INTERVALS OF RADIAL INTERPLANETARY MAGNETIC FIELDS AT 1 AU, THEIR ASSOCIATION WITH RAREFACTION REGIONS, AND THEIR APPARENT MAGNETIC FOOT POINTS AT THE SUN

Neugebauer & Goldstein (1997) and Neugebauer et al. (1997) studied observations from the International Sun/Earth Explorer 3 (ISEE-3) and first recognized periods of nearly radial interplanetary magnetic field (IMF) in the interplanetary medium that extended for 6 hr and longer. They define "nearly radial" as |B_R/B| > 0.90 where B_R is the radial component and B is the magnitude of the measured IMF. The traditional view of the IMF direction (Parker 1958, 1963), wherein the rapid acceleration of the solar wind produces a radial magnetic field above the acceleration region which is then wound by solar rotation to produce a spiral in interplanetary space, would seem to render radial fields at 1 AU and beyond inaccessible. While these intervals of radial fields at 1 AU were initially associated with trailing regions behind interplanetary coronal mass ejections (ICMEs), they later came to be associated more generally with periods of decreasing wind speed known as rarefaction regions. Jones et al. (1998) summarized Ulysses observations of similar events lasting 6 hr or longer and found them almost absent from the high-latitude (|Θ_lat| > 40°) observations. Wang et al. (2003) extended the analysis to 10 AU using Voyager data and provides some controversy regarding whether the duration of these events increases with heliocentric distance and whether they can occur outside of regions of declining wind speed.

Rarefaction regions represent a well-established dynamic in the solar wind where fast moving plasma (fast wind) moves ahead of slow moving plasma (slow wind) to create a relative void between the two (Gosling et al. 1972). They are common during solar maximum in so far as falling wind speeds and densities form a counterpart to many fast-moving transient observations (for instance the regions behind coronal mass ejections, CMEs, generally show a decreasing wind speed) and they form the counterpart to the corotating interaction region in solar minimum which is a recurring stream interface where fast wind pushes into slow wind from behind, creating a region of compression. In these traditional rarefaction regions the IMF remains within the stream that originates at the foot point of the field. Thus, the IMF adheres to the Parker (1958, 1963) prediction for the winding of the IMF. When Neugebauer & Goldstein (1997) originally identified the region behind CMEs as the location of the radial IMF intervals they suggested disconnection of the IMF foot point from the Sun as a possible explanation for the non-Parker field orientation, but Neugebauer et al. (1997) dismisses this interpretation along with field line draping (McComas et al. 1998b). Jones et al. (1998) appears to maintain the foot point disconnection theory.

So the question becomes how to produce an IMF that is aligned with the velocity gradient? Such a gradient is most likely generated by a time-dependent event of some kind at the foot point of the magnetic field such as caused by reconnection (Gosling & Skoug 2002) or by a transient evolution of the solar wind boundary in the solar atmosphere (Wang 1994; Sheeley et al. 2007). Gosling & Skoug (2002), Schwadron (2002), Murphy et al. (2002), Schwadron & McComas (2005), and Riley & Gosling (2007) have provided several such explanations and while offering some varied opinions as to the mechanism all cite a need for an abruptly changing wind speed at the foot point of the magnetic field line. See Figure 5 of Riley & Gosling (2007) for a demonstration of how this idea leads to a radial IMF at 1 AU. Interchange reconnection at the foot points of the IMF, which is magnetic reconnection between an open field line and a loop, can provide the rapid change in solar wind speed at the source of the IMF that is necessary to produce an IMF that is aligned with the velocity gradient. By having the foot point of the IMF first rooted in a fast wind source and then through whatever means move abruptly to a slow wind source, the IMF is forced to cross the rarefaction region and magnetically join the fast and slow winds in interplanetary space (Schwadron 2002; Murphy et al. 2002; Schwadron & McComas 2005; Schwadron et al. 2005). The changing wind speed in association with the source of the IMF can be accomplished either by a rapid change in the wind produced at the base of the IMF or by moving the base of the IMF from one wind source to another. The latter is generally discussed in terms of interchange reconnection. Either way, the resulting expansion draws the IMF into a radial configuration within the rarefaction region.

