Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.atnf.csiro.au/research/multibeam/papers/Webster_iauc171.ps.gz Äàòà èçìåíåíèÿ: Mon Oct 26 03:03:39 1998 Äàòà èíäåêñèðîâàíèÿ: Sun Dec 23 10:14:46 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ð ï ð ï ð ï ð ï

The Parkes Multibeam Blind HI Survey
R.L. Webster, V. Kilborn and J.C. O'Brien
School of Physics, University of Melbourne, Parkville, Vic, Australia,
3052
L. Staveley­Smith
Australia Telescope National Facility, PO Box 76, Epping, NSW,
Australia, 2121
M.E. Putman
Mount Stromlo and Siding Springs Observatory, Australian National
University, Weston Creek PO., ACT, Australia, 2611
G. Banks
Department of Physics and Astronomy, University of Wales, Cardiff,
PO Box 913, Cardiff, CF2 3YB, Wales, UK
Abstract. A thirteen­beam HI receiver has been constructed for the
Parkes radio telescope. When this instrument is used in active scanning
mode, it can rapidly survey large areas of sky, with a relatively uniform
sensitivity. The Multibeam Working Group, comprising about 30 as­
tronomers from more than a dozen institutions, is undertaking a blind HI
survey of the entire southern sky. The status of the survey is described,
with some of the first scientific results.
1. The Survey
Surveys of large areas of sky in HI are time­consuming, and require either a
dedicated telescope (e.g. Dwingeloo Obscured Galaxy Survey (Henning et al.
1998)) or the multiplexing advantage of a multibeam system. The multibeam
receiver at Parkes is a purpose­built instrument with 13 receivers, arranged in
an hexagonal grid (Staveley­Smith 1997). The beam centres are separated by
about two beam widths. With an appropriate orientation, the full array will
Nyquist sample as the telescope scans the sky.
Two surveys are underway: the HI All­sky Survey (HIPASS) which will
survey the region ffi ! 0 ffi (though a northern extension is being discussed), and a
more sensitive Zone­of­Avoidance survey (ZOA), within 5 ffi of the galactic plane.
Details of the HIPASS survey will be described. For a more complete description
of the ZOA survey see Henning et al (1998). The HIPASS survey uses active
scanning where the telescope is driven at 1 ffi per minute in declination strips,
recording a spectrum every 5 sec. Each region of sky is scanned 5 times, with
small offsets between each set of scans. Data is band­pass corrected online, and
1

both the raw and the preprocessed data are archived (Barnes et al. 1998). On
the completion of the full set of scans, data is gridded onto the sky in cubes of
8 ffi \Theta 8 ffi , with 4 0 pixels.
In mid­1998, the survey was ¸ 40% complete. Extensive modelling of the
gridding procedure has resulted in a fast robust gridder (Barnes 1998). The
survey parameters are given in the Table 1.
Table 1. Survey parameters, updated from Staveley­Smith (1997)
Parameter Value
Declination range ffi ! 0 ffi
Equivalent Integration time 500 sec
Velocity Range \Gamma1200 to 12; 700 km sec \Gamma1
Channel Resolution 13:2 km sec \Gamma1
Velocity Resolution 18:0 km sec \Gamma1
Positional Accuracy (3oe) ¸ 2 \Gamma 5 0
Detection Limit (3oe) 40 mJy per beam
HI Mass Limit (3oe) 1:4 \Theta 10 6 d 2
Mpc M fi
Angular Resolution ` ¸ ? 7 0
Limiting Column Density (3oe) 7 \Theta 10 17 HI cm \Gamma2 per velocity channel
Number of Data Cubes 388
At the present time, galaxy finding is by eye. Cubes are scanned in velocity­
position space for regions of enhanced HI emission. This process is both time­
consuming and statistically unquantifiable. The Multibeam Working Group is
investigating automated galaxy­finding methods which will increase the speed of
this process and provide quantifiable detection limits. Nearly all galaxy detec­
tions are unresolved on the sky. Thus we are searching for point sources which
are distributed in velocity space. Our most promising techniques are based on
wavelet transforms, using an point source profile with variable filtering in veloc­
ity space. The current flux limit for detection by eye is ¸ 4 Jykm sec \Gamma1 , which
is equivalent to ¸ 9oe for a velocity width of ¸ 70 km sec \Gamma1 . The principle is­
sue for the galaxy finder is to efficiently find galaxies in data with non­gaussian
noise. However we are also investigating new methods of filtering noise from the
data. An automated finder will provide a powerful tool with which to explore
the data cubes and will be necessary before an all­sky catalogue of galaxies can
be published.
In order to fully realise the scientific program associated with the survey,
a complimentary followup program is required. Synthesised images of selected
HI detections are being obtained at the ATCA . These images not only provide
more accurate positions for optical identifications, but also resolve the HIPASS
detections both spatially and in velocity (Kilborn et al. 1998). Optical B and
R­band images are being obtained on 1­metre telescopes at Siding Springs and
2

