SHELS: A Complete Galaxy Redshift Survey with R ≤ 20.6
Margaret J. Geller, Ho Seong Hwang, Daniel G. Fabricant, Michael J. Kurtz, Ian P. Dell'Antonio, Harus Jabran Zahid
aa r X i v : . [ a s t r o - ph . GA ] M a y SHELS: A Complete Galaxy Redshift Survey with R ≤ Margaret J. Geller
Smithsonian Astrophysical Observatory,60 Garden St., Cambridge, MA 02138 [email protected]
Ho Seong Hwang
Smithsonian Astrophysical Observatory,60 Garden St., Cambridge, MA 02138 [email protected]
Daniel G. Fabricant
Smithsonian Astrophysical Observatory,60 Garden St., Cambridge, MA 02138 [email protected]
Michael J. Kurtz
Smithsonian Astrophysical Observatory,60 Garden St., Cambridge, MA 02138 [email protected]
Ian P. Dell’Antonio
Department of Physics, Brown University,Box 1843, Providence, RI 02912 [email protected]
Harus Jabran Zahid
Institute for Astronomy, University of Hawaii at Manoa,2680 Woodlawn Dr., Honolulu, HI 96822 [email protected]
ABSTRACT
The SHELS (Smithsonian Hectospec Lensing Survey) is a complete redshiftsurvey covering two well-separated fields (F1 and F2) of the Deep Lens Survey toa limiting R = 20.6. Here we describe the redshift survey of the F2 field (R.A. = 09 h m s and Decl. = +30 ◦ ′ ′′ ). The survey includes 16,294 newredshifts measured with the Hectospec on the MMT. The resulting survey of the4 deg F2 field is 95% complete to R = 20.6, currently the densest survey to thismagnitude limit. The median survey redshift is z = 0 .
3; the survey provides aview of structure in the range 0.1 . z . .
6. A movie displays the large-scalestructure in the survey region. We provide a redshift, spectral index D n < z < .
38. To demonstrate potential applications ofthe survey, we examine the behavior of the index D n
1. Introduction
During the 1980s redshift surveys of significant portions of the nearby galaxy distributionushered in the age of mapping the universe (Davis et al. 1982; Geller & Huchra 1989).Surveys with a variety of strategies rapidly explored the low redshift universe further andbegan to open our view of the intermediate redshift universe (e.g. Rowan Robinson et al.1990; Lilly et al. 1995; Shectman et al. 1996; Yee et al. 1996; da Costa et al. 1998). LeF´evre et al. (2013) summarize (their Figures 24 through 26 and associated references) thecharacteristics of the remarkable array of redshift surveys that now cover enormous volumesnearby and probe the universe to large redshift.The Sloan Digital Sky Survey (SDSS hereafter; Ahn et al. 2014), large-area southernsurveys to comparable or somewhat greater depth (e.g. Shectman et al. 1996; Colless et al.2001; Jones et al. 2009: Baldry et al. 2010), and probes of the distant universe have been 3 –the basis for revisions and new tests of our understanding of the growth of structure in theuniverse. However, largely as a result of the available instrument/telescope combinations,there are few dense surveys covering the redshift range 0 . < z < . in two well-separated 4 deg fields; the survey is essentially completeto a limiting R = 20.6. The survey takes its name from the original goal of comparing thematter distribution revealed by a foreground redshift survey with the weak lensing maps ofthe Deep Lensing Survey (DLS hereafter; Wittman et al. 2006). SHELS covers two of thefive 4 deg fields of the DLS, F1 and F2. We discuss the survey of F2 here and plan to reportlater on the intrinsically less dense survey field, F1 (the redshift survey is not yet complete).When complete the two well-separated regions should provide an improved measure of theimpact of cosmic variance relative to contiguous field covering the same area on the sky.Taken together the two SHELS fields will include ∼ ,
500 redshifts for galaxies with R ≤ z = 0 .
3. The SHELS F2 field is 95% complete to R=20.6 and is currently the most denselysampled region to this limit.The AGES survey (Kochanek et al. 2012) focuses on AGN evolution and the galaxyredshift survey covers an area and redshift range most comparable with SHELS. Kochaneket al. (2012) also used the 300-fiber Hectospec (Fabricant et al. 2005) on the MMT tomeasure redshifts. Their galaxy redshift survey covers 7.7 deg and includes galaxies withI <
20; the survey includes AGN to fainter limits. The median redshift is z = 0 .