The goal of this study is to identify the likely source regions that form the radial IMF intervals seen at 1 AU. We have searched the Advanced Composition Explorer (ACE) data (Stone et al. 1998; Garrard et al. 1998) from late 1997 into early 2012 to find extended intervals when the IMF remains nearly radial. We perform a series of superposed epoch analyses that reveal both the mean behavior of solar wind parameters before and after the radial IMF intervals as well as the underlying variances. We then correlate these intervals with the solar wind speed, density and flux before, during, and after the radial intervals to show that they fall consistently within rarefaction intervals. We likewise correlate the thermal ion composition before, during and after the radial events to show that the rarefaction intervals in question and the radial IMF periods within them are formed primarily by the interaction of two "slow wind" sources with polar coronal hole sources playing only a minor role. Our findings are consistent with past interpretations that sudden changes in solar wind parameters at the foot points of the magnetic fields lead to intervals of radial IMF at 1 AU by spanning the resultant rarefaction regions, but point to two wind sources that are both nominally slow wind sources and seldom involve the magnetic fields of polar coronal holes. We argue that this is a reasonable result for low-latitude observations and does not conflict with earlier conclusions involving polar coronal holes where measurements were recorded at high heliographic latitudes.

The ACE/MAG and ACE/SWEPAM data (Smith et al. 1998; McComas et al. 1998a) recorded from late 1997 until early 2012 were used here by first averaging the measurements down to 1 hr resolution so that short-term excursions of the IMF are not considered. The analysis is performed in the familiar (R, T, N) coordinate system used by Parker (1963) to define the mean interplanetary magnetic field. Here, R is the unit vector in the radial direction from Sun to observer, T lies in the plane of solar rotation and is directed in the sense of rotation, and N = R × T completes the right-handed coordinate system. We wrote a code that scans the hourly averages to identify prolonged intervals of nearly radial IMF according to the same definition used in the earlier studies (Neugebauer & Goldstein 1997; Neugebauer et al. 1997; Jones et al. 1998; Wang et al. 2003): |B_R/B| > 0.90 where B_R is the radial component and B is the magnitude of the measured IMF. We will simply call these intervals of "radial IMF." For comparison, the normal Parker spiral at 1 AU has the angle between the IMF and the radial direction Θ_BR ∼ 45° ± 10° which yields |B_R/B| ∼ 0.71. Figure 1 shows the distribution of event durations obtained by that search. We found 226 intervals when the radial IMF persists for at least 6 hr and these intervals form the basis for our study. Periods of radial IMF shorter than 6 hr were obtained, but only intervals of radial IMF longer than 6 hr are used in this study. Most events are less than 10 hr in duration, although a few extend for as long as 30 hr.

Zoom In Zoom Out Reset image size

Figure 2 shows a typical event used in this study. There is a 5 day period of slowing solar wind speed forming a rarefaction region from day 27 to 32. The IMF intensity is constant, but the direction is not. From 29.375 to 29.875 hourly averages of the IMF are consistently in the range |B_R/B| > 0.90. This is marked by the horizontal bar between the top two panels. The nearest other events are 54 days prior and 72 days after this event. This event also demonstrates that data coverage is not perfect for all data sets and all events. The SWEPAM measurement of proton density is lost during the latter half of the interval. This in no way significantly impacts the conclusions of the paper.

Zoom In Zoom Out Reset image size

Figure 3 shows another example of radial IMF that points to a limitation in our analysis. In this 6 day period there are three intervals where |B_R/B| > 0.90 for more than 6 hr. They appear to be associated with the same rarefaction interval due to their close proximity and the failure of the wind to establish a significant duration of fast wind between them; or it may be a compound interval as it seems to show two different and clearly recognizable slopes in wind speed breaking at the start of day 91. This does raise the question of whether and how many of our events are distinct and independent. Remember, we are finding radial IMF periods and then associating them with a particular type of flow. The existence of two such events within a single rarefaction may point to them being part of the same "event," or may not. This is unclear.

Zoom In Zoom Out Reset image size

Our analysis method is rather severe. If just 1 hr of data fails the |B_R/B| > 0.90 test then whatever follows is considered to be a separate event. Figure 4 shows the number of hours between the end of one event and the start of the next. In 183 pairings, out of the total 225, the spacing exceeds 72 hr (3 days). Surely, these can be reasonably considered to be separate events. As for the rest, only 25 pairings are separated by less than 24 hr. These are the most suspect. That's 11% of the events, which is not enough to significantly alter the conclusions of this analysis, and some fraction of these pairings are very likely separate events.

Zoom In Zoom Out Reset image size

Figure 5 shows the number of events that were found in each year. The events are largely uniformly distributed across the solar cycle, although relatively few events are seen in 2003 and 2004. The years 1997 and 2012 contain relatively few events used in this study because the spacecraft was launched in 1997 September and at the time this work was performed only the first few months of 2012 had been processed. We then began a systematic study of the average solar wind properties before and after the periods of radial IMF.