Figure 1. Plot of velocity width against peak intensity for 274 galax­
ies detected in the southern in the galactic pole region.
Las Campanas. Our data will also be combined with data from other all­sky
surveys at different wavelengths, eg. near­IR.
2. HI Mass Function
The HI mass function has been determined for two regions of sky which have
been fully scanned. The \Gamma74 region is an 8 ffi ­wide declination band centred on
ffi = \Gamma74. A total of 99 galaxies have been detected in 0:242 steradians of sky
or about 4% of the southern sky. Confused sources have been omitted from the
catalogue. Four different people have searched the cubes for galaxies. Figure 1
plots the peak flux of detected galaxies as a function of measured velocity width
at 20% of the peak intensity. A clear cutoff is shown at v ¸ 50 km sec \Gamma1 . This
cutoff is simply the velocity resolution of the survey. The cutoff in peak intensity
is also relatively constant, and may be due to the efficiency of the eye in detecting
relatively high peaks against a noisy background. There is a concentration of
galaxies in the lower left­hand corner of the plot, suggesting that galaxies in this
region will be missed due to the detection limits. High resolution observations
from the ATCA are being used to quantify how these observational cutoffs limit
the parameter space (inclination, velocity width, peak flux, etc) in which galaxies
are observed.
Of the galaxies detected in the \Gamma74 region, 14% are new detections and
36% have new redshifts. There are a total of ¸ 50% new HI detections. Only
those galaxies with a total flux greater than 4 Jy km sec \Gamma1 are included in the
determination of the HI mass function. The \Sigma 1
VMax method is used to determine
the mass function in bins of 0:5 in the logarithm of the mass. Distances are
3

log M HI
q
h
(Mpc
dex
)
log
­3
­1
a = -1.35
log M* = 9.5
q * = 0.014
3
Figure 2. HI mass function 99 galaxies detected in the ­74 region.
determined using the Hubble law and a value of H 0 = 100 km sec \Gamma1 Mpc \Gamma1 . A
Schechter function has been fitted by eye. The functional form is
`(x) dx = ` \Lambda x ff e \Gammax dx (1)
where `(x) is the number of galaxies per decade of mass per Mpc 3 , x = MHI =M \Lambda
where M \Lambda is the characteristic mass defining the knee of the mass function, and
ff is the low mass slope of the mass function.
The low mass end of the mass function is still poorly determined. There
are only a couple of galaxies in the lowest mass bin, and the completeness of the
sample has not been determined. In addition, the \Sigma 1
VMax method is sensitive
to clustering, whereas a more robust method, such as a maximum likelihood
method is not. The fitted values of the Schechter funtction are ` \Lambda = 0:014
galaxies per Mpc 3 per decade of mass, log 10 M \Lambda = 9:5 and ff = \Gamma1:35. The HI
mass function is plotted in Figure 2. These values can be compared with other
recent determinations. For example, Zwann et al. (1997) find ff = \Gamma1:2 and
log 10
M \Lambda = 9:55.
Finally, and importantly, so far all of the HI detections have an optical
counterpart. Thus we are not detecting HI galaxies which do not contain stars,
to the limiting HI surface density of the survey. This is an important result for
theories of star formation.
3. Cen A Group of Galaxies
The Cen A group of galaxies was extensively surveyed by C“ot'e, Freeman and
Quinn (1997). Optical candidates were selected from UK Schmidt plates, and
followed up with pointed observations at the Parkes radio telescope. This
4

