ON THE NATURE OF THE FIRST GALAXIES SELECTED AT 350 μm

Submillimeter-selected galaxies (SMGs) were discovered in pioneering lensed and blank-field surveys (e.g., Smail et al. 1997; Barger et al. 1998; Eales et al. 1999) with the 850 μm optimized Submillimeter Common User Bolometer Array (SCUBA; Holland et al. 1999) and later in similar surveys with millimeter detectors (e.g., MAMBO, BOLOCAM, etc.; Bertoldi et al. 2000; Laurent et al. 2005). SMGs mainly comprise massive, star-forming galaxies (see, e.g., Fox et al. 2002) around z ∼ 2 (Chapman et al. 2005) with the bulk of the emission generated at rest-frame far-IR wavelengths. SMGs are thus part of the IR-luminous galaxy population, which includes galaxies found in the local universe by the Infrared Astronomical Satellite (IRAS) All Sky Survey (Soifer et al. 1984; Joseph & Wright 1985; Soifer et al. 1987) and other galaxies detected in mid- and far-IR bands, most notably (in terms of number selected) with the Infrared Space Observatory (Kessler et al. 1996), the Spitzer Space Telescope (Werner et al. 2004), and the Akari Infrared Satellite (Murakami et al. 2007) out to z ∼ 2 (see, e.g., Rowan-Robinson et al. 1997; Puget et al. 1999; Aussel et al. 1999; Elbaz et al. 2002; Chary et al. 2004; Lonsdale et al. 2004; Le Floc'h et al. 2004, 2005; Yan et al. 2004; Matsuhara et al. 2006).

The majority of SMGs have been selected at long submillimeter–millimeter bands (500–1300 μm). Shorter submillimeter wavelengths (200–500 μm) are more demanding for ground-based observers. For example, on a good night at Mauna Kea, atmospheric transmission is about ∼30% at 350 μm but ≳80% at 850 μm (Serabyn et al. 1998). Despite this, the first galaxy selected purely by 350 μm emission—SMM J143206.65+341613.4 (=SSG 1, Short Submillimeter Galaxy 1)—was discovered (Khan et al. 2005) in a deep, blank survey of the Boötes Deep Field with the second generation Submillimeter High Angular Resolution Camera (SHARC II; Dowell et al. 2003; Moseley et al. 2004) along with a second detection, SMM J143206.11+341648.4 (=SSG 2; Khan 2006; Khan et al. 2007). This survey, reaching 13 mJy and currently the deepest at 350 μm, obtained the first constraints on the 350 μm source counts. Three additional 350 μm selected galaxies have been found in serendipitous SHARC II follow-up observations of 850 μm and 1100 μm SMGs in the Lockman Hole (Laurent et al. 2006; Coppin et al. 2008).

Short-wavelength submillimeter surveys are expected to principally select star-forming galaxies at 1 < z < 3 (see, e.g., Pearson & Khan 2009; Khan et al. 2007; Khan 2006), the epoch of peak star formation in the universe (e.g., Hopkins & Beacom 2006). Source count models that reproduce the observed 350 μm counts include a predominantly IR-luminous galaxy population evolving with redshift (e.g., Pearson & Khan 2009; Franceschini et al. 2009), but more observations are needed to verify these predictions.

This paper presents a detailed characterization of each of the five 350 μm selected galaxies following the approach of Khan et al. (2005). Multiwavelength data are presented, and spectral energy distribution (SED) template fitting is used to provide constraints on the photometric redshifts, thermal parameters (IR luminosity and dust temperature), and energy diagnostics. Additionally, the first mid-IR spectrum of a 350 μm selected galaxy, obtained using Spitzer's Infrared Spectrograph (Houck et al. 2004), is given. The properties of the sample are compared with 850 and 1100 μm selected SMGs. Table 1 identifies the five sources and includes short nicknames used for convenience. The WMAP first year cosmological parameters (H₀ = 71 km s⁻¹ Mpc⁻¹, Ω_m = 0.27, Ω_Λ = 0.73; Bennett et al. 2003) are used throughout this work.

