A SUCCESSFUL BROADBAND SURVEY FOR GIANT Lyα NEBULAE. II. SPECTROSCOPIC CONFIRMATION

We targeted a total of 26 Lyα nebula candidates (15 first priority, 5 second priority, and all 6 third priority candidates). Long-slit spectroscopic observations were obtained using the MMT and the Blue Channel Spectrograph during 2007 May, 2008 April and June (Table 1). The observations used the 300 l mm⁻¹ grating with 10/15 wide slits resulting in a resolution FWHM of 2.6/3.4 Å and a wavelength range of Δλ ≈ 3100–8320 Å.

During our 2007 run, we chose slit orientations based primarily on the morphology of the candidate (i.e., aligned with the major axis of the emission as estimated from the B_W-band morphology), and as a secondary criterion attempted to intersect a nearby bright reference object if possible. In practice, however, we found that the faintness of our candidates and the short duration of our spectroscopic exposures necessitated a positional reference. As a result, slit orientations during the 2008 runs were chosen to always include a positional reference object, and consequently do not always trace the major axis of the B_W morphology. During these later runs, we dithered the target along the slit by ≈5'' between exposures to minimize the effect of any bad pixels. The full list of targeted candidates and the results of the spectroscopic observations are given in Table 2.

We reduced the spectroscopic data using IRAF.⁵ We subtracted the overscan and bias, and applied a flat-field correction using normalized observations of the internal quartz flat-field lamps. Twilight flats were used to determine the illumination correction for the science frames. We removed cosmic rays from the two-dimensional sky-subtracted data using xzap.⁶ One-dimensional spectra were generated using the task apall and optimal variance-weighted extraction (Valdes 1992); the spectral trace was determined using bright unresolved sources on the slit. We determined the wavelength solution to an rms precision of ≈0.08–0.18 Å using HeArNe and HgCd comparison lamps, and then corrected the data for any slight systematic offset using the night sky lines as a reference. The final wavelength calibration is accurate to ±0.42 Å. The relative flux calibration was based on observations of the standard stars BD+33 2642, BD+26 2606, BD+28 4211, Feige 34, and Wolf 1346.⁷ For each night, we applied a gray shift to compensate for any variable gray (i.e., independent of wavelength) extinction that may have affected a given standard star observation relative to one taken under better conditions. The sensitivity functions derived from individual standard star exposures were consistent to within ≲0.1 mag.

Due to the faintness of the candidates and the fact that we are searching for luminous Lyα nebulae at z ≈ 2–3, the aim of our follow-up spectroscopic program was to look for strong, high equivalent width line emission. A single strong line in the blue can be identified as Lyα rather than an unresolved [O ii]λλ3726,3729 doublet (the only other possibility at these wavelengths) due to the fact that a detection of [O ii] would be accompanied by stronger detections of [O iii]λλ4959,5007 and Hα as well. Candidates with strong Lyα emission at z ≈ 2–3 in the B_W band are easily detectable with the MMT/Blue Channel down to a 5σ limit of ≈1–7 × 10⁻¹⁷ in 30 minutes (assuming a Lyα line with FWHM_obs = 12 Å). This corresponds to limiting line luminosities of ≈0.3–5 × 10⁴² erg s⁻¹, which are below the typical luminosities of giant Lyα nebulae. However, continuum-only sources are much fainter and require longer integration to yield high signal-to-noise ratio spectra, more than was available during our spectroscopic campaign. To allow us to target the largest number of candidates, we therefore carried out a quick reduction of the data in real time and continued integrating on each of the 26 targeted candidates up until the point where we could either confirm the presence of a line or confirm the presence of continuum with no strong lines. The deeper spectroscopy necessary for studying the spectral properties of our confirmed Lyα nebulae in detail as well as for detecting absorption features in continuum-only sources is left to future observations with larger telescopes.

