The March Sky


The Sloan Digital Sky Survey (SDSS)

Dr Loveday set out the aims of the survey: it was to make a definitive map of the local universe over one quarter of the celestial sphere, concentrated primarily around the northern galactic pole. Five spectral bands were to be covered: three optical, one near IR, and one near UV. It would catalogue 50 million galaxies down to mag 22, obtaining spectra for a million of them. The collaboration included 150 scientists from Japanese, American and European institutions. The observing site had been selected as the Apache Point Observatory in New Mexico, with the primary instrument a wide-angle reflector with 2.5m primary mirror with 3° field, on an alt-azimuth mount with roll-off enclosure. Two spectrographs would be used to undertake the spectral component of the survey.

The sky was to be surveyed in narrow longitudinal strips. By surveying along these strips at the same rate as the Earth's rotation, it was possible to minimise movement of the instrumentation, and thus measure continuously, making maximal use of the available observing time. The integration of each plate lasted 55 seconds, with cameras tuned to each spectral band staggered in east-west orientation in the image plane, such that, as the Earth rotated, and the right-ascension of observation increased, objects proceeded from one camera to the next. All of the electronics worked simultaneously, with different areas of sky being measured in each colour. The cameras consisted of four-million-pixel 2-inch square CCD arrays cooled by liquid nitrogen to -80°C. The totality of the proposed survey area was 45 such strips, each taking roughly two nights. Thus the project would take place over 90 perfectly clear nights, which at the Apache Point Observatory are around 50% of all nights.

The spectroscopic component of the SDSS was achieved using a metal plate filter in the image plane of the telescope, with holes drilled at the positions of sources, and fibre-optics used to pipe light to a spectrograph. The fibres had to be placed by hand, and with up to 640 fibres connected to any one such filter, the average turnaround time to set up each plate was around 15 minutes. The spectrometers covered the range 3900-9100Å with light throughput efficiency 20-25%.

The sample of objects for spectroscopic examination included the brightest 900,000 galaxies in the survey, those down to mag 17.77, and around 100,000 luminous red galaxies, down to mag 19.5. These objects correspond to those out to around redshift 0.4. As of 2003 March, 60% of the survey area had been charted, with 370,000 spectra recorded.

The first public data release, "Data Release 1", would include 2067 square degrees of survey (25%) and 186,240 spectra. Publication had been scheduled for January, but had been postponed after a bug was found in the magnitude calibration. It was presently on schedule for a March 31 release. The available data products were to include survey images in FITS and JPEG format, spectra in FITS and GIF format, raw object lists, and finding charts.

Dr Loveday demonstrated how the survey data could be plotted onto pie diagrams with the Earth at the centre, and galaxies at increasing redshifts plotted at increasing distances from the centre, with redshift 0.1 on the edge of the plot. This revealed a frothy distribution of galaxies with voids and great-wall-type features bounding them. A decrease in the density of galaxies at increasing redshift was accounted for because the limiting absolute magnitude of observation deteriorated with distance. Such information about galactic distributions was useful in constraining our modelling of galactic evolution, as well as the values of cosmological parameters. In theory, the distribution of galaxies is the result of the gravitational interaction of luminous matter with dark matter – an interaction of which the details remain highly speculative at the present time. However, by modelling the kinds of distribution we would expect from different kinds of interaction, Dr Loveday explained how we could use the SDSS to select the most favoured interaction models.

Scientific highlights of the data included the identification of several luminous quasars beyond redshift 5. Furthermore, a surprising feature of the overall distribution of galaxies was a trend of brightening average absolute magnitude and decreasing density at higher redshifts, which remained after account had been taken for line-of-sight effects. A possible explanation was a local underdensity extending out to redshift 0.3. If real, the presence of structure on such large scales would be a result of profound cosmological consequence. However, there was close correspondence between the density/redshift relationship observed in north and south hemispheres, placing us very close to the centre of the proposed underdensity. This seemed remarkably coincidental, and a more attractive explanation was therefore a significant time-evolution of the galactic distribution between redshift 0.1 and the present day. This in itself was equally profound: galactic formation models suggest such structures to develop well before redshift 1, and hence we would not necessarily have expected to observe a great deal of change since redshift 0.1.

Following prolonged applause for Dr Loveday's comprehensive and clear account of the SDSS, the President adjourned the meeting until May 21 at the same venue.


Dominic Ford

© 2003 Dominic Ford / The British Astronomical Association.




Color scheme