The effects of both obscuration by the gaseous disk of the Galaxy and the limited sensitivity of currently available H I surveys have important consequences for the observed properties of the HVC phenomenon. We have identified those consequences in this paper. Obscuration leads to apparent localized enhancements of object density, as well as to systematic kinematic trends that need not be inherent to the population of CHVCs. A varying resolution and sensitivity over the sky substantially complicates the interpretation of the observed distributions. Taking account of both these effects in a realistic manner is crucial to assessing the viability of models for the origin and deployment of the anomalous-velocity H I. Our discussion leads to specific predictions for the numbers and kinematics of faint CHVCs which can be tested in future H I surveys.
As shown in Sect. 5, a straightforward empirical model in which CHVCs are dispersed throughout an extended halo centered on the Galaxy does not provide the means to discriminate between distances typical of the Galactic Halo and those of the Local Group. Comparable fit quality is realized for distance dispersions ranging from about 30 to 300 kpc. In addition to requiring a relatively large number of free parameters, such empirical models beg a number of serious physical questions. In the first instance: how is it that H I clouds can survive at all in a low-pressure, high-radiation-density environment without the pressure support given by a dark halo? Presumably such "naked'' Galactic Halo H I clouds would either be very short-lived or require continuous replenishment, since the timescales for reaching thermal and pressure equilibrium are only about 107 years (Wolfire et al. 1995). Realistic assessment of such a scenario must await more detailed simulations that track the long-term fate of gas, for example after tidal stripping from the LMC/SMC, within the Galactic Halo. Only by including more physics will it be possible to reduce the number of free parameters and determine meaningful constraints on this type of scenario. This class of model also suffers from a number of shortcomings in describing the observed distributions, namely that the object density enhancement coupled with high negative velocities seen in the Local Group barycenter direction are not reproduced.
The Galactic Halo simulations returned formally acceptable values of characteristic distance as low as some 30 kpc. There is, however, a growing body of independent evidence based on high-resolution imaging of a limited number of individual CHVCs that such nearby distances do not apply. Braun & Burton (2000) discussed evidence from Westerbork synthesis observations of rotating cores in CHVC? 204.2 + 29.8 + 075 (using the de Heij et al. 2002 notation for a semi-isolated source) whose internal kinematics could be well modeled by rotation curves in flattened disk systems within cold dark matter halos as parameterized by Navarro et al. (1997), if at a distance of at least several hundred kpc. Similar distances were indicated for CHVC? 115.4 + 13.4 - 260 on the basis of dynamical stability and crossing-time arguments regarding the several cores observed with different systemic velocities, but embedded in a common diffuse envelope. The WSRT data for CHVC 125.3 + 41.3 - 205 likewise supported distances of several hundred kpc, based on a volume-density constraint stemming from the observed upper limit to the kinetic temperature of 85 K. Burton et al. (2001) found evidence in Arecibo imaging of ten CHVCs for exponential edge profiles of the individual objects: the outer envelopes of the CHVCs are not tidally truncated and thus are likely to lie at substantial distances from the Milky Way. For plausible values of the thermal pressure at the core/halo interface, these edge profiles support distance estimates which range between 150 and 850 kpc.
The Local Group deployment models of Sect. 4 offer a more self-consistent and physically motivated scenario for the CHVC population. Dark-matter halos provide the gravitational confinement needed to produce a two-phase atomic medium with cool H I condensations within warm H I envelopes, and provide in addition the necessary protection against ram-pressure and tidal stripping to allow long-term survival. The kinematics of the population follow directly from an assumed passive evolution within the Local Group potential. While three free parameters (the distance scalelength, the mass function slope, and the upper mass cut-off) were then tuned to explore consistency with the observations, only the distance was effectively a "free'' parameter. The mass function slopes of the best fits have values of -1.7 to -1.8, in rough agreement with the value of -1.6 favored by Chiu et al. (2001) for the distribution of the baryonic masses in their cosmological simulations. The somewhat steeper slopes and therefore larger baryonic fractions favored by our model fits might be accomodated by recondensation onto the dark-matter halos at later times.
The H I upper mass cut-off introduced in the Local Group models can
also be externally constrained. In addition to satisfying the
observational demand that no H I column densities exceeding a few times
1020 cm-2 are seen in the CHVC population (consistent with
the absence of current internal star formation), there is the observed
lower limit of about
for the H I mass seen in
a large sample of late-type dwarf galaxies (Swaters
1999). The upper mass cut-off favored by the simulations, of
about
,
is essentially unavoidable given these two
constraints.
The spatial Gaussian dispersion which is favored by these simulations is quite tightly constrained to lie between about 150 and 200 kpc. The implication for the distribution of object distances is illustrated in Fig. 27 in the form of a histogram of the detected objects from model #9. The distribution has a broad peak extending from about 200 to 450 kpc with a few outliers extending out to 1 Mpc due primarily to the M 31 sub-population. The filled circles in the figure are the distance estimates for individual CHVCs found by Braun & Burton (2000) and Burton et al. (2001). Although very few in number, these estimates appear consistent with the model distribution, also peaking in number near 250 kpc.
We have made the simplifying assumption that the baryonic matter in our
model clouds is exclusively in the form of H I , rather than being
partially ionized. It is reassuring that the best-fitting models have
peak column densities which are sufficiently high that the objects
should be self-shielding to the extragalactic ionizing radiation field
for
as noted above. Since the
neutral component requires a power-law slope of about -1.7 to fit
the data, it seems likely that the total baryonic mass distribution
might follow an even steeper distribution, since the mass fraction of
ionized gas will increase toward lower masses.
