The Hubble Deep Field North SCUBA Super-map—I. Submillimetre maps, sources and number counts

Colin Borys 0 1 Scott Chapman 0 Mark Halpern 1 Douglas Scott 1 0 California Institute of Technology , Pasadena, CA 91125 , USA 1 Department of Physics & Astronomy, University of British Columbia , Vancouver, BC , Canada A B S T R A C T We investigate the emission of submillimetre-wave radiation from galaxies in a 165 arcmin2 region surrounding the Hubble Deep Field North. The data were obtained from dedicated observing runs from our group and others using the SCUBA camera on the James Clerk Maxwell Telescope (JCMT), and combined using techniques specifically developed for low signal-to-noise ratio source recovery. The resulting 'Super-map' is derived from approximately 60 shifts of JCMT time, taken in a variety of observing modes and chopping strategies, and combined here for the first time. At 850 m we detect 19 sources at >4 , including five not previously reported. We also list an additional 15 sources between 3.5 and 4.0 (where two are expected by chance). The 450-m map contains five sources at >4 . We present a new estimate of the 850- and 450-m source counts. The number of submillimetre galaxies we detect account for approximately 40 per cent of the 850-m submillimetre background, and we show that mild extrapolations can reproduce it entirely. A clustering analysis fails to detect any significant signal in this sample of SCUBA-detected objects. A companion paper describes the multiwavelength properties of the sources. 1 I N T R O D U C T I O N The Hubble Deep Field North (HDF-N) (Williams et al. 1996), a small region of the sky targeted by the Hubble Space Telescope (HST), has stimulated the study of the high-redshift Universe ever since the data were released in 1995. The original optical image of the HDF is one of the deepest ever obtained, and resolves thousands of galaxies in an area of a few arcmin2. However, since optical images capture only a narrow part of the spectrum, and since the rest-frame wavelength range detected depends on the redshift, it is necessary to supplement the HST image with data at other wavelengths in order to obtain a more complete understanding of galaxy evolution. In the years since the HDF image became public, deep pointings using radio, X-ray, near-infrared (IR) and mid-IR telescopes have been conducted. Additionally, optical spectroscopy has been carried out on hundreds of suitable objects in the field, thereby obtaining redshifts for most of the brighter HST-detected objects. The original HDF field is the size of a single WFPC2 field of view, roughly 2 2 arcmin2. HST also obtained shallower observations in fields adjacent to this, extending the region of study to roughly 6 6 arcmin2. These flanking fields have also been covered by other telescopes, and in fact over time the region associated with the HDF has been extended to approximately 10 10 arcmin2 in most wavebands. For an excellent review of the HDF-N region and its impact on the optical view of astronomy, refer to the article by Ferguson, Dickinson & Williams (2000). Despite these extensive observations, fully understanding the high-redshift Universe is hindered by the presence of dust, which re-processes radiation and emits it in the far-infrared (FIR). The importance of this is clearly demonstrated in the population of galaxies detected by the Submillimetre Common User Bolometer Array (SCUBA; Holland et al. 1999). Even with the tremendous observational effort of the past 5 years, we have made only modest progress with SCUBA-detected galaxies beyond the conclusions drawn from the pioneering work of Hughes et al. (1998) and Smail, Ivison & Blain (1997). SCUBA mapping surveys have constrained the number counts, and show that significant evolution is required in the local ultraluminous infrared galaxy (ULIRG) population in order to explain the abundance of high-redshift SCUBA sources. Having detected these sources, the goal now is to characterize them and try to answer fundamental questions concerning their nature. What powers their extreme luminosities? What is their redshift distribution? What objects do they correspond to today? One way to do this is by performing pointed photometry on a list of highredshift sources detected at other wavelengths, but this obviously M97BU65 M98AC37 M98AH06 M98BC30 M98BU61 M99AC29 M99AH05 M99BC41 M99BC42 M00AC16 M00AC21 M00AH07 M00BC01 M01AH31 P, J P P S P S J P P, J, S P P J J J introduces some biases. An alternate approach, one taken by many groups, is to perform a blank field survey, and compare the submillimetre map against images taken with other telescopes at other wavelengths. Arguably, the best field to do this is in the Hubble Deep Field region, which has more telescope time invested in observations than any other extragalactic deep field. In addition to the data already available, observations with the HST-ACS and upcoming confusion-limited SIRTF observations, make this an appealing region to target with SCUBA. For all of these reasons it is worthwhile to combine the available submillimetre data in this part of the sky. 2 O B S E RVAT I O N S Observations with SCUBA require the user to chop the secondary mirror at a rate of 7.8125 Hz between the target position and a reference (also called off) position. The difference between the signals measured at each position removes common-mode atmospheric noise. The direction and size of the chop throw is adjustable, and is typically set such that the off position does not change over time due to sky rotation. A map made in such a way will then exhibit a negative copy of the source in the off position. In raster-scan mode, the whole telescope moves on the sky in order to sample a region larger than the array size. A map made from data collected in this mode is commonly referred to as a scan-map. An alternative way to sample a large area is to piece together smaller jiggle maps. In this mode, the telescope stares toward the target and the secondary mirror steps around a dither pattern designed to fill in the space on the image plane between the conical feedhorns. Photometry mode is very similar to jigglemapping, except the dither pattern is smaller in order to spend more time observing the target object. The trade off is that the image plane is not fully sampled, and there will be gaps in a map made from photometry data. Jiggle-mapping and photometry also differ from scan mapping because they use two off positions in the chopping process. Therefore, there will be two negative echoes of each source. Our group has produced a shallow scan-map of the region at 450 and 850 m (Borys et al. 2002, hereafter Bo02) using SCUBA. Given the high profile of the region, however, groups from the UK and Hawaii also targeted the HDF to exploit the rich multiwavelength observations available. The 3 3 arcmin2 area centred on the HDF itself, studied originally by Hughes et al. (1998, hereafter H98), was recently re-analysed by Serjeant et al. (2003, hereafter S02) and we use their published da (...truncated)