31 nearlyidentical to SHELS. AGES uses a complex sparse sampling strategy in several photometricbands. To the SHELS limit, R = 20.6, the number density of AGES redshifts is ∼ − whereas the SHELS density averaged over the two fields is ∼ − .Thus SHELS, which is essentially complete to its magnitude limit, complements AGES byproviding a sample with a simple selection in apparent magnitude.We have used subsets of the SHELS data for a variety of projects that take advantage ofits straightforward selection. As initially intended, we have compared the matter distributionin the universe marked by galaxies in the redshift survey with the projected distributionmeasure with a weak lensing map (Geller et al. 2005; Geller et al. 2010). Cross-correlationof the two maps shows that the lensing map images the matter distribution traced by galaxiesin the redshift survey (Geller et al. 2005). Comparison of clusters of galaxies identified inthe redshift survey and from the weak lensing map suggests that the threshold for clusterdetection in weak lensing maps should be more conservative to yield reliable samples of realsystems (Geller et al. 2010; Utsumi et al. 2014).The uniform spectroscopy for the SHELS survey provided a basis for computation of 4 –the H α luminosity function (Westra et al. 2010) and for a determination of the faint end ofthe galaxy luminosity function at z ≤ . µ m (Hwang et al. 2012b) and to measure the mass-metallicity relation for galaxies in the redshift range 0 . < z < .
38 (Zahid et al. 2013).Here we provide the metallicities and stellar masses for galaxies in this redshift range.The SHELS data have also provided a partial basis for calibration of spectral indicatorsdetermined from Hectospec spectra (Fabricant et al. 2008). The data have also been part ofthe foundation for a demonstration that central velocity dispersions can be measured reliablywith the Hectospec (Fabricant et al. 2013). We plan to report central velocity dispersionsfor galaxies in both the F1 and F2 DLS fields in a separate paper.We describe the data in Section 2. We include the definition we use for the spectralindex D n . < z < .
38. Wedescribe the survey completeness in Section 2.4. We discuss both the observed and rest-framecolor distribution for the surey galaxies in Section 2.5. To provide an astrophysical contextand a partial guide to potential uses of the data we provide here, we briefly review the wayprevious workers have used the indicator D n n = 70 km s − Mpc − , Ω Λ = 0.7 and Ω m = 0.3 throughout. All quotederrors in measured quantities are 1 σ .
2. The Data
The SHELS redshift survey covers two of the five fields of the ambitious Deep LensSurvey (DLS; Wittman et al. 2006); these fields are denoted F1 and F2 and each covers ∼ &
90% to the magnitude limit, R ∼ .
6. This level of completeness is comparable withthe SDSS Main Galaxy Sample with limiting r = 17.77 (Strauss et al. 2002; Park & Hwang2009). 5 –The F1 field is centered at R.A. = 00 h m s and Decl. = 12 ◦ ′ ′′ . Theredshift survey of this field is not yet complete, but there are ∼ ≤ . & = 09 h m s and Decl. = +30 ◦ ′ ′′ . For this field we base our redshift survey on the DLS photometry.The DLS photometric data, with an effective exposure time of 14,500 seconds on the Mayall4-meter in < . ′′ seeing, reach a 1 σ limit for source detection in R of 28.7 magnitudesarcsec − (Muller et al 1998; Wittman et al 2006).The redshift survey (Geller et al. 2005; Geller et al. 2010; Westra et al. 2010; Geller etal. 2012; Hwang et al. 2012b) covers the F2 field to a limiting apparent magnitude R = 20.6.In the complete survey region of F2 with R ≤ & ≤ . . z . . The DLS selected all five of their fields, including F2, to exclude apparently brightnearby galaxies and to avoid known rich clusters with redshift z . .