2.1. Observations

The SHARC II observations, data reduction, and source extraction procedure for the Boötes Deep Field survey are discussed in detail by Khan et al. (2007). Further details of source extraction are given by Khan (2006). Monte Carlo simulations for determining the survey completeness also provide a measure of the flux boosting. Approximately, 4000 artificial sources of random intensity and position were inserted into the raw data map, then extracted using the same procedure as for the real sources (Khan et al. 2007). For recovered ⩾3σ sources with input flux densities >15 mJy, the ratio of the measured flux density to the input flux density indicates that the average flux boosting is 1.07 ± 0.23. Given the multiwavelength confirmation on SSG 1 and SSG 2, low outliers can be ruled out. Therefore, we proceed without applying a correction factor to these sources and assume that this also holds for the sources in the Lockman Hole. For the Lockman Hole, published 350 μm flux densities were combined with public B, R, I, and z photometry from the Subaru Lockman Hole survey (Dye et al. 2008), K band imaging from UKIDSS (Lawrence et al. 2007), archival Spitzer IRAC (Fazio et al. 2004) and MIPS (Rieke et al. 2004) imaging, and 1.4 GHz data from Biggs & Ivison (2006). These compiled photometric data are presented in Table 2.

Table 2. Photometry for Counterparts of 350 μm Selected Galaxies

λobs	SSG 1	SSG 2	LH 350.1	LH 350.2	LH 350.3
B (μJy)	<0.3	0.1 ± 0.02	2.1 ± 0.020	0.07 ± 0.007	0.24 ± 0.009
R	0.8 ± 0.1	0.3 ± 0.1	3.6 ± 0.033	0.27 ± 0.013	0.52 ± 0.014
I	2.2 ± 0.1	< 0.9	5.5 ± 0.051	0.47 ± 0.017	0.65 ± 0.018
z	⋅⋅⋅	⋅⋅⋅	9.2 ± 0.085	1.09 ± 0.040	0.78 ± 0.036
J	10.5 ± 1.1	< 7.9	11.7 ± 0.65	2.7 ± 0.45	1.3 ± 0.41
K	42.0 ± 1.4	< 11.5	32.2 ± 0.89	7.6 ± 0.56	<1.5
3.6 μm	73.8 ± 0.3	5.6 ± 0.3	75.5 ± 7.0	21.1 ± 2.0	10.5 ± 1.1
4.5 μm	58.5 ± 0.5	7.4 ± 0.5	74.8 ± 6.9	23.0 ± 2.2	13.2 ± 1.5
5.8 μm	37.7 ± 2.1	8.1 ± 2.1	54.2 ± 7.0	13.0 ± 4.5	16.2 ± 5.2
8.0 μm	45.8 ± 2.2	3.0 ± 2.2	80.7 ± 8.6	40.1 ± 6.0	18.9 ± 5.6
24 μm	523.8 ± 58.7	187.2 ± 29.4	576 ± 13a	204 ± 24a	<80.2b
70 μm (mJy)	< 40	< 40	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
160 μm	< 100	< 100	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
350 μm	23.2 ± 7.9	17.1 ± 6.4	28.4 ± 9.2	37.0 ± 13.4	32.8 ± 8.9
850 μm	⋅⋅⋅	⋅⋅⋅	<6	<6	<6
1100 μm	⋅⋅⋅	⋅⋅⋅	1.1 ± 1.4	−0.2 ±1.4	2.4 ± 1.4
1200 μm	0.142 ± 0.4	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
20 cm (μJy)	49.8 ± 11.4	53.1 ± 11.4	52.5 ± 5.2	32.0 ± 4.6	15.8 ± 4.8

Notes. For non-detections, the flux density at the radio position is used. Upper limits are 3σ. These data are used in the STARDUST2 SED template fitting. ^aFlux density is deblended using [3.6] μm positions. ^bFlux density cannot be deblended, used as a limit instead.

