It has been known for more than 50 years that there are significant amounts of neutral hydrogen around the Milky Way that do not follow Galactic rotation – the high velocity clouds (HVCs) (Muller et al., 1963; Wakker et al., 2004). For many years the lack of information on their distances has allowed speculation that some of them might be ∼1 Mpc or more away from the Milky Way and thus have MHI = 107−8 M⊙ (Blitz et al., 1999; Braun & Burton, 1999). However, the discovery of a similar population around M31, M33, and other galaxies (Thilker et al., 2004; Westmeier et al., 2008; Grossi et al., 2008; Putman et al., 2009; Keenan et al., 2016; Miller et al., 2009), suggests that the phenomenon is likely common and, as importantly, confined to the inner CGM of spirals. This does not mean that some of the more compact, more isolated HVCs might not be free-floating independent systems, but the total H I mass within the CGM of normal galaxies (aside from the products of tidal interaction) must be relatively small (Pisano et al., 2007; Pisano et al., 2011). The HVC system of the Milky Way has been estimated to contain an H I mass of ∼ 3 × 107 M⊙ (Putman et al., 2012). The HVC population likely contains as much ionized gas as neutral gas (Shull et al., 2009; Shull et al., 2011; Lehner & Howk, 2011).
One of the most important facts about HVCs is that they do not contain stars (Hopp et al., 2007). In this sense, compact HVCs that are discovered to be dwarf galaxies (Rhode et al., 2013), are simply objects that have been mis-classified as HVCs. Another important property is that their velocities are not really that high: none have velocities near or greater than the escape velocity of the Milky Way (Wakker et al., 2004). Putman et al. (2012) estimate that the larger Milky Way HVCs have kinematics that can be fit by a prograde rotational component about one-third the circular velocity of the Sun, and an inflow component of a few tens of km s−1 , though one prominent cloud almost certainly has a large retrograde motion (Lockman, 2003).
3.1. High Velocity Clouds in M31 and M33
The HVC population of M31 is shown in Figure 3. The typical M31 HVC has MHI of a few 105 M⊙, and would not be detectable as an individual object in most measurements of more distant galaxies. The right panel of Fig. 3 emphasizes the proximity of HVCs to that galaxy's disk, and the difficulty one would have in disentangling them from the disk-halo layer in more distant systems. The HVCs of M31 seem to be confined to within 50 kpc of that galaxy, though there are indications that some may be found to greater distances to the north (Westmeier et al., 2008; Wolfe et al., 2016). The total H I mass in the M31 HVCs is 2 × 107 M⊙; when extraplanar gas possibly associated with the disk is included, the amount may be as high as 5 × 107 M⊙ (Westmeier et al., 2008).
 |
Figure 3. From Westmeier et al. (2008). Left panel: the HVC system of M31 in relationship to the bright H I in the disk. Right panel: integrated H I over all velocities with contours at log10(NHI) 18, 18.5, 19, 19.5, ... . Many of the HVCs seem to blend into the disk, but in maps at different velocities it is clear that they are distinct objects. This is how M31 would appear to current instruments were it ∼ 10 Mpc distant, except that the two lowest contours would be missing. As noted by Putman et al. (2012) this image of M31 bears a strong resemblance to the H I image of the galaxy NGC 891 (Oosterloo et al., 2007) which has been observed with about the similar linear resolution as M31. |
There are also HVCs around M33, though here confusion with that galaxy's disk and possibly unrelated extraplanar gas makes it difficult to separate the populations cleanly (Grossi et al., 2008; Putman et al., 2009; Keenan et al., 2016). The total H I mass in the M33 HVCs is 3.5 × 107 M⊙ using only the data from the most recent study (Keenan et al., 2016). If we include clouds that may be located in the disk-halo region (Grossi et al., 2008, Putman et al., 2009) the total M33 HVC H I mass increases to ∼ 5 × 107 M⊙.
Figure 4 shows the velocity and location of the HVCs of M31 and M33 with respect to M31. At the distance of these galaxies one degree on the sky corresponds to about 15 kpc. This figure includes some objects in M33 (marked in blue) that are probably part of the disk-halo interface and not true HVCs (Putman et al., 2009; Keenan et al., 2016), but are included to show that aspects of these two populations can blend and they often occupy very similar areas of position and velocity space. This figure shows another interesting aspect of HVCs. The velocity spread of those associated with M33 is considerably smaller than those associated with M31. If HVCs are condensations from the CGM this would follow naturally, as M33 is less massive than M31 and so presumably has a CGM that rotates more slowly.