1. The F2 field isdistinctive because it contains a complex of rich clusters at z ≃ .
3; these systems includeAbell 781 (Abell 1958).Wittman et al. (2006) describe the photometric data reduction pipeline for the DLS.The R band source list for the F2 field is the basis for our galaxy catalog. The auto-matic object identification algorithm produces a complete F2 catalog of objects with surfacebrightness µ ,R ≤ . − isophote.To construct the catalog for SHELS spectroscopy, we examined all of the objects withR ≤ . ∼
3% are stars with a median apparent magnitude R ∼ R <
15) stars produce contaminated catalogs where detected objects have incorrect pho-tometry and size estimates. Brighter stars affect a larger area, but effectively measuringphotometry for these bright stars in the DLS images is impossible. We thus develop anempirically calibrated measurement of the size of the affected area based on the USNO-A2(Monet et al. 2003) m V magnitudes for stars. We measured the regions with significant( > σ ) background enhancement around bright stars in one subfield of F2 (the p11 sub-field), and compared the radius of the region with the USNO-A2 V magnitude to developan empirical relation for the exclusion regions. The best-fitting relation is: r exc = 1 . m V − . arcsec (1)where r exc is the radius of the excluded region.For the few brightest stars, we had to adjust the size of the region based on the colorof the star (not surprising given the central wavelength difference between the DLS R filterand the photographic V filter used in USNO). For most stars, however, we use the exclusionregion specified by the USNO magnitude.In addition to the power-law PSF “halos”, very bright stars have significant bleed trailsin the DLS images. Although these trails constitute a smaller portion of the image, theyalso contaminate the photometry of galaxies near them. Using the SDSS, we recoveredphotometry for a few very bright galaxies with saturated cores and for 252 objects withisophotes overlapping a bleed trail but within the complete survey region.Table 1 lists the central coordinate and the radius for each of the masked regions. Figure1 shows the distribution of the masked regions superimposed on the survey region (left panel)and the cumulative area they cover as a function of their radii (right panel). The total areaof the DLS F2 field is 4.19 deg . The masked regions cover 0.21 deg , or about 5% of thearea. The complete redshift survey covers the unmasked region of 3.98 deg .We included all 13,249 galaxy candidates with Kron-Cousins R ≤ ≤ . We acquired spectra for the objects with the Hectospec (Fabricant et al. 1998, 2005) onthe MMT from April 13, 2004 to December 21, 2009. The instrument deploys 300 fibers over 7 –a 1 ◦ field. The Hectospec observation planning software (Roll et al. 1998) enables efficientsample acquisition. We filled unused survey object fibers with targets fainter than the surveylimit.The SHELS spectra cover the wavelength range 3,700 — 9,100 ˚A with a resolution of ∼ z ))of 48 km s − ; for the 238 pairs of emission-line objects, the mean internal error is 24 km − .The comparison pairs have redshifts ≤ − (also normalized by (1 + z )), but these objects are much brighter than the samplelimit.During the pipeline processing, we visually inspect all of the spectra and assign a qualityflag of “Q” for high-quality redshifts, “?” for marginal cases, or “X” for poor quality to theredshift determination. We report only high quality “Q” redshifts.To give a view of the quality of the spectroscopy, Figure 2 shows the distribution ofthe Tonry & Davis (1979) r T D statistic as a function of redshift. The statistic r T D measuresthe relative amplitude of the cross-correlation peak and thus provides an estimate of theerror in the redshift. We compute the error in the redshift according to the prescription inKurtz & Mink (1998). The maximum r T D value declines for apparently fainter galaxies atlarge redshift. The scatter of values with r T D . r T D .The F2 redshift survey in the complete region (unmasked) includes 12,705 galaxies witha measured redshift and with R ≤ ≤ r T D value, the redshift source, and a flag indicating whether the object is in the maskedregion. Objects designated 1 are within the masked region and we do not include theseobjects when we analyze the data. Their redshifts may obviously be useful for some otherpurposes, but their DLS photometry is unreliable and completeness is difficult to evaluatein these regions. Table 3 also includes derived quantities that we have used for analysis hereor in previous papers. These quantities are D n < z < . n n < n >
3; these values result from poor qualityspectra that are just adequate to yield a redshift. For each galaxy we also provide a stellarmass provided that the procedure outlined in Section 2.3 converges.Table 4 lists 2994 redshifts for galaxies with
R > . ≤ n As in Fabricant et al. (2008), we define D n f ν units) inthe 4000-4100 ˚A and 3850-3950 ˚A bands (Balogh et al. 1999). D n n n code writtenby Arnouts & Ilbert to SDSS five-band photometry. Fits of stellar population models to the ∼ arnouts/LEPHARE/lephare.html ∝ e − t/τ for τ = 0.1, 0.3, 1.2, 3, 5, 10, 15,and 30 Gyr. We allow 0 < E(B-V) < . < z < . α is outside the Hectospec bandpass for z ≥ .