Download table as:  ASCIITypeset image

Additional 1.2 mm photometry of SSG 1 was obtained using MAMBO in 2005 January for ∼3 hr in photometry mode. The horn antenna design produces incomplete sampling of the field, and because there was no jiggling, SSG 2 was not observed. The data were reduced using the standard package (mopsic).

New VLA observations of the Boötes Deep Field were obtained on 2006 April 11. The VLA was in A configuration, giving a synthesized beam size of 136 × 150 at 1.5 GHz. Wide field imaging mode was used in order to avoid bandwidth smearing: two 25 MHz IFs of two polarizations each with seven channels per IF. The phase center was located 10'' north of SSG 1 to avoid any possible phase center artifacts. Standard wide field imaging techniques were employed, including self-calibration and three-dimensional corrections. The final image was generated with AIPS parameter robust = 1, giving an rms noise ≈10 μJy beam⁻¹. Automatic source search and flux density measurements were done with AIPS task sad. Both SSG 1 and 2 were detected, but neither was spatially resolved.

Initial Spitzer/IRAC and MIPS data were obtained during surveys of the NOAO Deep Wide Field in Boötes (Eisenhardt et al. 2004). Significantly deeper observations were later obtained in 2006 February as part of the IRAC GTO program (Program ID 520). The IRAC observations consisted of six dithered 100 s frames in each IRAC field of view, covering the SHARC II survey area in all four IRAC bands. Multiplexer bleed and other detector artifacts were removed by applying the Spitzer Science Center's (SSC) artifact mitigation code to the S14 version of the automated IRAC pipeline Basic Calibrated Data products. These cosmetically enhanced frames were then mosaiced using IRACProc v4.0beta (Schuster et al. 2006) to ensure correct treatment of the noise for both point and extended sources. The MIPS observations consisted of eight 30 s cycles at 24 μm only and covered the entire 350 μm map area. The data were reduced following standard procedures (Gordon et al. 2005), and source flux densities were measured with point spread function fitting. The 99% confidence region of SSG 1 contains two visible-wavelength sources. The new 24 μm imaging shows that the source chosen as the most likely submillimeter counterpart by Khan et al. (2005) is responsible for ∼75% of the 24 μm emission.

The Spitzer/IRS observation of SSG 1 was designed according to the recommended SSC setup. For each slit, there were six pointings along the slit at 24'' spacing, all centered on the slit in its narrow dimension. This is equivalent to the normal point source method except that the target is observed in six different slit positions instead of the usual two, lessening the effects of flat-fielding errors and bad pixels. In 3.1 hr (the total for the AOR, including peakups and overheads), there were five cycles of LL1 (19.5–38.0 μm) and seven of LL2 (14.0–21.3 μm) with 120 s ramps at each of the six slit positions. The data were reduced using the pipeline S15, cleaned using irsclean, and extracted with smart (Higdon et al. 2004). Two extraction techniques were compared: the first method involved extracting the spectra from the individual images and taking the median spectrum; the second method was to align the images, and then extract the spectrum. There was little difference in the final spectrum using either technique, and the spectrum is shown in Figure 1. The most prominent features in the SSG 1 spectrum are due to polycyclic aromatic hydrocarbons (PAHs). For quantitative analysis, PAHs can be measured from integrating the flux above a baseline, but this method tends to underestimate the flux density as compared to a profile decomposition method such as the one used by pahfit (Smith et al. 2007; see also Galliano et al. 2008 for more detailed consideration of the two approaches). While both methods give similar PAH ratios and similar spatial variations of PAH strengths, absolute flux densities will differ due to assumptions about the PAH profiles.

Zoom In Zoom Out Reset image size

2.2. Counterpart Identification

The additional deep 24 μm and 1.4 GHz imaging of the Boötes field yielded counterparts in these bands for SSG 1 and 2 with low probabilities of chance association (Khan et al. 2007). The more precise positions were then used to identify optical and mid-IR counterparts. Finding counterparts for the three sources in the Lockman Hole is not straightforward: Figure 2 shows thumbnail images of the fields at various wavelengths. As can be seen, there exist multiple radio detections for two of the 350 μm objects. However the most likely counterparts can be determined with the following rationale.

