While the Parker model describes the HMF remarkably well, there are a number of “second-order” alterations or additions required to fully explain observations. One such observation is the existence of shock-accelerated particles at latitudes higher than where CIR shocks are located. In the declining phase of the solar cycle, Ulysses observed CIRs to be confined to within 40° of the solar equator, about 10° greater than the 30° maximum latitude of the HCS at the time, but the energetic protons and electrons associated with CIR shocks were observed at much higher latitudes (Roelof et al., 1997). While solar wind particles are typically frozen to field lines, one possible explanation is that these more energetic particles effectively diffuse across the magnetic field as a result of scattering off magnetic waves and inhomogeneities (Kóta and Jokipii, 1995).
Alternatively, if the photospheric connectivity of heliospheric field lines changes in a systematic fashion, particles accelerated at CIRs close to the HCS could be expected at high latitudes without the need for strong cross-field diffusion. Such a framework was described by Fisk (1996). It results from combining a number of observations of the solar magnetic field, most importantly, the photospheric plasma and magnetic field are known to rotate differentially with latitude, from a rotational period of approximately 25 days at the equator, to over 30 days in the polar regions. Coronal holes, on the other hand, are observed to rotate rigidly about the rotation axis but with an axis of symmetry which is tilted with respect to the rotation axis (e.g., Bird and Edenhofer, 1990, and references therein). The non-radial overexpansion of the magnetic field within coronal holes described at the end of Section 2.3 also takes place about this symmetry axis. If this symmetry axis and rotational axis are aligned, HMF footpoints drift around the rotation axis resulting in a Parker-like heliospheric field in which the HMF traces out cones of constant latitude even if the field lines are rooted at a higher latitude in the photosphere. If, however, the symmetry axis of the magnetic structure is inclined to the rotational axis, the HMF becomes more complex and a field line from a particular moving photospheric source can make large excursions in latitude over time in the heliosphere due to experiencing different amounts of overexpansion in the corona. A further consequence of magnetic inclination to the rotation axis is that reconnection between the open HMF within coronal holes and closed coronal loops at the edges of coronal holes (referred to as “interchange reconnection,” e.g., Crooker et al., 2002) allows the HMF footpoints to saltate across the photosphere against differential rotation (Nash et al., 1988; Wang and Sheeley Jr, 2004). See Fisk et al. (1999) for more detail.
The effects of the above model should be most systematic in the stable tilted dipole configuration of the solar corona often encountered in the declining and minimum phases of the solar cycle, such as was charactersitic of the 1992 – 1997 Ulysses data. Attempts have been made to detect systematic latitudinal components in the Ulysses HMF data (Zurbuchen et al., 1997; Forsyth et al., 2002) but it was found likely that the amplitude of the signal would be too small to stand out from the general variability of the field. Alternative observational evidence for the resulting deviations from the Parker model can be found in rarefaction regions behind CIRs, where the HMF is found to be systematically more radial than an ideal Parker spiral for the observed solar wind speed (Murphy et al., 2002). This has been interpreted as the changing solar wind speed at the HMF footpoint (Schwadron, 2002), which could be expected if the HMF footpoint moves across the coronal hole boundary at the trailing edge of the fast stream as a result of differential rotation. Indeed, a similar mechanism has been proposed as the source of the slow solar wind (Fisk, 2003).
Solar wind intervals have also been identified in which the HMF is Parker spiral-aligned, but the suprathermal electron strahl is directed toward the Sun (Kahler et al., 1996; Crooker et al., 2004b, and references therein). This must result from the HMF being locally inverted, most likely as a result of interchange reconnection in the corona opening up a previously closed coronal loop (Owens et al., 2013), possibly a further signature of the circulation of the HMF.