38 and because aperture effects may be largefor z . R
23 = [
OII ] λ OIII ] λ , Hβ (2)where we assume the recombination value, 3, for the ratio of the [OIII] λ λ O
23 = [
OIII ] λ , OII ] λ α and [NII] λ > β and [OII] λ Figure 4 shows the differential completeness of the survey as a function of the limitingR-band magnitude (upper panel). The completeness declines very slowly to R ≃ ≤ ≤ < R < . ≤ z med, . = 0 .
30 (see Table 2).The two panels of Figure 4 show the two-dimensional completeness of the survey in12 ′ × ′ pixels. Yellow dots indicate galaxies without a redshift measurement. In the central8 × ≤ . < R ≤ . × <
90% complete, but all of these pixels aremore than 80% complete. For this fainter sample, the edges and corners contain pixels witha completeness . g − r ) model color as a functionof the DLS R-band magnitude for these objects. The color range for these objects is broadand comparable with the observed range for the entire sample. There is no significant trendin color with magnitude.Figure 6 shows a cone diagram projected along the R.A. direction. The pointsrepresenting individual galaxies are color-coded with the value of D n n z ∼ . Figure 7 shows the observed colors ( g − r ) and ( r − i ) of the survey galaxies as a functionof redshift. The expected impact of the redshift is obvious. The right-hand panel of Figure7 shows the rest frame color of the survey galaxies K-corrected to z = 0 .
35, somewhatgreater than the median survey redshift. We compute the K-correction using the SDSS 11 – ugriz photometry and the kcorrect code of Blanton & Roweis (2007). The choice of z = 0 . z & . z ∼ . z ∼ . − . z . . z ∼ .
3. D n The available photometric data along with the SHELS spectroscopy for the F2 fieldprovide a uniform, well-defined sampling of galaxy properties in the redshift range 0 . . z . .
6. In Sections 3.1 and 3.2 we display indicative quantities derived from the data as a briefguide to some possible future applications of this dataset.We focus here on the continuum feature D n n n n < r < z ∼ .
09, Kauffmann et al (2003) demonstrate that the distribution ofD n n n δ absorption have smaller mass-to-light ratios(these galaxies have undergone a burst of star formation in the last Gyr). In Sections 3.1and 3.2 we show the D n n n . < z < . z > .
75. In the extendedGroth strip covering 0.6 deg , the sample is simply magnitude limited (R AB < .
1; Newmanet al. 2013). Bundy et al. (2006) base their analysis on 943 galaxies on the range 0.4 < z < . n ∼ n δ to explore the star formationhistories of the galaxies in their sample. They demonstrate that the fraction of galaxies thathave experienced a burst of star formation in the last Gyr increases with redshift over therange 0 . < z < .
8. The color cuts defining the survey restrict the range of star formationhistories. The SHELS F2 survey can be used to calibrate color selected surveys because ithas the advantage that there are no cuts other than the magnitude limit.In analyzing the extensive zCOSMOS samples that cover the 2 deg COSMOS field(Scoville et al. 2007; Lilly et al. 2009), Moresco et al. (2010) extract ∼ . < z < < I < .
5. They use D n n ∼
50 km s − error in the Hectospec absorption-line redshifts aresmaller than the ∼
100 km s − errors for the zCOSMOS redshifts. This difference can beimportant when measuring the velocity dispersion in group environments and within theweb-like structures that define the galaxy distribution.More recently Moresco et al. (2013) have used the zCOSMOS 20K redshift survey torevisit the evolution of ETGs. Interestingly, they find a relationship between the apparentevolution of the sample and the basis for the sample definition (e.g. spectral features, color,and/or morphology). Their samples imply strong evolution of galaxies with stellar masses . M ⊙ and slow or no evolution for more massive galaxies. The most massive galaxies arethe least sensitive to the sample definition. Their Figure 6 shows stacked spectra binned inmass for ETGs defined in different ways. These spectra for z < . n D n n n z ∼ .
1. Thusthe SDSS 3 ′′ aperture subtends a diameter of 5.5 kpc and includes &
20% of the light evenfor luminous galaxies (see Figures 6 and 7 of Kewley et al. 2005). Although the 1.5 ′′ Hec-tospec fiber subtends a diameter of only 2.8 kpc, the median ratio of D n n n n n n n n z . .
2, D n n n . .
8) result fromlower signal-to-noise-spectra.The right-hand panel of Figure 8 shows normalized histograms of the D n n n < .