• 1.  
  LH 350.1. The 1.4 GHz image shows two potential counterparts, but only the southern one is within the 9'' SHARC II beam. In the visible images this appears to be a blend of two sources. The photometric redshift for this southerly source was obtained using hyper-z (Bolzonella et al. 2000) in conjunction with the BRIzJK [3.6][4.5] photometry, similar to the approach of Dye et al. (2008). The best-fitting solution is a starburst template at z = 1.19^+0.33_−0.14. The northern 1.4 GHz object is not associated with any visible-light detection, although it is prominent in the IRAC and MIPS images. Those images also show two more sources: an easterly object whose SED suggests it is a star or a low-redshift elliptical galaxy, and well to the west, the QSO RDS 054A at z = 2.416 (Schmidt et al. 1998), which is not expected to make a strong contribution to the 350 μm emission.
• 2.  
  LH 350.2. The 1.4 GHz image shows three sources within the beam. However, the two outermost objects are well resolved in the visible imaging, suggesting that they are likely to be relatively nearby. Photo-z fitting puts both at z ∼ 0.4, and SED template fitting puts a contribution to the 350 μm emission of <20% (assuming a dust temperature, T_d > 30 K). Therefore, the very red middle object is likely to be the predominant source of the 350 μm emission with a best-fitting photo-z of 1.21^+0.13_−0.20.
• 3.  
  LH 350.3. There is a strong 24 μm and weak 1.4 GHz source within 6'' of the 350 μm position. The B through 4.5 μm images suggest that the mid-IR flux is a combination of two sources with the radio position lining up on the redder, western object. Best-fitting photometric redshifts are z = 2.47^+0.51_−0.97 and z = 0.56^+0.28_−0.39 for the western and eastern objects, respectively. There remains the possibility of a merging system because the secondary solution for the eastern source is z = 2.60^+0.32_−0.33. Based on the radio identification, the western object is taken as the main source of the submillimeter emission.

Zoom In Zoom Out Reset image size

The multiwavelength photometry in Table 2 was used to derive photometric redshifts and thermal emission parameters. Following the same approach as Khan et al. (2005), the stardust2 template fit was used (P. F. Chanial 2009, in preparation) to simultaneously obtain the photometric redshift z_phot, dust temperature T_d, and the 8–1000 μm IR luminosity L_IR. Templates were based on local starbursts, and χ² was minimized through a Levenberg–Marquardt technique. For SSG 1, the redshift obtained for the IRS spectrum in Figure 1 was used in the fit, but z_spec = 1.05 agrees well with the z_phot = 1.03 from the template fitting. The best-fit parameters are given in Table 3, and the best-fitting templates are shown in Figure 3.

Zoom In Zoom Out Reset image size

For SSG 1, SSG 2, and LH 350.3, the 350 μm flux density is well fitted by the template, but in the cases of LH 350.2 and 3, the quality of the fit at 350 μm is clearly affected by the longer submillimeter data, which constrain the slope on the Rayleigh–Jeans side of the spectrum. Additionally, the 24 μm flux density is useful in determining which side of the peak the 350 μm data point lies. Usually, the mid-IR is not used to constrain the rest-frame far-IR continuum emission because specific dust grains and active galactic nuclei (AGNs) can have a major effect on mid-IR flux densities. Usage of the MIPS band was checked for a sample of SMGs (from Coppin et al. 2008) by comparing template fits using measurements at 350 μm, 850 μm, and radio with measurements at 24 μm, 350 μm, and radio plus long-submillimeter limits. The two approaches agree within the stardust2 uncertainties. This indicates our thermal parameters are not noticeably biased by the inclusion of the mid-IR data (see P. F. Chanial 2009, in preparation, for more details).

