1 Introduction

Since the discovery of the H  I high-velocity clouds by Muller et al. (1963), different explanations, each with its own characteristic distance scale, have been proposed. It is likely that not all of the anomalous-velocity H  I represents a single phenomenon, in a single physical state. Determining the topology of the entire population of anomalous-velocity H  I is not a simple matter, and the task is all the more daunting to carry out on an all-sky basis because of disparities between the observational survey material available from the northern and southern hemispheres. The question of distance remains the most important, because the principal physical parameters depend on distance: mass varying as d², density as d^-1, and linear size directly as d. Most of the H  I emission at anomalous velocities is contributed from extended complexes containing internal sub-structure but embedded in a common diffuse envelope, with angular sizes up to tens of degrees. Such structures include the Magellanic Stream of debris from the Galaxy/LMC interaction and several HVC complexes, most notably complexes A, C, and H. The complexes are few in number but dominate the H  I flux observed.

The Magellanic Stream comprises gas stripped from the Large Magellanic Cloud, either by the Galactic tidal field or by the ram-pressure of the motion of the LMC through the gaseous halo of the Galaxy. It therefore will be located at a distance comparable to that of the Magellanic Cloud, i.e. some $50\rm\;kpc$ (see e.g. Putman & Gibson 1999). The distance to Complex A has been constrained by van Woerden et al. (1999) and then more tightly by Wakker (2001) to lie within the distance range 8<d<10 kpc. If, as seems plausible, the other large complexes also lie at distances ranging from several to some 50 kpc, they will have been substantially affected by the radiation and gravitational fields of the Milky Way.

Another category of anomalous H  I high-velocity clouds are the compact, isolated high-velocity clouds discussed by Braun & Burton (1999). CHVCs are distinct from the HVC complexes in that they are sharply bounded in angular extent at very low column density limits, i.e. below 1.5$\times$10¹⁸ cm^-2 (de Heij et al. 2002). This is an order of magnitude lower than the critical H  I column density of about 2$\times$10¹⁹ cm^-2, where the ionized fraction is thought to increase dramatically due to the extragalactic radiation field. For this reason, these objects are likely to provide their own shielding to ionizing radiation. Although not selected on the basis of angular size, such sharply bounded objects are found to be rather compact, with a median angular size of less than 1 degree.

An analogy of the CHVC ensemble with that of the dwarf galaxy population in the Local Group is suggestive, and illustrates the hypothesis that is under discussion here. Some few Local Group dwarf galaxies also extend over large angles. The Sgr Dwarf Spheroidal discovered by Ibata et al. (1994) spans some $40\hbox{$^\circ$ }$; presumably it was once a rather conventional dwarf, but its current proximity to the Milky Way accounts for its large angular size. This proximity has fundamentally distorted its shape, and will determine its further evolution. The streams of stars found in the halo of the Galaxy by Helmi et al. (1999) probably represent even more dramatic examples of the fate which awaits dwarf galaxies which transgress into the sphere of the Galaxy's dominance. Analogous stellar streams have been found in M 31 by Ibata et al. (2001) and by Choi et al. (2002), as well as in association with Local Group dwarf galaxies by Majewski et al. (2000), indicating that accretion (and subsequent stripping) of satellites is an ongoing process. If a selection were to be made of the dwarf galaxy population in the Local Group on the basis of angular size, the few large-angle systems which would be selected, namely the LMC and SMC, and the Sgr dwarf, and - depending on the flexibility of the selection criteria - perhaps the ill-fated coherent stellar streams in the Galactic halo, would represent systems nearby, of large angular extent, and currently undergoing substantial evolution. Those systems selected on the basis of being compact and isolated, on the other hand, would represent dwarf galaxies typically at substantial distances and typically at a more primitive stage in their evolution. Regarding distance and evolutionary status they would differ from the nearby, extended objects, although at some earlier stage the distinction would not have been relevant.