Stimulated by asymmetries noted in the latitudinal gradients of cosmic rays in 1995 solar minimum Ulysses observations, there has been continuing interest as to whether there is a north-south asymmetry present in both the solar and heliospheric magnetic fields. It was noted at the time that these results could be explained by a 10° southward displacement of the heliomagnetic equator and, hence, the heliospheric current sheet (e.g., Simpson et al., 1996). Although initial analysis of Ulysses magnetic field data did not support such a large displacement, Smith et al. (2000) showed that Wind data in the ecliptic were consistent with a 10° displacement at this time, the effect at Ulysses being masked by temporal changes. Subsequent analysis of HMF data at 1 AU (Mursula and Hiltula, 2003) and at Ulysses (Erdős and Balogh, 2010) have yielded results consistent with a long term trend of a few ( 2 – 3) degrees southward displacement of the HCS. During the Ulysses mission, this displacement has been consistently southward, independent of the reversals in the solar magnetic dipole polarity in alternate solar cycles. As discussed in the review of Smith (2008), the interpretation and comparison of these and similar studies requires care in separating spatial and temporal effects.
Coronal mass ejections (CMEs) are huge eruptions of solar plasma and magnetic field. As CME initiation and release frequently involves magnetic reconnection, CMEs are often spatially and temporally collocated with solar flares, though it is now clear that flares do not trigger CMEs (Harrison, 1995). CMEs move out through the corona and into the heliosphere where fast (slow) CMEs are accelerated (decelerated) towards the ambient solar wind speed (e.g., Gopalswamy et al., 2000; Cargill, 2004). These interplanetary CMEs (ICMEs) can be observed both remotely with white-light heliospheric imagers as density perturbations (e.g., Davis et al., 2009) and directly with in situ magnetic field and particle detectors. ICMEs produce the largest deviations from the Parker spiral magnetic field and are the primary source of strong meridional HMF in the near-Earth solar wind, making ICMEs particularly geoeffective (e.g., Gosling, 1993; Schwenn, 2006, and references therein). The out-of-ecliptic magnetic field can result both from the ICME structure itself (see Section 4.2.1) and from distortion of the ambient HMF (e.g., Jones et al., 2002, see also Section 2.5).
There are a number of plasma, magnetic field, compositional and charge-state signatures used to identify ICMEs from in situ data, though no one signature is either necessary or sufficient for classification. Ion charge state and elemental abundance signatures are generally consistent with CMEs forming in the hotter corona, below the bulk solar wind acceleration height (e.g., Gloeckler et al., 1999; Lepri and Zurbuchen, 2004). Plasma density and pressure are usually lower than the bulk solar wind, suggesting ICMEs undergo greater expansion than the bulk solar wind (e.g., Cane and Richardson, 2003). See Neugebauer and Goldstein (1997) and Wimmer-Schweingruber et al. (2006) for thorough reviews of ICME signatures and their implications for the formation and evolution of ejecta. In this review, we concentrate on the magnetic signatures of ICMEs, and how ICMEs relate to the HMF in general.
Magnetic clouds (MCs) are a subset of ICMEs primarily characterised by a large-scale, smooth rotation in the magnetic field direction, and are typically associated with an increase in field magnitude and a decrease in small-scale field variance (Burlaga et al., 1981; Klein and Burlaga, 1982). These signatures have been interpreted and modeled as a magnetic flux rope (Burlaga, 1988; Lepping et al., 1990). Figure 11 shows an example of a magnetic cloud, observed by the Wind and ACE spacecraft in August 1998. Approximately 5 days of data are shown, with the MC boundaries shown as the solid vertical lines. The second panel from the top shows the magnetic field magnitude, which is significantly enhanced above that in the ambient solar wind. The following two panels show the in- and out-of-ecliptic plane angles of the magnetic field, respectively. The smooth rotation in the magnetic field direction is clearly identifiable, resulting in large out-of-ecliptic magnetic fields at the front and rear portions of the MC. It’s likely such magnetic clouds are the largest events in a spectrum of flux-rope associated solar ejecta (Moldwin et al., 2000; Rouillard et al., 2011).
