2.3. The observed SEDs of dusty galaxies

Information about the submm SEDs of galaxies has been gathered from targeted mm and submm observations of samples of low-redshift far-IR-selected galaxies from the IRAS catalog, (Andreani and Franceschini, 1996; Dunne et al., 2000; Lisenfeld et al., 2000; Dunne and Eales, 2001), and from far-IR and submm observations of high-redshift galaxies (see Fig. 2). The most extensive local survey (SLUGS; Dunne et al., 2000) consists of 850-µm SCUBA observations of 104 galaxies selected from the low-redshift IRAS Bright Galaxy Sample (BGS; Soifer et al., 1987). After fitting single-temperature B SEDs to the galaxies, Dunne et al. found that = 1.3 ± 0.2 and T_d = 38 ± 3 K described the sample as a whole, with a natural dispersion in the properties from galaxy to galaxy. This IRAS-selected sample could be biased against less dusty galaxies. Dunne et al. are currently addressing this issue by observing a complementary sample of B-band selected low-redshift galaxies, which should be representative of optically luminous low-redshift galaxies as a whole.

Note, however, that when fitting only a few datapoints, there is a significant correlation between values of and T_d that can account for the data (left panel of Fig. 3). This can lead to ambiguity in the results, further emphasizing the difficulty in associating the dust mass or temperature inferred from a galaxy SED with the real physical properties of the galaxy.

Figure 3. An illustration of some of the issues involved in describing the SEDs of dusty galaxies. On the left is a probability contour plot that shows the 0.5, 5 × 10-3 and 5 × 10-5 probability contours for a fit to an SED model defined by the variable parameters and Td with a fixed value of = - 1.95, taking into account four SED datapoints for the galaxy NGC 958 as shown in the right-hand panel (Dunne and Eales, 2001). Note that 1 Jy = 10-26 W m-2 Hz-1. Note that there is a very significant degeneracy in the fitted parameters. Adding additional data points with small errors close to the peak of the SED at 200 µm reduces the extent of the probability contours by about 50%, but they remain elongated in the same direction. Note that > 2 is not expected physically. On the right the data are compared with fitted single-temperature SEDs. The solid line is the best fit to the data. The dashed lines correspond to SEDs from the ends of the probability `banana' shown in the left-hand panel. Note that without the 450-µm point, the thick dashed curve describes the best-fit SED, which is defined by a significantly greater dust temperature. This SED is similar to that of a typical luminous IR galaxy, whereas the best fitting model with all four data points is much more like the SED of the Milky Way. Note that the shift in the best-fit model on adding 450-µm data is generally less significant than in this case.

The addition of 450-µm data for 19 of the 104 galaxies in the SLUGS sample (Dunne and Eales, 2001), tends to split the galaxy SEDs into two categories: those that retain a definite 40-K spectrum after including the 450-µm data, and those for which cooler single-temperature SEDs, more similar to the SEDs of normal spiral galaxies, then provide a better fit. The first group are typically the more luminous galaxies in the sample, while the second includes 3 of the 5 lowest luminosity galaxies from the sample. Dunne and Eales (2001) propose a two-temperature model to account for the changes in light of the new 450-µm data; however, a cooler single-temperature model with a larger value of provides a fit of similar quality. The results for one of the most significantly different fits is shown in the right-hand panel of Fig. 3. With the addition of the 450-µm data, the nature of the SEDs of low-redshift, low-luminosity galaxies become more diverse. However, the more luminous galaxies, which are likely to be the most similar to typical high-redshift submm galaxies, are still described reasonably well by the original Dunne et al. (2000) 38-K SED.

An alternative approach is to determine an SED that can describe the observed flux density distribution of galaxies in the far-IR and submm wavebands, which are sensitive to galaxies at low, moderate and high redshifts (Blain et al., 1999b; Trentham et al., 1999; Barnard and Blain, 2002). Using the B functional form, values of 1.5 and T_d 40 K are required to provide a good description of the data, rather similar to the values derived for temperatures of individual low-redshift luminous dusty galaxies in Dunne et al. (2000) and Lisenfeld et al. (2000), and for both the small number of high-redshift submm-selected galaxies with known redshifts and mid-IR spectral constraints (Ivison et al., 1998a, 2000a) and typical high-redshift QSOs (for example Benford et al., 1999). These temperatures are significantly less than those determined for the most extreme high-redshift galaxies (Lewis et al., 1998), and significantly greater than the T_d = 17 K inferred from the maps of the Milky Way made using the all-sky survey from the FIRAS instrument on the Cosmic Background Explorer (COBE) satellite in the early 1990's (Reach et al., 1995). Note that there are examples of moderate-redshift infrared-selected galaxies with both hotter and colder typical dust temperatures than 40 K: see Deane and Trentham (2001) and Chapman et al. (2002d) respectively. At present it seems likely that a 40-K dust temperature is a reasonable assumption for high-redshift submm-selected galaxies.

Inevitably, however, there will be a population of hotter high-redshift galaxies (Wilman et al., 2000; Trentham and Blain, 2001). These galaxies would be underrepresented in existing submm surveys, but may make a significant contribution to the 240-µm background radiation intensity (Blain and Phillips, 2002). Further observational information to test the assumption of a 40-K dust temperature is keenly awaited. As we discuss below, in Section 2.6, the assumed dust temperature has a significant effect on the selection function of submm galaxy surveys, and on the properties that are inferred for the galaxies that are found in these surveys.