Zoom In Zoom Out Reset image size

The automated codes find intervals of protracted magnetic field orientations within the ACE/MAG data where |B_R/B| > 0.9. Figure 6 shows the first example of the superposed epoch analyses that we will repeat in this paper. Before describing the results, let us first describe how we perform this analysis and the others that follow that are similar to it. We have 226 intervals of nearly radial IMF with varying duration no less than 6 hr. For this reason, we measure time as defined by before the start or after the end of any given event. For a fixed time before or after the event we sum across the 226 events to get an average and standard deviation for that measurement at that time relative to the event. We also obtain a coverage factor because not all instruments yield measurements at all times. From the standard deviation and coverage we obtain an error of the mean. Means, coverage, standard deviations and errors of the mean all vary with time and reflect the distribution of up to 226 independent measurements recorded at that time relative to the radial IMF intervals.

Zoom In Zoom Out Reset image size

Figure 6 (top) shows the average value for the quantity |B_R/B| computed across the 226 intervals where time is measured prior to the onset and after the end of the radial IMF period. Note the break in the figure where times before and after the events coincide. Figure 4 shows that the times of radial IMF are generally isolated and without similar occurrences within 3 days. The middle and bottom panels show the data coverage (percentage of total 226 intervals with data during the corresponding hour) and standard deviation across the data sets as a function of time before and after the radial IMF intervals from which the uncertainties in the top panel are computed. Note the rapid onset and disappearance of the period of radial IMF.

2.1. Wind Speed and Density

Figure 7 (top) shows the average wind speed computed across the 226 events for a given time before the onset and after the end of the radial IMF interval following the same format as Figure 6. Note that the wind speed prior to the radial IMF averages about ∼460 km s⁻¹ and is greater on average than the wind speed after the radial IMF which averages ∼390 km s⁻¹. Figure 7 (middle) shows what percentage of the total 226 hourly samples contained measurements used in computing the above average wind speed. SWEPAM coverage of the solar wind is sometimes limited by several factors including the periodic download of a complete ion distribution so that this number is rarely 100%. Figure 7 (bottom) shows the computed standard deviation for each time. The uncertainties plotted for example times in the top panel are here and elsewhere the computed error of the mean as derived from the computed standard deviation and number of independent hourly samples used in computing the mean wind speed.

Zoom In Zoom Out Reset image size

While Figure 7 clearly demonstrates a slowing wind speed in the times when we find radial IMF, the standard deviations of the underlying distributions are on the order of 100 km s⁻¹. This seems fairly typical of the normal distribution of low-latitude winds that vary between 300 and 600 km s⁻¹. This suggests that it may be worthwhile examining those underlying distribution functions. Figure 8 plots the distribution of wind speeds taken for each event when averaged over two disjoint time intervals: 15–25 hr prior to the onset of the radial IMF periods and 15–25 hr after the end of the radial IMF periods. Generally, this is far enough removed from the periods of radial IMF to remove the measurement from the rarefaction interval and place it into the fast wind sample before the rarefaction and the slow wind sample after the rarefaction. The time interval from before the rarefaction clearly shows a minor fraction of fast winds (say <600 km s⁻¹, for instance) that might be associated with polar hole sources, but the bulk of the distribution shows little evidence of the 800 km s⁻¹ winds seen at high latitudes. The slow wind distribution appears fairly typical of slow winds, but does have a minor component of what might normally be considered fast wind samples. Both samples appear to be admixtures of the fast winds normally associated with coronal hole sources and slow winds normally associated with low-latitude sources, but both distributions appear to be dominated by the latter source.

Zoom In Zoom Out Reset image size

To further demonstrate that the periods of radial IMF studied here are, in fact, rarefaction intervals, we repeat the above analysis using the proton density as obtained from the SWEPAM instrument. Figure 9 shows the results of that analysis. The plot begins with an average density that is typical of low-latitude fast winds at 1 AU, decreases during the rarefaction times when a gradient is seen in the average wind speed shown in Figure 7, and then increases to higher values after the rarefactions to a mean that is typical of the low-latitude slow wind. Again, coverage and the computed standard deviation are used to compute the error of the mean, which is shown for sample times before and after the periods of radial IMF. All of this looks like typical rarefaction region behavior (Gosling et al. 1972) where an expanding solar wind causes reduced plasma density except that these intervals were chosen to possess nearly radial magnetic fields over a persistent interval in a manner that is atypical of most rarefaction intervals.

