The Milky Way gives us a close view of process like the galactic fountain, which lifts gas several kpc away from a spiral galaxy's disk (Shapiro & Field, 1976; Bregman, 1980). Figure 2 shows Milky Way H I along a cut ∼ 20∘ in longitude through the Galactic plane approximately along the tangent points. It is easy to find H I loops, filaments, clouds and a diffuse component extending many degrees away from the plane. In this figure each degree of latitude corresponds to a displacement ≈ 135 pc from the midplane, so the H I emission at b = ± 4∘ arises in neutral gas at a height z ∼ 0.5 kpc, several H I scale heights above the main H I layer (Dickey & Lockman, 1990). In the inner Milky Way about 10% of the H I can be found more than 0.5 kpc from the disk (Lockman, 1984). Some of this H I is contained in discrete clouds that can be identified to z ≈ 2 kpc (Lockman, 2002; Ford et al., 2010). Some of the extraplanar H I is also organized as large "supershells", whose tops can reach |z| > 3 kpc. They can be ∼ 1 kpc is size and contain 3 × 104 M⊙ of neutral gas (Heiles, 1979; McClure-Griffiths et al., 2002; Pidopryhora et al., 2007).
Figure 2. Cut through the Galactic plane at velocities approximately along the tangent point showing the vertical extent of Galactic HI. In this projection a latitude of 8∘ corresponds to a distance ∼ 1 kpc from the Galactic plane. These observations can resolve structures down to a few 10s of pc in size.
Supershells certainly have their origin in supernovae and stellar winds (Tomisaka & Ikeuchi, 1986), and can often be linked directly to sites of star formation (Pidopryhora et al., 2007; Kaltcheva et al., 2014). The cloud-like component of extraplanar H I has a less certain origin, though it may result from the breakup of shells. Often referred to as "disk-halo" clouds, they appear to be related both spatially and kinematically to the disk, though the population extends well into the lower halo (Lockman, 2002; Ford et al., 2010). Some of the larger Milky Way disk-halo clouds with MHI ≈ 500 M⊙ have been imaged at 2-3 pc resolution and show a variety of shapes with some sharp boundaries and evidence of a two-component thermodynamic structure (Pidopryhora et al., 2015). Note that from the above discussion on sensitivity, these disk-halo clouds could not be detected beyond the Milky Way as individual objects. Within the Milky Way there is a general correlation between the amount and vertical extent of disk-halo clouds and the spiral arms (Ford et al., 2010) although there is no detailed correlation with individual star-forming regions.
A similar extraplanar H I layer is seen in a number of galaxies, often containing 10% of the disk H I mass, though the percentage has large variations from galaxy to galaxy (Sancisi et al., 2008). While it is not possible to isolate individual disk-halo clouds in other galaxies, the better vantage afforded in observations of extragalactic systems allows the kinematics of this component to be analyzed more completely. The disk-halo (or extraplanar) gas often shows evidence of a vertical lag in rotational velocity of 10−20 km s−1 kpc−1 but with a large range; sometimes an inflow toward the center of a galaxy is also inferred at the level of 10-20 km s−1 (Fraternali et al., 2002; Sancisi et al., 2008; Zschaechner et al., 2015). This extraplanar gas does not show large deviations from prograde galactic rotation, and constitutes a dynamic neutral galactic atmosphere.
The disk-halo gas extends to heights where it may mix with material accreting through or cooling from the CGM (Putman et al., 2012). Thus the presence of neutral gas many scale-heights away from the plane is not unexpected, and distinguishing between gas that is "recycled" as much of the disk-halo material must be, and gas that has never been in the disk, may not be straightforward in the absence of other information. Although the disk-halo clouds are concentrated to the disk in the sense that their numbers increase towards the disk, there is no way to know if any individual cloud had its origin in the Milky Way, or has been accreted, or is a combination of both processes (Marasco et al., 2012). It would be very interesting to have information on the elemental abundances in disk-halo clouds at different heights.
Any gas accreting onto the Milky Way has to pass through the extended disk-halo layer before it reaches the inner disk. One massive cloud passing through this layer shows evidence of disruption, and is discussed in section 3.2.