An interesting question to consider is whether the extrapolated mass
distributions of our Local Group CHVC models can also account for the
number of galaxies currently seen. In Fig. 28 we plot the
mass distribution of objects in one of the best-fitting Local Group
models, model #9 of Table 4. The thin-line
histogram gives the mass distribution of the model population after
accounting for the effects of ram-pressure and tidal stripping. The
thick-line histogram gives the observed CHVC distribution that results
from applying the effects of Galactic obscuration and sensitivity
limitations appropriate to the LDS and HIPASS properties in the
northern and southern hemispheres, respectively. The hatched histogram
gives the inferred total baryonic (H I plus stellar) mass distribution
of the Local Group galaxies tabulated by Mateo (1998), assuming
a stellar mass-to-light ratio of
.
M 31 and the Galaxy, with baryonic masses of
some 10
,
are not included in the plot. The diagonal
line in the figure has the slope of the model H I mass function of
.
The figure demonstrates that the low-mass populations of
these models are roughly in keeping with what is expected from the
number of massive galaxies together with a constant mass function slope
of about
.
At intermediate masses,
107-10
,
there is a small deficit of cataloged
Local Group objects relative to this extrapolated distribution, while
at higher masses there is a small excess. Conceivably this may be the
result of galaxy evolution by mergers.
It is important to note that the distribution of objects shown in
Fig. 28 is only the current relic of a much more extensive
parent population. As shown in Table 5, about 75% of
the CHVC population in these models is predicted to have been disrupted
by ram pressure or tidal stripping over a Hubble time, contributing about
of baryons to the Local Group environment and
the major galaxies.
One of the most suggestive attributes of the CHVC population in favor
of a Local Group deployment is the modest concentration of objects
which are currently detected in the general direction of M 31, i.e.
in the direction of the Local Group barycenter. These objects have
extreme negative velocities in the GSR reference frame. While this is a
natural consequence of the Local Group models it does not follow from
the empirical Galactic halo models, nor is it a consequence of
obscuration by Galactic H I. Putman & Moore (2002)
have made some comparisons between numerical simulations of dark matter
mini-halos in the Local Group with the
distributions of HVCs and CHVCs, and were led to reject the possibility
of CHVC deployment throughout the Local Group. Our discussion here has
shown that such comparisons require taking explicit account of
detection thresholds in the available survey observations, as well as
of the vagaries of obscuration caused by the H I Zone of Avoidance. The
Putman & Moore investigation did not take these matters into account.
The modest apparent amplitude of the M 31 concentration relative to
the Galactic population as seen with present survey sensitivities
provides the best current constraints on the global distance scale of
the CHVC ensemble. There follows a testable prediction, namely that
with increased sensitivity a larger fraction of the M 31 population of
CHVCs should be detected. This prediction was made explicit in
Fig. 23, where one of our model distributions was
shown as it would have been detected if HIPASS sensitivity were
available in the northern sky. For that particular model, some 250
additional detected objects are predicted, of which the majority are
concentrated in the
region centered on M 31. The
ongoing HIJASS survey of the sky north of
(Kilborn
2002), which is being carried out using the 76-m Lovell
Telescope at Jodrell Bank to about the same velocity coverage, angular
resolution, and sensitivity as the HIPASS effort, should allow this
prediction to be tested.
We have omitted the part of the sky around the south Galactic pole in
our fitting of Local Group models to the observations, because of the
extreme velocity dispersions measured in this direction. The
nearest external group of galaxies, the Sculptor Group, is located in
the direction of the south Galactic pole. If the CHVCs are distributed
around the major Local Group galaxies, then plausibly the same sort of
objects could be present in the Sculptor Group. Putman et al.
(2002) mention detection of clouds in the direction of the
southern part of the Sculptor Group. Because no similar clouds were
detected in the northern part of this Group, they consider it unlikely
that this concentration of CHVCs is associated with the Sculptor Group.
We note, however, that rather than being a spherical concentration of
galaxies, the Sculptor Group has an extended filamentary morphology,
which ranges in distance from
in the south
to
in the north. Putman et al. assumed that the HIPASS
sensitivity would allow detection of H I masses
of
throughout the Sculptor Group. But in
Fig. 13 we show the actual distance out to which
HIPASS can detect H I masses given a realistic cloud model and
detection threshold: even the most massive and rare objects in our
simulated distributions, with
,
can
only be detected out to
.
It is therefore only the near
portion of the Sculptor filament that might be expected to show any
enhancement in CHVC density with the currently available sensitivities.
It is also interesting to consider whether the simulated Local Group
model populations would be observable in external galaxy groups at even
larger distances. In Fig. 29 we show one of our
best-fitting Local Group models, model #9 of
Table 4, projected onto a plane as in
Fig. 20. In Fig. 20, the surviving
clouds were distinguished by H I flux; in Fig. 29,
the distinction is by H I mass. We indicate with grey dots those
objects that were deemed to have been disrupted by ram-pressure or
tidal stripping. The red and black dots indicate the remaining objects
in the population, with the red dots representing objects that exceed
and the black dots those that
fall below this mass limit. The choice of a limiting mass of
over a linewidth of 35 km s-1 was made to
represent what might be possible for a deep H I survey of an external
galaxy group. In this example, some 95 objects occur which exceed this
mass limit distributed over a region of some
Mpc
extent. For a limiting mass of
over 35 km s-1, the number drops to 45. It is clear that a very good mass
sensitivity will be essential to detecting such potential CHVC
populations in external galaxy groups. Current searches for such
populations, reviewed by Braun & Burton (2001), have
generally not reached a sensitivity as good as even
over 35 km s-1, so it is no surprise that such
distant CHVCs have not yet been detected.
Acknowledgements
The Westerbork Synthesis Radio Telescope is operated by the Netherlands Foundation for Research in Astronomy, under contract with the Netherlands Organization for Scientific Research.
Copyright ESO 2002