Figures 2–6 show the postage stamps, two-dimensional spectra, and one-dimensional spectra of the Lyα sources; the measured properties are listed in Table 3. The spectral extraction apertures were chosen to maximize the signal-to-noise ratio of Lyα. Redshifts were determined from the centroid of a Gaussian fit to the observed Lyα line. No correction was included for Lyα absorption, a potential source of bias in our redshift estimates. The B_W sizes, isophotal areas, and surface brightnesses were measured above the median 1σ surface brightness limit of the entire NDWFS survey from the original B_W images using SourceExtractor (detect_minarea = 5, detect_thresh = 28.9 mag arcsec⁻²; Bertin & Arnouts 1996). The Lyα sizes were measured from the two-dimensional spectra using SourceExtractor above the 1σ surface brightness limit at the location of Lyα (detect_minarea = 5, detect_thresh ≈1 × 10⁻¹⁸ erg s⁻¹ cm⁻² Å⁻¹ arcsec⁻²; see Table 3). The B_W sizes along the slit can underestimate the Lyα sizes measured from the two-dimensional spectra; in the case of PRG3, our deepest spectrum, the B_W size underestimates the Lyα size by a factor of ≳1.3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3. Lyα Nebula Measurements

Parameter	PRG1	PRG2	PRG3	PRG4	LABd05
Aperture (arcsec)	1.5 × 5.04	1.5 × 7.84	1.0 × 5.60	1.5 × 1.68	1.0 × 4.48
λLyα, obs (Å)	3249.61 ± 0.39	3971.41 ± 0.13	3813.28 ± 0.90	3511.23 ± 0.67	4444.99 ± 0.31
Redshift	1.6731 ± 0.0003	2.2668 ± 0.0001	2.1368 ± 0.0007	1.8883 ± 0.0005	2.6564 ± 0.0003
FLyα (10−17 erg s−1 cm−2)	44.1 ± 4.0	49.2 ± 1.1	10.2 ± 1.2	10.3 ± 1.2	19.0 ± 0.9
LLyα (1042 erg s−1)	8.2 ± 0.7	19.3 ± 0.4	3.5 ± 0.4	2.6 ± 0.3	10.9 ± 0.5
Lyα EWrest (Å)	257.1 ± 29.1	127.3 ± 6.3	47.1 ± 6.4	88.7 ± 11.8	115.5 ± 9.5
Lyα FWHMobs (Å)	9.19 ± 0.60	8.52 ± 0.19	23.36 ± 7.90	6.51 ± 0.89	15.44 ± 0.70
Lyα σv (km s−1)	361.2 ± 23.7	273.9 ± 6.1	782.1 ± 264.3	236.8 ± 32.2	443.3 ± 20.0
$F_{{\rm C}\,\mathsc{iv}\lambda 1550}$ (10−17 erg s−1 cm−2)a	2.1 ± 1.1	1.8 ± 0.8	<0.8	<0.8	<5.4
$F_{{\rm He}\,\mathsc{ii}\lambda 1640}$ (10−17 erg s−1 cm−2)a	5.8 ± 1.0	1.8 ± 0.9	<0.9	0.7 ± 0.5	1.4 ± 1.3
$F_{{\rm C}\,\mathsc{iii}\lambda 1909}$ (10−17 erg s−1 cm−2)a	4.8 ± 0.9	3.0 ± 1.1	<1.5	<1.4	<1.8
BW diameter along slitb (arcsec)	8.96	10.12	6.72	6.75	9.32
BW isophotal areab (arcsec2)	40.9	73.2	45.3	38.7	54.4
BW surface brightnessb (mag arcsec−2)	27.2	27.0	26.8	27.0	27.0
Lyα diameter along slitc (arcsec)	9.24	12.04	8.96	3.92	8.96
Lyα diameter along slitc (kpc)	78.3	99.0	74.4	33.0	71.3
Approximate Lyα isophotal aread (arcsec2)	43.4	103.7	80.4	>5.9e	50.3
Approximate total LLyαf (1042 erg s−1)	47.2 ± 4.3	170.2 ± 3.7	49.6 ± 6.1	>2.6e	122.8 ± 6.0