Figure 3. Members in the Cen A group, where the new HIPASS
detections are the larger symbols. Both a spatial plot and a plot in
velocity space are shown.
work provides a good basis to test the multibeam's survey capabilities. A
total of 29 galaxies were detected in the HIPASS data in the veolcity range
200 \Gamma 900 km sec \Gamma1 . Of these, 18 had previously been detected, principally by
C“ot'e et al. 10 new galaxies were found with fluxes greater than our fiducial
cutoff of 4 Jy km sec \Gamma1 . Of these 5 were previously catalogued, and 5 were not.
Thus the HIPASS observations have increased the number of galaxies detected
in the group by ¸ 50%. Figure 3 shows the spatial distribution of galaxies in
the Cen A group, with the new members indicated by the larger symbols. Fig­
ure 3 also plots the group members as a function of velocity. From this figure it
appears that putative members at the extreme ends of the velocity range may
not belong to the gravitationally relaxed core of the group. Indeed if traditional
cluster member algorithms are applied (Yahil and Vidal 1977), then only those
galaxies in the velocity range 300 \Gamma 700 km sec \Gamma1 satisfy the 3oe cutoff criterion.
A Schechter function can be fitted to mass distribution for the galaxies in the
group, assuming that they are all located at a distance of 3:5 Mpc. Values of
log 10
M \Lambda = 9:3 and ff = 1:3 are obtained.
In addition, the region around all the optically verified members of the
group was exhaustively searched by eye, resulting in a detection of ESO 272­
G025 with a flux of 1:2 Jykm sec \Gamma1 , which is a ¸ 3oe detection. Figure 4 shows
the HIPASS profile of this galaxy. A blind eye search would not have found
this galaxy, however this detection provides a benchmark with which automated
finding algorithms can be tested.
4. Galaxy Formation
Our preliminary datasets all give a slope for the low mass end of the mass
function of ff ¸ \Gamma1:3, irrespective of whether we search in the field or in a
5

000000000000000000
000000000000000000
000000000000000000
000000000000000000
000000000000000000
000000000000000000
000000000000000000
111111111111111111
111111111111111111
111111111111111111
111111111111111111
111111111111111111
111111111111111111
111111111111111111
Figure 4. HIPASS profile for ESO272­G025, an optically detected
member of the Cen A group. This is a 3oe detection.
denser galaxy environment. Improved galaxy detection algorithms may increase
the detections at the low mass end and a more sophisticated treatment of the
incompleteness will certainly provide a more robust measurement of ff. Thomas
(1997) has predicted the shape of the mass function under a range of cosmological
assumptions. In a standard CDM cosmology, if the baryons are located in CDM
halos, then using the Press­Schechter formalism, a low mass slope of ff = \Gamma1:8
is calculated for the mass function. The optical galaxy luminosity function has
a faint end slope of ff = \Gamma1:25 and this function can be matched to the CDM
halo mass function at the knee if a M=L = 15h M fi =L fi is assumed for normal
galaxies. At the high mass end, the dynamical timescale is greater than the
cooling time. Thus we would not expect massive optically luminous galaxies to
form. At the low mass end, the CDM halos have a very different distribution
from the optically catalogued galaxies. This would result if there was a natural
bias of light with respect to the dark matter. In addition, if the low mass end
of the HI mass function has a different slope to the faint end of the luminosity
function, this would support the idea that the star formation rate in HI halos
is not simply a function of total mass. However a full exploration of the HI
parameters of detected galaxies is required before strong conclusions can be
drawn from these results.
5. Dynamics of the Local Group
With the completion of the first of the five scans, a full mosaic of the southern
galactic cap (ffi ¸ ! 62 ffi ) was generated (Putman et al. 1998). In order to improve
the detection of HI structure near the Milky Way, the raw data was reprocessed
using a modified bandpass correction (Barnes 1998).
The origin of the Magellanic Stream has remained controversial since its
discovery by Matthewson et al. (1977). Currently two models for the stream
are considered viable: tidal distortion which predicts a leading arm in conjuction
with the trailling Magellanic Stream, and ram pressure stripping by gas in the
halo of the Milky Way. Figure 5 shows the full mosaic of the HI distribution.
A leading arm is clearly seen stretching from the bridge between the Large
6