5) and older stellarpopulations (D n > . < z < . n n n n < . n z/z max . Here z max is the redshift where thegalaxy apparent magnitude is too faint for inclusion in the survey (Kauffmann et al. 2003).This approach places all galaxies in the interval at the same relative redshift. Because therange of absolute magnitudes we consider in each panel is narrow, the main effect of thescaling is to smooth out the impact of large-scale structure. Clearly, galaxies at greater z are generally at greater z/z max .The known evolutionary effects are evident in Figure 9. At greater z/z max , the fractionof galaxies with small D n − . < . M R < − . − . < . M R < − . z = 0 . − . < . M R < − .
0, there are few galaxies with large D n To gain more insight into the uses of SHELS for exploring issues in galaxy evolution,we show (Figure 10) the distribution of stellar masses of SHELS galaxies as a function ofredshift (left-hand panel). We color code the galaxies with the value of D n n ugriz spectral energy distribution. We determine thestellar masses from the photometry alone; we take only the redshift from the spectroscopicdata. Thus it is interesting to explore Figure 10 in some detail. For stellar masses M star & M ⊙ , the SHELS survey is deep enough to sample the mass range well to z ∼ < M star < M ⊙ , the typicalvalue of D n star > M ⊙ . Similar evolution is apparent in, for example, thePRIMUS survey (Moustakas et al. 2013)The spectral evolution of the galaxy population is not nearly as evident in the right-handpanel of Figure 10 where we display the absolute magnitude . M R as a function of redshiftcolor-coded by D n z/z max . Figure 11 shows the result. For the most luminous galaxies, − . < . M R < − . R ≃ − . < . M R < − .
0, there is asecondary peak at log(M/L) R ≃ − . − . < . M R < − .
0, galaxies with youngstellar populations predominate and the mass-to-light ratios are generally low.Comparison of the plots in Figures 9 and 11 seems at first glance to present a puzzle.In the magnitude interval − . < . M R < − .
0, the distribution of D n R peaks at ∼ . < D n <
2. 16 –The dependence of log(M/L) R and D n n ≤ R is a steep function of D n n R . In other words, a peak in the distribution ofD n R . For D n > .
5, thecorresponding (M/L) R range is small; here a broad distribution in D n R . The three panels of Figure 12 show that the relationship between D n n < M star / M ⊙ < . ), the distributiondevelops a relatively large tail with D n ≤ . z . This effect is much more pronounced for galaxies in the stellar massrange 10 < M star / M ⊙ < where most of the objects in the survey lie. In the bins oflowest stellar mass, the galaxies are predominantly star-forming.Figure 14 shows the SHELS F2 rest-frame spectra summed in bins of 0.1 in z and 0.5 dexin stellar mass corresponding to the histograms in Figure 13. We show only the rest framewavelength range 3600 -5300 ˚A that we sample throughout the redshift range of the survey.The summed spectra include galaxies with 0.1 < z < < M star / M ⊙ < . . Theinput spectra are flux corrected to units of relative F λ (Fabricant et al. 2013), preservingthe relative detected fluxes. We reject spectra with with D n < . n > .
0. InFigure 14 we normalize the final fluxes to a value of 100 at 4500 ˚A.The summed spectra show once again the increasing impact of star formation at higherz and lower stellar mass. In each panel, labeled by the redshift and stellar mass interval, wealso list the value of D n λ n λ λ n
4. The Mass-Metallicity Relation Through Summed Spectra for . < z < . n < z < .
38 where we have access to H α .Zahid et al. (2013, 2014) place the SHELS MZ relation in the context of other surveys.The data show that the mass-metallicity relation flattens as the universe evolves. They alsoshow that the gas-phase metallicity saturates; the saturation level is independent of redshift.Zahid et al. (2013) use the SHELS data for the F2 field as a partial basis for determinationof the MZ relation at 0.2 < z < .