 |
Figure 4. The HVC systems of M31 and M33 (red circles), where the velocity with respect to the Local Group Standard of rest, VLGSR, is plotted against the angular distance from M31 in the direction of M33. To the samples of M31 and M33 HVCs (Westmeier et al., 2008; Keenan et al., 2016) are added clouds that may be in the halo of M33 (blue circles) (Grossi et al., 2008; Putman et al., 2009). It can be difficult to distinguish between the two groups, which may actually blend together. The systemic velocities of M31 and M33 are indicated. Diamonds mark the M31-M33 clouds discussed in section 5.2. They have velocities more in common with the systemic velocity of M31 and M33 than of their system of HVCs. There are no stars in any of these clouds. |
A key fact that has been established through distance measurements to Milky Way HVCs and studies of M31 and M33 is that HVCs are not free-floating in the Local Group, but are concentrated around the large spirals. If the HVCs are even in approximate pressure equilibrium with a galaxy's CGM they must be orders of magnitude more dense than the hot gas around them and thus falling toward the disk. There is some evidence for the interaction of Galactic HVCs with the CGM in the form of distortions of the cloud shapes (Brüns et al., 2000), and there is rather spectacular evidence for the direct accretion of H I onto the Milky Way from one HVC: The Smith Cloud.
3.2. The Smith Cloud – Accretion in Action
In the same year that the discovery of HVCs was announced, a short paper appeared reporting observations of a peculiar H I feature that now appears to be a very important object, the Smith Cloud (Smith, 1963). Figure 5 shows an H I channel map from a recent 21cm survey made with the Green Bank Telescope (GBT). The cloud has a good distance estimate (Wakker et al., 2008), and thus a well defined mass and size; it is moving toward the Galactic plane, which it should intersect in ∼ 30 Myr if it is survives as a coherent entity (Lockman et al., 2008). Its mass in H I is ∼ 2 × 106 M⊙ and it has an ionized component with a similar mass detected in faint Hα emission (Hill et al., 2009). There is evidence that it has a magnetic field ≈ 8 µG (Hill et al., 2013) and a S/H abundance that is about one-half Solar (Fox et al., 2016). Its total space velocity is below the escape velocity of the Milky Way, and the largest component is in the direction of Galactic rotation. It appears to be entering the Milky Way at a rather shallow angle (Lockman et al., 2008, Nichols & Bland-Hawthorn, 2009). The Smith Cloud is thus adding angular momentum to the disk. It has no detectable stars (Stark et al., 2015).
 |
Figure 5. Channel map of the Smith Cloud from new GBT data. All the emission in this figure is associated with the Cloud, which has other components extending down to latitude −40∘ . The part of the Cloud displayed in this image is about 4 kpc in length. The Galactic plane lies above this figure, and the Cloud is moving toward it at an angle. |
The brightest, most compact component of the Cloud lies about 3 kpc below the Galactic plane. It is thus in the disk-halo transition, and it appears that it is encountering a clumpy medium. The Cloud has holes that are matched by small fragments at 60 km s−1 lower velocity, consistent with the velocity of the Milky Way's halo at that location (Fig. 6). This object shows clear signs of interaction with the Milky Way's extraplanar gas.
 |
Figure 6. A velocity-position plot down the major axis of the Smith Cloud from Lockman et al. (2008). Arrows mark lumps or streams of H I that have been decelerated by interaction with the Milky Way halo. Besides a general stripping and deceleration of the cloud edges (H I indicated by dashed arrows), it appears that the Cloud is encountering dense lumps in the Milky Way disk-halo interface that remove chunks of the Cloud (noted with solid arrows). Some of these correspond in detail to voids in the body of the Smith Cloud. |
There are many puzzles surrounding the Smith Cloud. If it has condensed from the CGM than why does it have a S/H metallicity ratio higher than typical HVCs (Fox et al., 2016, Wakker et al., 2004)? If it originated from the Galaxy, then how did it acquire such a large peculiar motion and mass, which implies a kinetic energy ∼ 5 × 1053 ergs (Marasco & Fraternali, 2017)? Does it require a significant dark matter component to maintain its stability, as suggested by some investigations (Nichols & Bland-Hawthorn, 2009; Nichols et al., 2014)?
If the Smith Cloud were at the distance of M31 it would appear to the HVC surveys with a peak NHI around 2 × 1019 cm−2, a few times higher than the M31 HVCs, but not especially anomalous. With an MHI of 2 × 106 M⊙, the Smith Cloud has about the median H I mass of the M33 HVCs and lies at the upper range of those around M31. Given the observational uncertainties, the Smith Cloud would be an inconspicuous addition to either galaxy (Lockman et al., 2008; Westmeier et al., 2008; Keenan et al., 2016). However, as the brighter parts of the Smith Cloud lie only ∼ 3 kpc away from the Galactic plane, if it were at the distance of M31 or M33 it would lie projected on the disk of those galaxies, and from even the most favorable vantage would be separated by only 12′ from their disk. It would almost certainly be considered part of the disk-halo interface and not cataloged as an HVC.
Some HVCs in M31 and M33 have linewidths > 50 km s−1 , suggesting large internal motions or disruption (Westmeier et al., 2008; Keenan et al., 2016). If the Smith Cloud were at the distance of M31, our poorer linear resolution would blend its different components, raising its H I linewidth from the < 20 km s−1 that we typically measure in its brighter parts, to 40 km s−1 .