The possibility that some of the high-velocity clouds might be essentially extragalactic has been considered in various contexts by, among others, Oort (1966, 1970, 1981), Verschuur (1975), Eichler (1976), Einasto et al. (1976), Giovanelli (1981), Bajaja et al. (1987), Wakker & van Woerden (1997), Braun & Burton (1999), and Blitz et al. (1999). It is interesting to note that the principal earlier arguments given against a Local Group deployment, most effectively in the papers cited above by Oort and Giovanelli, were based on the angular sizes of the few large complexes and on the predominance of negative velocities in the single hemisphere of the sky for which substantial observational data were then available. The more complete data available now, however, show about as many features at positive velocities as at negative ones. It is also interesting to note that the papers by Eichler and by Einasto et al. cited above consider distant high-velocity clouds as possible sources of matter, including dark matter, fueling continuing evolution of the Galaxy. Blitz et al. (1999) revived the suggestion that high-velocity clouds are the primordial building blocks fueling galactic growth and evolution, and argued that the extended complexes owe their angular extent to their proximity.

Braun & Burton (1999) identified CHVCs as a subset of the anomalous-velocity gas that might be characteristic of a single class of HVCs, whose members plausibly originated under common circumstances and share a common subsequent evolutionary history. They emphasized the importance of extracting a homogenous sample of independently confirmed objects from well-sampled, high-sensitivity H  I surveys. The spatial and kinematic distributions of the CHVCs were found by Braun & Burton to be consistent with a dynamically cold ensemble spread throughout the Local Group, but with a net negative velocity with respect to the mean of the Local Group galaxies. They suggested that the CHVCs might represent the low-circular-velocity dark matter halos predicted by Klypin et al. (1999) and Moore et al. (1999) in the context of the hierarchical structure paradigm of galactic evolution. These halos would contain no, or only a few, stars; most of their visible matter would be in the form of atomic hydrogen. Although many of the halos would already have been accreted into the Galaxy or M 31, some would still populate the Local Group, either located in the far field or concentrated around the two dominate Local Group galaxies. Those passing close to either the Milky Way or M 31 would be ram-pressure stripped of their gas and tidally disrupted by the gravitational field. Near the Milky Way, the tidally distorted features would correspond to the high-velocity-cloud complexes observed.

The quality and quantity of survey material is important to interpretation of the CHVC population, which is a global one. The observational data entering this analysis involved merging two catalogs of CHVCs, one based on the material in the Leiden/Dwingeloo Survey (LDS) of Hartmann & Burton (1997), and the other based on the H  I Parkes All-Sky Survey (HIPASS) described by Barnes et al. (2001). Both of these surveys were searched for anomalous-velocity features using the algorithm described by de Heij et al. (2002, Paper I). This algorithm led to the LDS catalog of de Heij et al. for the CHVCs at declinations north of $-30\hbox{$^\circ$ }$, and to the HIPASS catalog of Putman et al. (2002) for those at $\delta<0\hbox{$^\circ$ }$. The surveys overlap in the declination range $+2\hbox{$^\circ$ }<\delta < -30\hbox{$^\circ$ }$, allowing estimates of the relative completeness of the catalogs. We are able to predict how a survey with the LDS parameters would respond to the CHVCs detected by HIPASS, and vice versa. In the subsequent simulations, we are able to sample the simulated material as if it were being observed by one of these surveys.

This paper is organized as follows. We describe the application of the algorithm in Sect. 2.1. In Sect. 2.2 we discuss various observational selection effects, and indicate how to account for these. We address obscuration by our own Galaxy in Sect. 2.2.1, the consequences of the differing observational parameters of the LDS and HIPASS data in Sect. 2.2.2, and the resulting differing degrees of completeness of the LDS and HIPASS catalogs in Sect. 2.2.3. We discuss the observable all-sky properties of the CHVC ensemble in Sect. 3, including the spatial deployment (in Sect. 3.1), the kinematic deployment (in Sect. 3.2), and the distributions of H  I flux and angular size (in Sect. 3.3). We then attempt to reproduce these properties by considering models, first based on Local Group distributions as discussed in Sect. 4 and then on distributions within an extended Galactic Halo as discussed in Sect. 5; these simulations are sampled as if being observed with the LDS and HIPASS programs. We discuss the conclusions that can be drawn from this analysis in Sect. 6.