MCs comprise between a third (Gosling, 1990) to a half (Cane and Richardson, 2003) of all ICMEs in near-Earth space, with some evidence of this fraction varying with the solar cycle (Riley et al., 2006b). At present, it is unclear whether or not all CMEs involve erupting flux ropes, but the signature is simply not seen in in situ observations because of sampling effects, in-transit distortion, etc. (e.g., Jacobs et al., 2009). Although constituting a minority of total ICMEs observed, MCs have received considerable attention for two reasons. Firstly, they drive the largest geomagnetic disturbances (Gosling, 1993; Richardson et al., 2002). Secondly, fitting a flux rope model to the single-point in situ data allows estimation of the large-scale magnetic properties of ICMEs to be estimated, notably the local flux-rope orientation and total magnetic flux content, information unobtainable by other methods and/or observations.
The total CME mass flux, as estimated from coronagraph observations, is only a minor contribution to the solar wind (Webb and Howard, 1994). Similarly, the magnetic flux carried by a single magnetic cloud is (Lynch et al., 2005; Owens, 2008), only a few percent of the estimated total open solar magnetic flux at any time (, e.g., Owens et al., 2008a). This is in rough agreement with the small fraction of time the near-Earth solar wind can be attributed to recognisable ICME material (Richardson et al., 2000). However, while a single CME is unlikely to have a significant contribution to the total open solar flux, the net CME contribution also depends on the time for which CME magnetic flux remains connected to the Sun, as discussed in Section 5.4. Indeed, there are a number of observations which suggest CMEs are intrinsically linked with the large-scale evolution of the HMF, as discussed here.
Particularly during solar minimum, when the streamer belt and heliospheric current sheet coincide, magnetic clouds are frequently encountered at magnetic sector boundaries. In such instances, the normally sharp transition from inward to outward magnetic polarities associated with a crossing of the HCS instead takes the form of a smooth rotation in the magnetic field direction associated with the passage of the MC’s flux rope (Crooker et al., 1998a). Clearly, the magnetic cloud polarity is determined by the large-scale solar magnetic field orientation and may be a means by which the large-scale field evolves, shifting the location of the sector boundary in response to a change in the photospheric magnetic flux (Crooker et al., 1998b, see also Section 5.4). Note, however, that many CMEs do not result in a permanent change to the HCS position (Zhao and Hoeksema, 1996). The relation between the large-scale solar magnetic field and ICMEs is further evidenced by the observed solar-cycle and hemispheric trends in magnetic cloud orientations and polarities. In-ecliptic observations of magnetic clouds show that their orientation and polarity follows the Hale law of sunspot polarity (Hale and Nicholson, 1925), where the polarity of the leading (in the sense of solar rotation), lower-latitude sunspot is the same as the dominant hemispheric polarity at the start of the solar cycle (Bothmer and Rust, 1997; Bothmer and Schwenn, 1998; Mulligan et al., 1998; Li et al., 2011). Many photospheric and coronal structures exhibit magnetic helicity ordered by hemisphere, with the northern (southern) hemisphere associated with left- (right-) handed helicity. In the heliosphere, magnetic clouds showing the same trends (Rust, 1994; Marubashi, 1997). Ulysses out-of-ecliptic observations of magnetic clouds over solar cycle 23 (Rees and Forsyth, 2003) further support the idea that ICMEs project the Hale cycle out into the heliosphere. These relations have important consequences for the means by which the solar cycle polarity reversal is communicated out to the heliospheric magnetic field, as discussed in Section 5.4.
Counterstreaming suprathermal electron observations of magnetic clouds indicate that approximately 50 – 60% of ICME-associated magnetic flux is formed of heliospheric loops with both ends attached to the Sun, with little change in this fraction between 1 and 5 AU (Gosling et al., 1987; Shodhan et al., 2000; Crooker et al., 2004a; Riley et al., 2004). Hence, Crooker et al. (2004a) and Riley et al. (2004) concluded that closed loops within ICMEs must add to the total open solar flux for long time scales (months to years). This is discussed further in Section 5.4.