For LH 350.3, the lack of a defining 1.6 μm rest-frame stellar continuum feature makes a significant AGN contribution plausible, although the relatively low radio flux density would argue against this. Because of the absence of AGN templates in stardust2, the LH 350.3 data were also fitted using another routine with a mixture of AGNs and starburst templates (Negrello et al. 2009). In this template fit, the favored solution is a starburst at z_phot = 3.8 (but without including the radio flux density). Therefore, we consider LH 350.3 to be, like the rest of this sample, predominantly star forming. Given the photometric redshift estimates from hyper-z, stardust2, and Negrello et al. (2009), it is possible that LH 350.3 is the most distant galaxy so far discovered at short-submillimeter wavelengths.

3.1. Energy Source Diagnostics

The different survey depths make it complicated to produce fair statistics for number counts. The Lockman Hole surveys (Laurent et al. 2006; Coppin et al. 2008) were targeted at known sources, and sensitivity for serendipitous sources varied with distance from the target source. Even the blank field survey in Boötes had sensitivity varying with location (see Figure 3 of Khan et al. 2007). However, simply considering the flux densities of the five detections and the total survey area (15.1 arcmin²),¹⁷ we estimate resolving ∼20% of the 350 μm contribution to the cosmic infrared background (CIB; Fixsen et al. 1998) at S > 17 mJy. This fraction is reasonable given that the bulk of the background will be generated by galaxies near the break in the differential counts. It is also consistent with the 30% estimate at S > 13 mJy from the source counts of Khan et al. (2007).

A lower limit on the star formation rate density (SFRD) can be found using the thermal parameters in Table 3 and the comoving volume in 1 < z < 3. The conversion from far-IR luminosity to SFR is given by Kennicutt (1998). The result for this epoch is 0.017 M_☉ yr⁻¹ Mpc⁻³, much lower than the values presented by Hopkins & Beacom (2006). This implies that the bulk of the 350 μm contribution to the SFRD is from less luminous infrared galaxies and normal galaxies, consistent with observations of IR-luminous galaxies in different bands (e.g., Le Floc'h et al. 2005) and the predictions from source count models (e.g., Pearson & Khan 2009).

Our 350 μm selected galaxies can be compared with sources selected at longer submillimeter wavelengths. Suitable samples that include 350 and 24 μm photometry come from Coppin et al. (2008), Laurent et al. (2006), and Kovács et al. (2006). The photometric redshifts and thermal parameters for these SMGs were again found using stardust2. Where there is no spectroscopic redshift, the photo-z was obtained from optical and mid-IR magnitudes given by Dye et al. (2008) and Clements et al. (2008). Following the L_IR–T_d distribution by Blain et al. (2004), there appear to be two trends separating the predominantly local IRAS galaxies from the higher redshift SMGs (see Figure 4). Compared with those selected at 850 and 1100 μm, the 350 μm selected sources have similar dust temperatures and give no indication of temperature bias according to selection wavelength. Deeper 800–1000 μm imaging (e.g., with ALMA—Wootten & Thompson 2009) of the 350 μm selected population would provide a more direct comparison of any selection biases.

Zoom In Zoom Out Reset image size

Mid-IR surveys to probe the 1 < z < 3 galaxy population have the advantage of better angular resolution, aiding counterpart identification (and hence more reliable redshift estimation), but a significant disadvantage is that without further far-IR observations, constraints on the rest-frame far-IR thermal emission must be indirectly inferred. The 350:24 flux density–redshift distribution for the SHARC II 350 μm selected SMGs compared with the 850 and 1100 μm selected SMGs is shown in Figure 5 alongside colors from various templates from Efstathiou et al. (2000). The overall distribution suggests that all samples are drawn from the same IR-luminous galaxy population, and the good agreement with the templates suggests these are predominantly luminous and ultraluminous infrared galaxies.¹⁸

