Watanabe,
I. Kondo, and Y. Uchida (1991), The Solar-A Mission: An overview,
Solar Physics, 136, 1 – 16, Springer, New York.
Phillips, J. L., S. J. Bame, W. C. Feldman, B. E. Goldstein, J. T. Gosling,
C. M. Hammond, D. J. McComas, M. Neugebauer, E. E. Scime, and
S. T. Suess (1995), ULYSSES solar wind plasma observations at high
southerly latitudes, Science,268, 1030±.
Reiff, P. H. (1983), The use and misuse of statistical analyses, in ASSL Vol
104: Solar-Terrestrial Physics: Principles and Theoretical Foundations,
p. 493, Springer, New York.
Rishbeth, H., and P. J. S. Williams (1985), Ionospheric radar: The system
and its early results, Mon. Not. R. Astron. Soc.,26, 478 – 512.
Wannberg, G., L.-G. Vanhainen, A. Westman, A. R. Breen, and P. J. S.
Williams (2002), The new 1420 MHz dual polarisation interplanetary
scintillation (IPS) facility at EISCAT, in Conference Proceedings, Union
of Radio Scientists (URSI).
Woch, J., W. I. Axford, U. Mall, B. Wilken, S. Livi, J. Geiss, G. Gloeckler,
and R. J. Forsyth (1997), SWICS/ULYSSES observations: The three-
dimensional structure of the heliosphere during solar minimum, Geophys.
Res. Lett.,24, 2885 – 2888.
Woo, R., and S. R. Habbal (1997), Extension of coronal structure into
interplanetary space, Geophys. Res. Lett.,24(10), 1159 – 1162.
Yamauchi, Y., M. Kojima, M. Tokumaru, H. Misawa, H. Mori, T. Tanaka,
H. Takaba, T. Kondo, and P. K. Manoharan (1996), Micro-turbulence in
the solar wind at 5–76 rs observed with interplanetary scintillation,
J. Geomagn. Geoelectr.,48, 1201 – 1217.

M. M. Bisi, Center for Astrophysics and Space Sciences, University of
California, San Diego, 9500 Gilman Dr., La Jolla, CA 92093, USA.
(mmbisi@ucsd.edu)
A. R. Breen, R. A. Fallows, and R. A. Jones, Institute of Mathematical
and Physical Sciences, University of Wales, Physical Sciences Building,
Penglais Hill, Aberystwyth, Ceredigion SY23 3BZ, UK. (raf@aber.ac.uk)
S. Rifai Habbal, Institute for Astronomy, University of Hawaii, USA.
A06101 BISI ET AL.: LARGE-SCALE STRUCTURE OF THE FAST SOLAR WIND
9of9
A06101