In addition to the large-scale, global features discussed thus far, single-point spacecraft observations reveal fluctuations in the HMF over all observable time scales. These are interpreted as a combination of spatial and temporal variations in the rest frame of the plasma, with waves, shocks, turbulence, tangential- and rotational discontinuities all likely contributing (e.g., Matthaeus et al., 1986; Horbury et al., 2001; Bruno et al., 2001; Bruno and Carbone, 2013). A complete discussion of such phenomena is beyond the scope of this review, but we provide a brief overview of the observations which pertain to the origin of the HMF itself.
The solar wind exhibits a large array of different wave modes, many of which directly perturb the heliospheric magnetic field (see Tu and Marsch, 1995 and Marsch, 2006). The HMF in the high speed solar wind, particularly the polar regions at solar minimum, is dominated by Alfvénic fluctuations, flowing predominantly anti-sunward in the plasma frame (Smith et al., 1995; Goldstein et al., 1995; Horbury et al., 1995). The solar wind is also highly turbulent, further adding to the spectrum of fluctuations in the HMF (Bruno and Carbone, 2013, and references therein). However, the full extent of this turbulence is debated, as magnetic field discontinuities could equally be the result of spacecraft encountering structures convected by the solar wind (Mariani et al., 1973; Tu and Marsch, 1993; Bruno et al., 2001; Borovsky, 2008). In this model of the HMF, the largest changes in the magnetic field direction are the result of crossing boundaries between large, coherent flux tubes, while the smaller fluctuations are true turbulent fluctuations within the flux tubes themselves. Such structures pass spacecraft with a time scale 10 minutes, thus the inferred flux tube size is in approximate agreement with super-granules on the Sun (Neugebauer et al., 1995). However, the weak association between large magnetic discontinuities and compositional changes (Owens et al., 2011b), mean they are equally likely to have formed by turbulence during transit, as be of solar origin.
The turbulent/filamentary nature of the HMF is important as it can place constraints on the solar wind formation mechanism. Figure 12 shows sketches of possible mechanisms for coronal heating, along with the implications for heliospheric magnetic field discontinuities (Cranmer, 2008). On the left, the corona is heated by reconnection between open solar flux and closed loops emerging through the photosphere (e.g., Fisk, 2003; Schwadron and McComas, 2003; Schwadron et al., 2006) and the heliospheric magnetic field will naturally become tangled due to foot point motions. On the right, the corona is heated by waves and/or turbulence (e.g., Cranmer and van Ballegooijen, 2005; Verdini and Velli, 2007). The heliospheric magnetic field can then become tangled by turbulent motions, either propagating directly from the corona or generated in transit. Of course, it may be possible that both mechanisms are at play.
The largest amplitude waves, driven by fast ICMEs or the interaction of fast and slow solar wind streams, can steepen into interplanetary shock waves (Balogh et al., 1995). CIR shocks and associated structures are discussed in Section 2.5. For fast ICMEs, shocks can form inside the corona. The region of compressed solar wind bounded by the shock and the ICME leading edge is referred to as the “sheath,” and is analogous to the planetary magnetosheaths (though see Siscoe and Odstrčil, 2008; Savani et al., 2011). Magnetic fields in ICME sheaths can frequently be strong enough to trigger geomagnetic activity in their own right (Owens et al., 2005), with a quarter (Richardson et al., 2001) to a half (Tsurutani et al., 1988) of all geomagnetic storms potentially attributable to ICME sheaths. As with CIRs, pre-existing structures or fluctuations in the upstream solar wind are swept up into the ICME sheath and compressed into planes perpendicular to the ICME leading edge or stream interface (Jones et al., 2002).
Large magnetic field discontinuities in the solar wind would seem to provide ideal conditions for magnetic reconnection. The relatively high plasma beta, however, argues against widespread reconnection in the solar wind. This debate was finally settled in 2005, when signatures of reconnection, in the form of large-scale reconnection outflow exhausts, were observed in the near-Earth solar wind (Gosling et al., 2005; Phan et al., 2006). The leading edge of fast ICMEs seems to be a preferential location for reconnection, but it is also regularly occurs at low HMF shear angles in low plasma beta fields, often found within ICMEs (e.g., Gosling et al., 2007).