2.6. Submm-wave selection effects
Deep submm-wave observations image the high-redshift Universe with very little contamination from low-redshift galaxies, and can potentially find a population of galaxies that is quite different to those detected in conventional deep optical surveys, and which could be undetectable in these surveys. The complementarity of submm and optical observations is illustrated by the very limited overlap between galaxies detected in the deep submm-optical image shown in Fig. 1. However, submm surveys are certainly subject to selection effects. In Fig. 5 the flux-density-redshift relation for a submm-luminous galaxy with a fixed bolometric luminosity is presented as a function of its SED parameters - Td, and . The relatively minor effects of different assumed cosmological models are also shown. Changing the dust temperature has the greatest effect. The inferred luminosity of a dusty galaxy for a fixed observed submm flux density goes up by a factor of 10 if the dust temperature is doubled, at all but the very highest redshifts. There is thus a significant potential bias in submm surveys against the detection of galaxies with hotter dust temperatures for a given bolometric luminosity.
Figure 5. Flux-density-redshift relations illustrating some of the uncertainties that apply to the interpretation of submm and far-IR surveys. In the top-left panel the relatively small effects of changing the world model parameters are shown. Note that changes to the volume element and flux density received counteract each other, and so the chosen cosmology has a small effect on the interpretation of surveys. The most significant effect - that of changing the dust temperature - is shown in the top-right panel. At 175 µm for galaxies at moderate redshifts, the effect of temperature is rather small. However, at 850 µm, the effects are very significant, and must be remembered when interpreting the results of 850-µm observations: doubling the dust temperature corresponds to increasing the luminosity associated with a given flux density by a factor of about 10. The less significant effects of changing the dust emissivity index or the mid-IR spectral index are shown in the bottom-left and bottom-right panels respectively.
This effect was noted by Eales et al. (1999), when investigating the evolution of galaxies in the context of the results of deep SCUBA surveys. They suggested that the submm galaxies may be cooler than the temperatures of about 60 K usually assumed, and so their significance as a population of strongly evolving high-redshift galaxies may have been overestimated.
As discussed in Section 2.3, a cooler dust temperature of 40 K is compatible with observations of the SEDs of individual submm galaxies with confirmed redshifts detected in submm surveys (Ivison et al., 1998a, 2000a) and with the results of targeted observations of luminous low-redshift IRAS galaxies and high-redshift QSOs. If this temperature is assumed, then the inferences about galaxy evolution made from the results of submm surveys (Blain et al., 1999b, c; Eales et al., 2000; Smail et al., 2002) should be reliable. However, until a large sample of submm galaxies with redshifts and multi-waveband SEDs is available, the possibility that a cold or hot population of high-redshift dusty galaxies could be missing from or misidentified in submm surveys cannot be ruled out (Eales et al., 1999; Blain and Phillips, 2002). The possible effects on inferred luminosities of different forms of the SED shown in Fig. 5 need to be taken seriously, especially when describing the properties of individual galaxies selected in submm surveys.
There is little reliable evidence for a systematic relationship between dust temperature and redshift. Observations of low-redshift IRAS galaxies (Andreani and Franceschini, 1996; Dunne et al., 2000), indicate that any variation of dust temperature with luminosity appears to be gradual. However, there is evidence for a significant and systematic change in the temperature of dusty galaxies with a wider range of luminosities, from about 20 K for low-redshift spirals (Reach et al., 1995; Alton et al., 2000; Dunne and Eales, 2001) to about 40 K for more luminous objects typical of the galaxies detected in the IRAS survey. Temperatures of up to 110 K are found for some extremely luminous high-redshift galaxies (Lewis et al., 1998).
We stress that there could be a significant selection effect in submm surveys that depends on the range of dust temperatures in the source population. The importance of such an effect can be quantified once a complete redshift distribution is available for a submm-selected galaxy sample.