HIPASS Data Release Help Page
The `HI Parkes All Sky Survey' (HIPASS) covers the whole southern sky as
well as northern declinations up to +25 degrees. HIPASS commenced in Feb
1997, and the last northern scans were taken in 2002. HIPASS was carried
out with the Australia Telescope National Facility's
Parkes 64-m telescope
equipped with a novel
21-cm multibeam system. The latter comprises a
cooled, 13 beam receiver and digital correlator (Staveley-Smith et al.
1996, Wilson et al. 1997 and Sinclair et al. 1997). The 64 MHz observing
band, centred on 1394.5 MHz, was divided into 1024 channels, providing a
velocity range of -1280 < cz < 12700 km/s with a channel width of approx.
13.2 km/s and a velocity resolution of 18 km/s. The HIPASS integration
time per beam is 450s. After gridding an r.m.s. noise of about 13.3 mJy
was achieved in the first set of HIPASS cubes. The observations and
techniques used to calibrate and image the data are described in Barnes
et al. (2001). The processing algorithms are successfully designed to be
statistically robust to the presence of interference signals and are
particular to imaging point (or nearly point) sources, not extended
emission. Specifically, a major improvement in image quality is obtained
by designing a median-gridding algorithm which uses the median estimator
in place of the mean estimator. While the average Parkes beam FWHM is
14.3 arcmin, the average gridded HIPASS beam is about 15.5 arcmin.
The full data release contains data from 538 data cubes (size: 8o
x 8o, 388 over the entire southern sky (DEC < +2o),
and 150 over the northern sky, (DEC < +25o).
The one-dimensional spectral data for a given position is available for
downloading in a variety of different formats.
The multibeam receiver is described by
Staveley-Smith et al. (1996) (PASA, 13, 243). The observations and data
reduction are described in Barnes et al (2001) (MNRAS,
The angular resolution of the gridded HIPASS data is 15.5 arcmin.
The channel spacing is 13.2 km s-1 and the velocity
resolution is 18.0 km s-1 (or 27.0 km s-1
if Hanning smoothing is applied).
This field takes the name of the object which you are searching for. When you
enter the name, spaces are not important with the search accepting both:
- HIPASS Spectra: due to space limitations, the grid spacing of the spectra available for downloading is currently 8 arcmin. You will be presented with the nearest HI spectrum to your requested position.
- HIPASS Cubes: are now available in CASDA.
NGC 5253 and NGC5253
Galaxy positions can be obtained from (a) the Lyon/Meudon Extragalactic Database (LEDA)
(a locally-stored 1996 version) or (b) the NASA Extragalactic Database (NED)
The SEARCH button initiates the catalogue search. If successful,
co-ordinates are entered in the RA and DEC box below. You can then
proceed to set the plot parameters, and press the PLOT button.
The RA and DEC field refers to the Right Ascension (RA) and Declination (DEC)
co-ordinates of the position whose HI spectrum you want to plot.
There are a two different ways you can enter these co-ordinates:
The nearest available spectrum will be plotted. The current pixel size is
8 arcmin, and the post-gridding resolution of the data is 15.5 arcmin.
Currently, J2000 is supported.
The x-axis can either be velocity (cz) in units of km s-1 or
frequency in units of MHz. In either case the reference frame is the
solar system barycentre (heliocentric).
The lower and upper limits to the x-axis range in the plotted units.
The plot style can either be "Curve" (data points joined by straight
lines) or "Histogram".
The data can be plotted in its original form (velocity resolution 18.0
km s-1), or can be Hanning-smoothed (velocity resolution 26.0
km s-1). In both cases, the velocity spacing between channels
is 13.2 km s-1.
The PLOT button initiates the plotting of the HI spectrum at the requested
position, with the requested plot parameters. If the spectrum is
unavailable or not yet released, a message is printed below the RA,DEC
entry box. Otherwise, the HI spectrum is displayed, and various download
options are listed.
Once a satisfactory HI spectrum has been plotted, it may be downloaded
in a variety of formats. The download name suggested to your browser
has got the source coordinates embedded:
e.g. HIPASS_132157-3637.ext , where ext is
one of gif, ps, ps.gz, txt or
fits. This does NOT reflect the official HIPASS name of the source.
GIF format (typical size 7k).
Postscript format (typical size 26k).
Compressed (gzip) postscript format (typical size 6k).
Ascii format (typical size 44k). The data is in 3 columns: channel
number, velocity (km s-1 in optical cz convention) or
frequency (MHz), and flux density in Jy beam-1. The
reference frame for the spectral axis is the solar system barycentre.