SMC
Bridge
Stream
LMC
Leading Arm
Galactic Plane
Figure 5. The southern galactic cap as viewed by HIPASS. Different
features are labelled.
Magellanic Cloud and the Small Magellanic Cloud towards the Milky Way. This
new image is a substantial piece of evidence in favour of the tidal theory for
the formation of the Magellanic Stream. This model of the interaction of the
LMC/SMC system with the Milky Way supports the idea of Moore et al. (1998)
that small disk galaxies are tidally harassed by the larger neighbours in their
local environment.
Interestingly, if we also look at the image of the Large Magellanic Cloud in
HI, it appears to be a regular spiral galaxy. In optical light, only the bulge is
readily visible, and the galaxy is classified as a Irregular. Thus the galaxy mor­
phology based on HI is different from the optical morphology. In addition there
is little evidence of tidal distortion in HI distribution in the Large Magellanic
Cloud.
6. Discussion
The HIPASS survey will be available as a public database, with the first cubes
released in late 1998. So far only about 4% of the southern sky has been com­
pleted, and the data searched albeit by eye. Our planned scientific program is
wide­reaching: an HI mass function derived from ? 4000 galaxies with masses
? 10 6 M fi ; a DEEP survey, pushing the limits of detectibility within the con­
straints of the scanning technique; and a bivariate brightness distribution, based
on HI detections with followup optical imaging. The survey will provide a full
inventory of the local HI distribution for regions with ¸ ? 10 18 HI cm \Gamma2 -- an ideal
7

laboratory for the study of galaxy formation. Progress on the survey can be
found at the website:
http://www.atnf.csiro.au/research/multibeam/multibeam.html.
Acknowledgments. The HIPASS survey is the result of the collaboration
of the Multibeam Working Group. The Parkes Observatory staff are thanked
for their continued assistance in the survey.
References
Barnes, D.G. 1998, PhD thesis, University of Melbourne
Barnes, D.G., Staveley­Smith, L., Ye, T. and Oosterloo, T. 1998 ADASS VII
(San Franciso)
Henning, P.A. et al, 1998, AJ in press
Henning, P.A., Staveley­Smith,L., Kraan­Korteweg, R.C. and Sadler, E.M.,
these proceedings
Kilborn, V., de Blok, E., Staveley­Smith, L. and Webster, R.L. 1998, these
proceedings
Matthewson, D.S., Schwarz, M.P. and Murray, J.D. 1977 ApJ 217, L5
Moor, B., Lake, G. and Katz, N. 1998 ApJ 495, 139
Putman, M.E. et al. 1998, Nature 394, 752
Staveley­Smith, L. 1997, PASA, 14, 111
Thomas, P.A. 1997 PASA, 14, 25
Yahil, A and Vidal, N.V. 1977 ApJ 214, 347
Zwaan, M., Briggs, F. and Sprayberry, D. ApJ 490, 173
8