Zoom In Zoom Out Reset image size

Studies of infrared galaxies are often limited by unavailability of data at wavelengths longer than 24 μm. In such cases, the IR luminosity can be estimated using only the 24 μm flux density and various SED templates (Chary & Elbaz 2001) combined with an estimated redshift. For the SMG samples, the thermal emission estimates found this way compare surprisingly well with those from stardust2 for z ≲ 1 as shown in Figure 6. For these galaxies, the dominant errors on both luminosity predictors will arise from the photo-z fitting; these errors are highly non-Gaussian and non-trivial to determine. To some extent, the agreement in Figure 6 can be attributed to small number statistics and selection. The overall variation in the ratio over a wider redshift range could also imply a heterogenous SMG population. This would be consistent with the color in Figure 5 but with a varying degree of AGN contribution. Another condition on this sample is the requirement of bright optical, mid-IR, and most importantly radio identifications; at high redshift, an AGN will boost the radio flux density and the emission in other bands (notably in the mid-IR), allowing this type of analysis to be performed. However, none of the sources in the sample appear to be strongly affected by an AGN. At higher redshifts multiwavelength analysis becomes much more difficult. Some of the higher-z 850/1100 μm galaxies could simply have incorrectly identified 24 μm counterparts (see, e.g., Younger et al. 2007).

Zoom In Zoom Out Reset image size

The rest-frame 12 μm flux density has been suggested to be a "pivot point" for SEDs (Spinoglio et al. 1995; see also Elbaz et al. 2002), giving a fixed L₁₂/L_FIR for both starbursts and AGNs. This correlation was determined from local IRAS sources, but at z ∼ 1, the rest frame 12 μm is shifted into the 24 μm band, and thus the relation at this redshift can be examined directly by our sample. We expect this relation to hold if the populations are similar. An observed difference in the populations is that distant mid-IR sources show stronger dust grain emission features around 12 μm compared to the local IR-luminous galaxy population. This difference could be due to differing AGN fractions (Wu et al. 2009).

Figure 7 shows the ratio of far-IR luminosity to observed Spitzer/MIPS 24 μm monochromatic luminosity as a function of redshift. The ratio flattens out at z ∼ 1 as the rest-frame 12 μm emission is shifted into the MIPS 24 μm band, but the dispersion increases at higher redshift. This suggests that the local empirical far-IR/12 μm relation could apply to earlier epochs where the galaxy is undergoing rapid evolution, making rest-frame 12 μm flux densities a better tracer of IR luminosity (Spinoglio et al. 1995) than other mid-IR emission features, e.g., 7.7 μm PAH selected at z ∼ 2. At z ∼ 1.4, the silicate absorption feature is redshifted into the MIPS 24 μm band and affects the flux density measurement. This effect is emphasized in Figure 5, where an excess in the 350/24 μm colors is evident in the model SEDs, and galaxies in the redshift range where the silicate feature would cause a deficit in the 24 μm flux (1.3 < z < 1.6) are notably absent from Figure 7.

Zoom In Zoom Out Reset image size

Various studies have suggested that the far-IR luminosity can be deduced solely from mid-IR observations of galaxies. Bavouzet et al. (2008) have shown that the far-IR luminosity of dusty galaxies observed with Spitzer correlate closely with the corresponding 8 and 24 μm mid-IR luminosities. These studies were limited to redshifts z ≲ 1 by the necessity to use Spitzer bands at 70 and 160 μm to calculate the far-IR luminosity. The mid-IR to far-IR luminosity relation was tentatively extended to higher redshifts (z ∼ 2) by stacking the far-infrared fluxes in the long-wavelength Spitzer bands, and these stacked results were used by Caputi et al. (2007) to calculate the bolometric luminosity function at z ≈ 2. In contrast, our submillimeter observations directly access the high-redshift universe without the need for stacking images to obtain far-infrared fluxes. From Figure 7, we observe a tight correlation between the far- and mid-IR flux densities in our sample between 1 < z < 2, but there is a large dispersion in the infrared colors at z > 2. For z > 2, the 24 μm band is sampling rest wavelengths shorter than 8 μm, and the far-IR luminosity cannot be accurately predicted from such observations. In order to confirm the mid-IR/far-IR relation for 1 < z < 2 and to test the relation to higher redshifts, large samples of SMGs will be required. Such samples will be available following the large surveys planned with Herschel SPIRE (at 250, 350, and 500 μm; Griffin et al. 2008).