One-dimensional image FITS format
(typical size 12k). The first axis contains the spectral data. The
second and third axes are dummy axes (i.e. of length one pixel)
containing the requested RA and Decl., respectively (not the actual RA
and Decl. of the nearest pixel).
Around zero velocity, or 1420.4 MHz, there are strong positive and
negative spectral lines. These are real signals produced by Galactic
hydrogen. Because the bandpass calibration of HIPASS data involves
spatial filtering, these data are not useful for measuring column
densities for Galactic gas. The positive and negative signals represent
the deviation, at the current position, from the median signal level
calculated in a declination strip of length 4o, centred at
the current position. So, a negative signal usually does not imply
Although interference is highly suppressed in HIPASS data, the final data
product still contains traces of interference, which you need to be wary of.
The prime interfering line is the 11th harmonic of the 128 MHz sampler clock
at 1408 MHz, or cz=2640 km s-1. This is a narrow line, although
Doppler corrections may of up to 0.15 MHz or 30 km s-1 may
broaden and shift this line. Interference from the GPSL3 beacon at 1380 MHz,
or cz=8800 km s-1 is only sporadic, but may bias the noise level
over a substantial frequency range (+/- 1 MHz).
Other residual narrow-band signals may be present in HIPASS
cubes, notably near 1400 MHz, or 4400 km s-1. Cubes in
the Sculptor region and the equatorial strip are known to contain
more interference lines. If you
suspect an interference signal, the best way to check is
to examine a few nearby spectrum (say about a degree away in position).
Radio continuum sources are suppressed in the data reduction procedure,
so that the median level of any spectrum should be close to zero flux
density. However, there are residual features which appear at the position
of very strong sources (e.g. Cen A itself). These features are commonly
a baseline curvature, a quasi-periodic 5.8 MHz ripple (arising
from the standing wave between the focus cabin and the vertex of the
Parkes telescope), and an increase of rms noise due to the increase of
The PMN catalogue, although at the higher frequency of 5 GHz gives a good
indication of the presence of discrete continuum sources. This catalogue is
available through NED.
For an unresolved source located at the centre of a pixel, the flux scale is
accurate to about 5%. If your source is not at the centre of a pixel, there
will be some flux loss due to the nearest-neighbour sampling algorithm. The
pixel size is 8 arcmin for data released on the web.
- When you have the full RA and DEC co-ordinates, enter
"hh mm ss" and "dd mm ss" respectively
eg. 10 01 00 and -87 00 00
- Colon (":") seperated co-ordinates are also accepted,
eg. 10:01:00 and -87:00:00
- Trailing zeros can be omitted,
eg. instead of "-87 00 00", "-87" is sufficient.
For resolved sources, the TOTAL flux density, spatially integrated
over the source, may greatly exceed the flux density per beam (which
is a surface brightness). For such cases, spatially integration of the
cube is the only way of obtaining an accurate flux. This functionality
is not yet provided.
On very bright radio continuum sources (flux densities greater than
about 100 Jy), there is saturation in the total power measurement
system, leading to an inaccurate and unpredictable flux density scale.
The velocity scale of the HIPASS data plotted here is with respect to the
solar barycentre (commonly, heliocentric) and in the usual optical
(cz) convention. Note that the spacing, in km s-1, between adjacent
channels changes with velocity.
The frequency scale of the HIPASS data plotted here is with respect to the
solar barycentre (commonly, heliocentric).
The canonical RMS noise in HIPASS cubes is 13-14 mJy beam-1,
but higher noise may be measured for cubes near the Galactic Plane and
in the north. The following figure is a guide to the RMS (in Jy) for each
cube. The HIPASS cube numbers appear under "Object" in the public data
release spectrum, though can also be found
This data is provided under the auspices of the Multibeam Survey Working Group.
The data version number is provided with FITS and ascii data files.
Users of this facility are requested to (1) acknowledge the ATNF in any
publications as follows:
The Parkes telescope is part of the Australia Telescope which is funded
by the Commonwealth of Australia for operation as a National Facility
managed by CSIRO.
and (2) cite an appropriate HIPASS observations and data reduction paper,
Barnes et al. (2001) [MNRAS 322, 486] and relevant HIPASS Catalogs.
Where possible, authors are requested to include one of the terms `ATNF',
or `Parkes Telescope', in the ABSTRACT of their
papers. This is to facilitate electronic searches for publications that
include ATNF data.
Data provided by the Multibeam Survey Working Group.
Pages created by M. Marquarding and L. Staveley-Smith.
Last Modified: 29-March-2010 by B. Koribalski