Figure 1: Fraction of a homogeneously distributed sample of test clouds that is not blended with Galactic emission. The sample was distributed on the sky with a Gaussian velocity distribution in the Local Standard of Rest system (upper panel), the Galactic Standard of Rest system (middle), and the Local Group Standard of Rest (lower). The average velocity of the Gaussian velocity distributions is $-50\rm\;km\;s^{-1}$ for each of the simulations; the velocity dispersion is $240\rm\;km\;s^{-1}$ for the LSR representation, and $110\rm\;km\;s^{-1}$ for the GSR and LGSR ones, in rough accordance with the observed situation. The obscuration by Galactic emission was simulated by removing that part of the sample for which the deviation velocity (measured with respect to the LSR) is less than $70\rm\;km\;s^{-1}$, based on a model of Galactic kinematics which incorporates observed properties of the gaseous disk, including its warp and flare. The apparent structures are a consequence of the non-uniform obscuration in position and velocity. The appearance of the H  I Zone of Avoidance differs when material is considered in different reference frames.

Figure 2: Demonstration of the influence of Galactic obscuration on observable properties of a CHVC population. The left-hand panels show the distributions of measured (after obscuration) versus actual mean velocity and velocity dispersion determined from a series of 1000 simulations of a population of 200 objects. The upper panel on the left represents a population with a Gaussian velocity distribution in the GSR reference frame, with $\mu = -50\rm\;km\;s^{-1}$ and $\sigma = 115\rm\;km\;s^{-1}$; the lower left-hand panel represents a population with a Gaussian velocity distribution in the LGSR frame, with $\mu = -55\rm\;km\;s^{-1}$ and $\sigma = 105\rm\;km\;s^{-1}$. Obscuration removes about 30% of the population and leads to both an overestimate of the dispersion and a more negative estimate of the mean velocity. The right-hand panels illustrate the degree to which populations in the GSR and LGSR frames could be distinguished via their statistical parameters. Measured and actual parameter differences between the GSR and LGSR frames are contrasted for 1000 simulated populations of 200 objects, half defined in the GSR frame and half in the LGSR frame. All populations have a mean velocity of $-50\rm\;km\;s^{-1}$ and a dispersion of $110\rm\;km\;s^{-1}$ in their reference frame. The measured parameter differences for the observed CHVC sample are indicated by the dashed lines. While the mean velocity does not provide significant distinguishing capability between the GSR and LGSR frames, the velocity dispersion does.

Figure 3: Demonstration of the effects of differing resolution and sensitivity in the LDS and HIPASS data on the extracted CHVC samples.   Upper left:  comparison of the velocity FWHM for the CHVCs found in the HIPASS catalog (black line) with the velocity widths of those found in the LDS catalog (red line). The velocity resolution of the LDS is 1.03 km s-1, but 26 km s-1 for HIPASS; after degrading the LDS data to the HIPASS velocity resolution, the dashed red line is obtained.   Upper right:  comparison of the angular FWHM for the CHVCs in the HIPASS catalog (black line) with those in the LDS catalog (red line). The angular resolution of the LDS is $36\hbox {$^\prime $ }$, but $15\hbox {$^\prime $ }$ for HIPASS; the dashed black line shows the HIPASS values after convolving with the LDS beam.   Lower left:  comparison of the total flux of CHVCs found in the HIPASS catalog (black line) with those in the LDS catalog (red line). After compensating the HIPASS detection rates for the lower LDS sensitivity, the dashed black histogram is obtained. The compensation was performed using the relative detection rates in the region of survey overlap (see Fig. 4).   Lower right:  comparison of the total fluxes for the semi-isolated objects (CHVC:s and CHVC?s) entered in the HIPASS catalog (black line) with those in the LDS listing (red line). Compensation of the HIPASS detections to the corresponding LDS sensitivity gives the dashed black line. (This figure is available in color in electronic form.)