38. Two aspects of the Hectospec spectroscopy limit theredshift range. For z . . z > .
38 H α is no longer within the Hectospec bandpass andthus the normal BPT diagram cannot be used to discriminate against AGN. Table 3 includesmetallicities for the individual galaxies in F2 used in the Zahid et al. (2013) analysis.Zahid et al. (2013) base their MZ relation on the analysis of individual spectra fromboth the F1 and F2 fields of the SHELS survey; the F1 field contibutes 40% of the spectra inthat study. Here, to investigate the robustness of the result, we construct summed spectra inmass bins for the F2 data alone. We then measure a metallicity for each summed spectrumand ask whether the results track the relation in Zahid et al. (2013).To construct the summed spectra in Figure 15 we apply the selection criteria describedin Section 2.3. In addition we remove 58 objects where the spectrum is contaminated bypoorly subtracted night sky lines. The final sample contains 2131 spectra.To construct the summed spectra, we first sort the spectra into equally populated bins ofstellar mass. Each mass bin (the limits are indicated in the Figure legend) is based on a stackof 213 galaxies. We then linearly interpolate the spectra and observational uncertainties toa common rest-frame wavelength vector based on the observed redshift. The interpolatedrest-frame wavelength vector has a 1.5 Angstrom resolution and covers the observed wave-length range 3500-9100 ˚A. At each resolution element, the stacked spectrum is the averageinterpolated flux of all spectra in the bin.The spectra in Figure 15 are ordered from top to bottom and from left to right beginningwith the lowest metallicity at the upper left. The legend gives the stellar mass range or eachbin and the KK04 metallicity. Changes in the spectra are obvious by visual inspection. The 18 –strength of [OII] λ α increases as themetallicity increases. There are corresponding decreases in the strength of [OIII] and H β .Table 7 lists the normalized flux as a function of rest-frame wavelength for these spectra.Figure 16 shows the MZ relation computed from the spectra in Figure 15 comparedwith the results of Zahid et al (2013). The two estimates of the relation are not independentbecause Zahid et al. (2013) include the F2 data, but analyzed in a completely different way.The comparison in Figure 16 demonstrates the robustness of the MZ relation to samplingand analysis issues. It also demonstrates the utility of the summed spectra here and, byanalogy, in Figure 14.
5. Conclusion
The goal of the SHELS project is to use redshift surveys complete to R = 20.6 to explorethe galaxy distribution and related spectroscopic properties of galaxies in the redshift range0.1 . z . .
6. The survey covers two widely-separated DLS fields. The total area of thesurvey will be ∼ . The first complete field, F2 field (R.A. = 09 h m s andDecl. = +30 ◦ ′ ′′ ) includes 12,705 galaxies with R ≤ clean surveyregion. The redshift survey is 95% complete to this limit. Currently the F2 survey is thedensest redshift survey to this limit. The median redshift of the survey z = 0 . z )) for an individual galaxy is .
50 km s − . The smallredshift errors make the survey very well-suited to evaluation of the local velocity dispersionin the full range of environments. The inclusion of [OII] λ n < z < n = 00 h m s and Decl. =12 ◦ ′ ′′ ), is not yet complete. We plan to provide similar data for F1 soon. It is interestingthat the ratio of galaxy counts to the limiting R = 20.6 in F1 versus F2 is ∼ Facilities:
MMT(Hectospec) 20 –
REFERENCES