Zoom In Zoom Out Reset image size

2.2. Thermal Ion Composition

So far we have performed an automated search for extended periods of radial IMF that produced observations within rarefaction intervals. Since the commonly recognized explanation for these periods invokes interchange reconnection in the low solar atmosphere, we now attempt to use thermal ion composition from the ACE/SWICS instrument (Gloeckler et al. 1998) to better understand the wind sources and from these infer the likely locations of the reconnection sites.

The composition instrument can be affected by low particle flux rates. Figure 10 shows the average thermal flux of protons as a function of time before and after the radial field intervals computed in the superposed epoch manner. As expected, the flux of thermal ions decreases in the radial IMF intervals, providing further confirmation that these intervals are indeed located within rarefaction regions. We will see below that this leads to reduced coverage of composition data at this time, but there are sufficient measurements to reach the necessary physical conclusions.

Zoom In Zoom Out Reset image size

Figure 11 (top) shows the average charge state composition for O⁺⁷/O⁺⁶ that climbs from ∼0.24 at the start of the rarefaction intervals to ∼0.30 by the end. While this suggests a rising electron temperature at the source it remains consistent with nominal "slow wind" or low-latitude sources and is not typical of polar coronal hole sources. Zurbuchen et al. (2002) studied Ulysses observations at high heliographic latitude and while they show that polar coronal hole sources have O⁺⁷/O⁺⁶ ∼ 0.01, they note that typical values are somewhat larger (von Steiger et al. 2001; Zurbuchen 2001). Slow wind sources show O⁺⁷/O⁺⁶ ∼ 0.1. Therefore, the above average values for O⁺⁷/O⁺⁶ that we show are not consistent with typical polar coronal hole sources and are much more consistent with slow-wind sources. Figure 11 (middle) and (bottom) show the coverage and standard deviation for the O⁺⁷/O⁺⁶ ratio from which the uncertainties plotted in the top panel are derived. The standard deviations are large and comparable to, if not greater than, the means.

Zoom In Zoom Out Reset image size

Since the standard deviations shown in the above figure are large enough to encompass some significant distribution of solar source conditions, including coronal hole sources, we select two intervals for more detailed examination. These are again 15–25 hr prior to the beginning of the radial IMF intervals and 15–25 hr after the end of the radial IMF intervals. These time selections place the samples outside or at the boundaries to the rarefaction intervals. Figure 12 shows the distribution of O⁺⁷/O⁺⁶ values for these two intervals. The two distributions are very similar although the fast wind sources ahead of the rarefactions do show a greater concentration at low values as is expected. The samples ahead of the rarefaction show 30 examples with O⁺⁷/O⁺⁶ < 0.025 and another 46 samples with 0.025 < O⁺⁷/O⁺⁶ < 0.075. There are 188 samples in total in this time range. This suggests that <20% of the fast wind sources are associated with coronal holes. In the slow wind observations that trail behind the rarefaction regions only 8 of the 191 events show O⁺⁷/O⁺⁶ < 0.025. It appears from this that coronal holes play a minor role in the formation of the radial IMF at 1 AU.

Zoom In Zoom Out Reset image size

von Steiger et al. (2010) examined Ulysses data including high-latitude observations that could reliably be traced to a polar coronal hole source and argues that the ratio (C⁺⁶/C⁺⁵) × (O⁺⁷/O⁺⁶) is an especially good determinant of coronal hole sources. Having examined O⁺⁷/O⁺⁶ already, we turn to C⁺⁶/C⁺⁵ before computing the product. Figure 13 (top) shows the average charge state composition ratio C⁺⁶/C⁺⁵ for the same data sets and times relative to the radial IMF periods. Again, as with the above oxygen ratio, the averages start low in the high-speed winds and end high in the low-speed winds. Coverage (middle panel) and standard deviations (bottom panel) are again used to compute sample errors-of-the-mean shown in the top panel. The standard deviations are again comparable to the means, so we again turn our attention to the distribution of values in the two 10 hr intervals prior to and after the rarefactions.

Zoom In Zoom Out Reset image size

Figure 14 shows the distribution of C⁺⁶/C⁺⁵ for the same two 10 hr time intervals before and after the rarefactions that were used above. Considering the target values of the next analysis, we can see that both distributions peak at relatively large values of ∼0.4 in the faster wind ahead of the period of radial IMF and ∼1.0 in the slower wind after the radial field period.