The discovery of the first 350 μm selected galaxies using SHARC II offers a different insight into IR-luminous galaxies than those selected in either the mid-IR or in longer submillimeter bands. The present sample comprises IR-luminous galaxies at 1 < z < 3 and resolves ∼20% of the 350 μm background at S > 17 mJy. The implied lower limit on the SFRD suggests that the bulk of the 350 μm contribution is from galaxies of lower IR luminosity than the ones detected at present flux densities.

The combination of 350 μm and mid-IR flux densities allows us to examine the relation of the mid- and far-IR emission in dusty galaxies at high redshift. There is surprisingly good agreement at z < 2, with the flattening of the IR:24 μm/(1 + z) luminosity ratio at z ∼ 1 evidence for the local empirical relation of rest-frame 12 μm IR luminosity tracing IR luminosity applying to distant IR-luminous galaxies. The wider dispersion at z > 2 suggests a limit to which the mid-IR can be used as a proxy for far-IR emission. We also find, by comparing our 350 μm selected sample to samples chosen at 850 μm and millimeter wavelengths, no evidence for a cold color temperature bias in SMGs. These results will be further tested through surveys with Herschel-SPIRE, which will detect SMGs in unprecedented numbers.

The present 350 μm selected sample, with 15 < S₃₅₀ < 40 mJy, is near the SPIRE 350 μm blank-field detection limit (e.g., Pearson & Khan 2009; Franceschini et al. 2009) and also the detection limits of current longer-wavelength submillimeter instruments. Counterpart identification is difficult even with the available angular resolution and will be far more difficult for space-based surveys with smaller telescopes. The achievable sensitivity and resolution of ground-based detectors will therefore provide a necessary complement to the space-based surveys, especially in characterizing the nature of the faint SMG population. Whether the first 350 μm selected galaxies are typical of the population will be probed through the combination of surveys with SPIRE and future surveys from the ground (e.g., SCUBA 2–450 μm; Holland et al. 2006).

We thank Tom Phillips and the CSO for observing time and assistance during our runs, and Darren Dowell, Colin Borys, Attila Kovacs, Rick Arendt, Dave Chuss, and Bob Silverberg for providing observing and data reduction support. We thank Glenn Laurent and Kristen Coppin for advice on their 350 μm Lockman Hole follow-up surveys, Mark Brodwin and Yen-Ting Lin for assistance in compiling the Boötes optical and near-IR photometry, and Rob Ivison for providing the radio maps for the Lockman Hole sources.

The Caltech Submillimeter Observatory is supported by NSF contract AST-0229008. This work is based in part on observations made with the Spitzer Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA. This work is based in part on data collected at Subaru Telescope, which is operated by the National Astronomical Observatory of Japan. The Subaru data are those used by Dye et al. (2008). This research has made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. This research draws upon data provided by Buell Jannuzi & Arjun Dey and by Richard Elston & Anthony Gonzalez (Elston et al. 2006) as distributed by the NOAO Science Archive. NOAO is operated by the Association of Universities for Research in Astronomy (AURA), Inc., under a cooperative agreement with the National Science Foundation. The UKIDSS project is defined by Lawrence et al. (2007). UKIDSS uses the UKIRT Wide Field Camera (WFCAM; Casali et al. 2007). We used data from the fourth data release (DR4).

S.A.K. is partially supported through FONDECYT Proyecto 1070992 and ALMA-Conicyt. Support for E.L.F.'s work was provided by NASA through the Spitzer Space Telescope Fellowship Program.

Facilities: CSO, Spitzer, UKIRT, Subaru, VLA, IRAM/MAMBO