Introduction

Depth and area coverage of the SXDS

The optical imaging from the SXDS covers 1.3 square degrees to depths of B=28.2, R = 27.5, i'=27.2, and z'=26.3 (5 sigma, point source) and will be a massive improvement over anything which has gone before, more than a magnitude deeper than any survey of comparable area, and covering nearly 1,000 times the area of the Hubble Deep Fields.

• CADIS: Calor-Alto Deep Imaging Survey (Meisenheimer et al. 1998)
• CFDF: Canada-France Deep Fields (McCracken et al. 2001)
• CFRS: Canada-France Redshift Survey (Lilly et al. 1995)
• CNOC: Canadian Network for Observational Cosmology (Yee et al. 2000)
• ESO: ESO Imaging Survey (Nonino et al. 1999)
• HDF: Hubble Deep Field (Williams et al. 1996)
• NOAO: NOAO Deep Wide-Field Survey
• SDSS: Sloan Digital Sky Survey (York et al. 2000)
• VIRMOS: VLT-VIRMOS Deep Survey (Le Fevre et al. 1998)
• WHDF: Herschel Deep Field (Metcalfe et al. 2001)

Survey depth

There is thus very limited benefit in using an 8-m telescope to undertake a survey which fails to go significantly deeper than existing/planned surveys, but how deep? Athough m(AB)=26 represents the limit for optical spectroscopy, even with 8-10-m telescopes, many objects fainter than this limit will be red and therefore accessible to the future generation of infrared spectrographs, such as FMOS, and optical-infrared colors (rather than simply lower limits as would arise from a shallower optical survey) will be of great benefit in understanding the stellar populations of such objects. In addition, the rapidly-maturing technique of photometric redshift determination allows many important inferences to be obtained in the absence of spectroscopic redshifts. This requires multi-color observations, and we propose to take BRIz' images to support this technique which, when combined, with future JHK data, will provide reliable photometric redshifts.

The ultimate limit to the depth which can be reached is set by confusion, which is a sensitive function of image size (while objects at faint magnitudes are all galaxies, they are typically compact). The faint-end slope of the Hubble Deep Field number counts indicates than a telescope such as Subaru which provides 0.6-arcsec images (median image quality with Suprime-Cam) can probe ~1.5 magnitudes fainter than one which provides 0.8-arcsec images. As most future surveys will utilize existing 4-m telescopes which are, in general, unable to supply excellent optical images, it makes sense to use the unique opportunity of the Subaru Key Project to survey to magnitudes which are too faint for other telescopes. A sensible limit is 5-10 times the confusion noise, which corresponds to m(AB)=28. We therefore wish to reach this limit in the R-band, with the limits for the other filters determined by typical source colors.

Survey size

Our observations will be of comparable depth to those of the HDFs (although, obviously, the image quality will be poorer). Even a single pointing with Suprime-Cam would produce a field two orders of magnitude larger than that of the combined HDF-N and HDF-S. This is vitally important, because the transverse extent of the HDFs are only 1 Mpc, and it is known that matter in the Universe is strongly clustered on these (and larger) scales. Consequently, attempts to measure global quantities (e.g., the star formation density as a function of redshift) are likely to produce misleading results and, for example, there is some evidence that the HDF samples a void in the star formation density at high redshift.

A single Suprime-Cam pointing will not cover the entire fields of view planned by the deep SCUBA and WFCAM surveys proposed for the SXDS field, and only covers a fraction of the XMM-Newton survey region. If the bright SCUBA sources are indeed the progenitors of massive ellipticals then they should be strongly clustered. This is an inevitable result of gravitational collapse from Gaussian initial density fluctuations. We expect that the bias for luminous proto-ellipticals will be even stronger, since here we select not just massive galaxies, but those that collapse especially early in order to generate the oldest stellar populations. This most extreme clustering corresponds to fluctuations in projected number density that are of order unity on the scale of a single SCUBA field, falling to 10% rms on 1-degree scales. Therefore, if we wish to test the hypothesis that SCUBA sources and EROs are a closely-related population, we require a survey of size approaching 1 degree. One of the primary goals of this survey is thus to measure the clustering properties of a meaningful sample of high-redshift objects over comoving scales reaching over 30 Mpc. Moreover Almaini et al. (2001) have also discovered that this large-scale structure in the X-ray population appears to be closely tied to structure seen in the SCUBA population.

Given the manifold increase in scientific value that is gained from multi-wavelength data, we believe that a larger field which encompasses the above surveys will provide the greatest impact. We therefore propose to observe a 1.3-square degree field (5 Suprime-Cam pointings) which encompasses more than 90% of the XMM-Newton survey region, and the entire SCUBA and WFCAM surveys. This field has a transverse extent of ~30 Mpc, and so will be virtually insensitive to large-scale structure in the Universe. It can therefore confidently be expected to provide an accurate measurement of the global properties of the Universe. By providing a rapid release of data products to the community, we expect most, if not all, of the follow-up studies performed on the HDF to be performed on the SXDS Field, and reveal the true properties of the high-redshift Universe.

Return to top...