Figure 15 (top) shows the average charge state composition ratio (C⁺⁶/C⁺⁵) × (O⁺⁷/O⁺⁶) as a function of time before and after the radial IMF periods. The target value for coronal holes is (C⁺⁶/C⁺⁵) × (O⁺⁷/O⁺⁶) < 0.01. The computed means for this product are never less than 0.3. The computed standard deviations are again large.

Zoom In Zoom Out Reset image size

Figure 16 shows the distribution of (C⁺⁶/C⁺⁵) × (O⁺⁷/O⁺⁶) for the same two time intervals before and after the rarefactions. It must be pointed out that nine of the fast wind intervals and eight of the slow wind intervals have values that are off scale and >2. The fast winds ahead of the rarefactions do have ratios <0.05 for 89 of the 186 events. The slow winds behind the rarefactions have ratios <0.05 in only 45 of the 185 events. On closer examination, there are 31 events before the rarefaction with the ratio <0.015 and only 6 after the rarefaction with the ratio <0.015. While coronal holes may be a contributing source for some of the events studied here, they do not form a majority of sources including a majority of fast wind sources.

We have shown that polar coronal hole fast winds of the type seen by Ulysses at high latitudes play at best a minor role in the formation of most radial IMF events associated with rarefaction regions at low latitudes. Those rarefactions appear to be formed primarily by two low-latitude slow wind sources of different speeds. This is not to say that coronal holes represent no fraction of the sources and it may be the case that low-latitude holes provide a different wind speed and correspondingly different composition signatures than polar coronal holes. In the end, Ulysses observations (Zurbuchen 2001; Zurbuchen et al. 2002; von Steiger et al. 2001, 2010) at high latitudes and ACE observations at low latitudes must represent the sources available to those locations. All that is required for a rarefaction region at low latitude is the existence of two wind sources with different wind speeds in close proximity to one another. In order to turn a rarefaction region into a radial IMF region it becomes necessary to associate a magnetic field line with both the fast wind ahead of the rarefaction and the slow wind behind it. Either this is done by changing the wind speed associated with a fixed foot point for the IMF or it is accomplished by moving the foot point from one wind source to another. Any such rarefaction that includes an interchange reconnection event that moves the IMF foot point from a fast wind source to a slow wind source would exhibit a radial IMF in interplanetary space. This would permit the wind sources to be steady on the time scale of the rarefaction formation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

One refinement of this analysis can be seen by limiting the analysis to the years 2006–2009. At this time the polar coronal hole extends to low latitudes and may play a larger role in the dynamics. This limits the analysis to 58 events. Of those 58 events 19 have the charge state ratio (C⁺⁶/C⁺⁵) × (O⁺⁷/O⁺⁶) < 0.01 in the fast winds prior to the radial IMF and 4 events have this ratio <0.01 in the slow winds after the radial IMF. This translates into 33% and 7%, respectively. This is twice the 16% of polar coronal hole sources seen in the fast winds in the complete set of events and equivalent to the 7% of polar coronal hole sources seen in the slow winds in the complete set of events. Either way, polar coronal holes play only a minor role in fast wind sources at low latitudes.

We have examined 226 radial IMF intervals lasting at least 6 hr. These events were found by a program searching for radial IMF events alone so that no bias is applied by us in the selection. It is established, based on wind speed and density data, that the radial IMF intervals exist within rarefaction flows. We find, based on composition data, that the sources of both the faster and slower wind that form the rarefactions are nominal slow wind low-latitude sources and that polar coronal hole sources appear to play a minor role. The only explanation for this that we can see is that the observations at low latitudes make use of whatever sources are available to form rarefactions and these sources tend not to be polar coronal holes. When the interchange instability thought by some to be responsible for the radial IMF operates between two sources of different wind speed at low latitudes the radial IMF is formed. Those sources are less likely to involve polar coronal holes than observations at high latitudes. It is not a dynamic selection process but is simply a matter of availability. However, it is strongly suggestive of the interchange instability being a common feature of the solar photosphere and corona that is not limited to the boundaries of polar coronal holes.

S.T.O. and C.W.S. are supported by Caltech subcontract 44A-1062037 to the University of New Hampshire in support of the ACE/MAG instrument. B.J.V. is supported by NASA Guest Investigator grant NNX09AG28G, NASA/SR&T grant NNX10AC18G, and NSF/SHINE grant ATM0850705. Support for N.A.S. is provided by the NSF FESD Sun-to-Ice Project. Support at LANL was provided under the auspices of the U.S. Department of Energy, with financial support from the NASA ACE program. T.H.Z. and L.Z. are supported by an ACE program subcontract from Caltech. S.T.O. is an undergraduate physics major at the University of New Hampshire.