Notes. ^aLine flux upper limits are 2σ values. ^bB_W sizes, isophotal areas, and surface brightnesses measured from the NDWFS imaging using SourceExtractor with a detection threshold of 28.9 mag arcsec⁻², the median 1σ B_W surface brightness limit of NDWFS. ^cLyα sizes measured from the two-dimensional spectra using SourceExtractor with detection thresholds of [2.5, 1.0, 0.8, 2.1, 1.5]× 10⁻¹⁸ erg s⁻¹ cm⁻² Å⁻¹ arcsec⁻², the respective 1σ line surface brightness limits at the position of Lyα. ^dApproximate Lyα isophotoal area computed by correcting the B_W isophotal area $A_{B_{W}}$ by a factor of ν², where ν is the ratio of the Lyα and B_W diameters measured along the slit. ^eAs the B_W emission is not an accurate tracer of the Lyα emission in PRG4, approximate luminosity and area estimates have been replaced with lower limits derived from the spectroscopic data alone. ^fApproximate total Lyα luminosity computed by scaling the Lyα luminosity measured within the spectroscopic aperture by a geometric correction factor of $f_{{\rm geo}}=A_{B_{W}}\times \nu ^{2}/(\omega \times d)$, where $A_{B_{W}}$ is the isophotal area of the source on the B_W image, ν is the ratio of the Lyα and B_W diameters measured along the slit, ω is the slit width, and d is the spatial extent of the spectral extraction aperture.

Download table as:  ASCIITypeset image

Estimates for the total Lyα isophotal area and total Lyα luminosity are also given in Table 3. The approximate total Lyα isophotal area was estimated by correcting $A_{B_{W}}$ by a factor of ν², where $A_{B_{W}}$ is the isophotal area of the source on the B_W image and ν is the ratio of the Lyα and B_W sizes measured along the slit. The approximate total Lyα luminosity was derived by scaling the Lyα luminosity within the spectroscopic aperture by the geometric correction factor $f_{{\rm geo}}=A_{B_{W}}\times \nu ^{2}/(\omega \times d)$, where ω is the slit width and d is the spatial extent of the spectral extraction aperture. We stress, however, that in using area corrections based on broadband imaging we are relying on the assumptions that the B_W emission roughly traces the Lyα emission in the source and that the relative factor between the B_W and Lyα sizes is the same throughout the object as it is along the slit; either of these assumptions may be violated in practice.

In what follows, we discuss each confirmed Lyα source in detail. Of the confirmed Lyα nebulae, four were originally categorized as having a diffuse morphology while one (PRG4) was categorized as having a group morphology (see Paper I).

4.1.1. PRG1

4.1.2. PRG2

4.1.3. PRG3

4.1.4. PRG4

4.1.5. LABd05

The dominant contaminants in both the first- and second-priority spectroscopic samples are sources with spatially resolved blue continuum emission but no visible emission lines. Despite the lack of strong line emission in the B_W band, our morphological broadband search selected these sources either due to sufficiently extended, blue continuum emission or due to a close projected grouping of blue galaxies. A few examples of these continuum-only sources are shown in Figure 7.

Zoom In Zoom Out Reset image size

Without deeper spectroscopy, we can only speculate as to the nature of these continuum-only sources. The largest cases within the candidate sample (sources 1+2 and 3; Paper I) are so spatially extended (≈15''–86'', which at z ≈ 1.2–2.9 would imply physical sizes of ≈130–710 kpc in the continuum) and irregular in morphology that they are almost certainly located within the Galaxy, perhaps low surface brightness Galactic reflection nebulae. Since low-redshift (z ≲ 1.2) blue star-forming populations or low surface brightness galaxies (LSBs) would be expected to show [O ii], [O iii], or Hα emission lines in our spectra, some fraction of the remaining continuum-only contaminants may in fact be galaxies or Lyα nebulae in the redshift desert (1.2 ≲ z ≲ 1.6), for which Lyα is blueward of the atmospheric cutoff (λ_obs ≲ 3100 Å) but for which [O ii] has been redshifted past the red end of our MMT/Blue Channel spectra (λ_obs ≳ 8320 Å). One of the Lyα nebulae confirmed by our survey (PRG1, at z ≈ 1.67) is in fact below the redshift where Lyα is covered by the B_W band. Instead, this source was selected by our survey primarily due to blue continuum emission, and it was only thanks to the excellent blue sensitivity of MMT/Blue Channel that we were still able to detect the Lyα emission at ≈3250 Å. The case of PRG1 lends credence to the hypothesis that at least a fraction of the continuum-only "contaminant" sources are in fact Lyα nebulae at 1.2 ≲ z ≲ 1.6.

At the same time, however, our expectation from Paper I was that Lyα nebulae in the redshift desert should make up roughly 25% of the candidate sample, under the optimistic assumption that the Lyα nebula number density does not evolve significantly with redshift. In practice, we found continuum-only detections represented a much larger fraction (75%) of the target spectroscopic sample, suggesting that this explanation may not be the full story. While the presence of continuum emission in the spectra does confirm that these continuum-only sources are indeed real astrophysical objects and not artifacts within the NDWFS imaging, deeper ground-based optical spectroscopy or UV spectroscopy from space will be required to confirm their origin on a case by case basis. At this stage, their nature remains mysterious.

This work is the first demonstration of the feasibility of conducting systematic surveys for large Lyα nebulae using deep broadband imaging data sets. The primary advantage of our unusual survey approach is the enormous comoving volume that can be surveyed using deep archival data sets. In addition, since this search technique is best used in the blue where the sky is dark, the resulting Lyα nebula sample is weighted to lower redshifts (z < 3) where we have the opportunity to undertake detailed studies of their properties. The obvious tradeoff is that our approach is not as sensitive to Lyα nebulae that are intrinsically faint, low surface brightness, or compact in morphology, as discussed in Papers I and III. Our search, therefore, provides a measurement of the bright end of the Lyα nebula luminosity function, nicely complementing standard narrowband surveys that probe to fainter luminosities.

The success rate for finding sources with Lyα emission was ≈27% for first priority and ≈20% for second priority candidates. Therefore, if we were able to target all the Lyα nebula candidates in our sample, we would expect to find a total of ≈18 Lyα nebulae (≈10 and ≈8 from the first and second priority sets, respectively). While one of the goals of our broadband survey for Lyα nebulae is to place constraints on the space density of these rare objects, a robust estimate of the space density requires a detailed analysis of the selection function and is beyond the scope of the present paper. Here, we briefly discuss our detection rate in the context of traditional narrowband Lyα nebula surveys.

Based on the results of the narrowband survey carried out at z ≈ 2.3 by Yang et al. (2009), in Paper I we estimated the expected number of Lyα nebulae in our survey volume to be ∼60–400, assuming a 100% detection rate, the same detection limit as Yang et al. (2009), and a constant volume density as a function of redshift. Instead, we have confirmed 5 Lyα nebulae, and scaling these results to the unobserved candidates, we expect to find only 18. While this estimate is extremely crude, it does suggest that the space density of the detected Lyα nebulae in our sample is lower than that of the Yang et al. (2009) sample. Possible reasons for this difference are that (1) the Yang et al. (2009) narrowband survey is more sensitive to fainter, and therefore less luminous, Lyα nebulae than our broadband survey; (2) the Yang et al. (2009) survey does not exclude Lyα nebulae with bright central sources whereas our survey does due to the nature of the morphological search algorithm; and/or (3) the Yang et al. (2009) survey is more sensitive to cosmic variance than our larger volume survey. We defer a more detailed discussion of the space density of Lyα nebulae implied by our survey to Paper III.

The power of a systematic survey is the opportunity it provides to find out what is common among a class of objects and also what the dispersion in properties is among members of that class. The four large cases in our sample (PRG1, PRG2, PRG3, and LABd05) span nearly an order of magnitude in total Lyα luminosity (50–170 × 10⁴² erg s⁻¹), show a range of Lyα equivalent widths (∼50–260 Å), and are at least 70–100 kpc in diameter. Morphologically, the four large Lyα nebulae all show clumps and knots of emission in the broadband imaging. The brightest compact knot in PRG1 is very red while that in PRG2 is remarkably blue. In addition, all four show what appears to be diffuse continuum emission in the ground-based spectroscopy. This could either be due to many unresolved clumps or due to a continuum component that is truly spatially extended. Analysis of HST imaging of one system (LABd05) lends support to the latter hypothesis, revealing that most of the continuum in this one source is unresolved even at high resolution (01; Prescott et al. 2012b). In three cases (PRG1, PRG2, and LABd05) there is evidence for emission in other lines (e.g., C iv, He ii, or C iii]).

Given that diffuse continuum emission will have a larger impact on the observed broadband color than line emission, one might ask if our survey is biased toward finding lower equivalent width sources than narrowband surveys. In fact, however, Figure 8 shows that our survey uncovered Lyα nebulae with rest-frame equivalent widths comparable to those of luminous Lyα nebulae found using standard narrowband surveys but over a much larger redshift range.

Zoom In Zoom Out Reset image size