Abell, G. O. 1958, ApJS, 3, 211Ahn, C. P., Alexandroff, R., Allende Prieto, C., et al. 2014, ApJS, 211, 17Baldry, I. K., Robotham, A. S. G., Hill, D. T., et al. 2010, MNRAS, 404, 86Balogh, M. L., Morris, S. L., Yee, H. K. C., Carlberg, R. G., & Ellingson, E. 1999, ApJ, 527,54Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, ApJS, 149, 289Blanton, M. R., Hogg, D. W., Bahcall, N. A., et al. 2003a, ApJ, 594, 186Blanton, M. R., Brinkmann, J., Csabai, I., et al. 2003b, AJ, 125, 2348Blanton, M. R., & Roweis, S. 2007, AJ, 133, 734Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Bundy, K., Ellis, R. S., Conselice, C. J., et al. 2006, ApJ, 651, 120Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ, 533, 682Chabrier, G. 2003, PASP, 115, 763Colless, M., Dalton, G., Maddox, S., et al. 2001, MNRAS, 328, 1039Cowie, L. L., Songaila, A., Hu, E. M., & Cohen, J. G. 1996, AJ, 112, 839da Costa, L. N., Willmer, C. N. A., Pellegrini, P. S., et al. 1998, AJ, 116, 1Davis, M., Huchra, J., Latham, D. W., & Tonry, J. 1982, ApJ, 253, 423Davis, M., Faber, S. M., Newman, J., et al. 2003, Proc. SPIE, 4834, 161Fabricant, D., Fata, R., Roll, J., et al. 2005, PASP, 117, 1411Fabricant, D. G., Kurtz, M. J., Geller, M. J., et al. 2008, PASP, 120, 1222Fabricant, D., Chilingarian, I., Hwang, H. S., et al. 2013, PASP, 125, 1362Foster, C., Hopkins, A. M., Gunawardhana, M., et al. 2012, A&A, 547, A79Freedman Woods, D., Geller, M. J., Kurtz, M. J., et al. 2010, AJ, 139, 1857 21 –Geller, M. J., & Huchra, J. P. 1989, Science, 246, 897Geller, M. J., Dell’Antonio, I. P., Kurtz, M. J., et al. 2005, ApJ, 635, L125Geller, M. J., Kurtz, M. J., Dell’Antonio, I. P., Ramella, M., & Fabricant, D. G. 2010, ApJ,709, 832Geller, M. J., Diaferio, A., Kurtz, M. J., Dell’Antonio, I. P., & Fabricant, D. G. 2012, AJ,143, 102Hwang, H. S., Geller, M. J., Diaferio, A., & Rines, K. J. 2012a, ApJ, 752, 64Hwang, H. S., Geller, M. J., Kurtz, M. J., Dell’Antonio, I. P., & Fabricant, D. G. 2012b,ApJ, 758, 25Jones, D. H., Read, M. A., Saunders, W., et al. 2009, MNRAS, 399, 683Kauffmann, G., Heckman, T. M., White, S. D. M., et al. 2003, MNRAS, 341, 33Kewley, L. J., Jansen, R. A., & Geller, M. J. 2005, PASP, 117, 227Ko, J., Hwang, H. S., Lee, J. C., & Sohn, Y.-J. 2013, ApJ, 767, 90Kobulnicky, H. A., & Kewley, L. J. 2004, ApJ, 617, 240Kochanek, C. S., Eisenstein, D. J., Cool, R. J., et al. 2012, ApJS, 200, 8Kurtz, M. J., & Mink, D. J. 1998, PASP, 110, 934Le F`evre, O., Cassata, P., Cucciati, O., et al. 2013, A&A, 559, A14Lilly, S. J., Le Fevre, O., Crampton, D., Hammer, F., & Tresse, L. 1995, ApJ, 455, 50Lilly, S. J., Le Brun, V., Maier, C., et al. 2009, ApJS, 184, 218Mignoli, M., Cimatti, A., Zamorani, G., et al. 2005, A&A, 437, 883Monet, D. G., Levine, S. E., Canzian, B., et al. 2003, AJ, 125, 984Moresco, M., Pozzetti, L., Cimatti, A., et al. 2010, A&A, 524, A67Moresco, M., Pozzetti, L., Cimatti, A., et al. 2013, A&A, 558, A61Moustakas, J., Coil, A. L., Aird, J., et al. 2013, ApJ, 767, 50 22 –Muller, G. P., Reed, R., Armandroff, T., Boroson, T. A., & Jacoby, G. H. 1998, Proc. SPIE,3355, 577Newman, J. A., Cooper, M. C., Davis, M., et al. 2013, ApJS, 208, 5Noeske, K. G., Faber, S. M., Weiner, B. J., et al. 2007, ApJ, 660, L47Park, C., & Hwang, H. S. 2009, ApJ, 699, 1595Roseboom, I. G., Pimbblet, K. A., Drinkwater, M. J., et al. 2006, MNRAS, 373, 349Rowan-Robinson, M., Lawrence, A., Saunders, W., et al. 1990, MNRAS, 247, 1Scoville, N., Aussel, H., Benson, A., et al. 2007, ApJS, 172, 150Shectman, S. A., Landy, S. D., Oemler, A., et al. 1996, ApJ, 470, 172Strauss, M. A., Weinberg, D. H., Lupton, R. H., et al. 2002, AJ, 124, 1810Tonry, J., & Davis, M. 1979, AJ, 84, 1511Utsumi, Y., Miyazaki, S., Geller, M. J., et al. 2014, ApJ, 786, 93Vergani, D., Scodeggio, M., Pozzetti, L., et al. 2008, A&A, 487, 89Westra, E., Geller, M. J., Kurtz, M. J., Fabricant, D. G., & Dell’Antonio, I. 2010, ApJ, 708,534Wittman, D., Dell’Antonio, I. P., Hughes, J. P., et al. 2006, ApJ, 643, 128Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, AJ, 140, 1868Yee, H. K. C., Ellingson, E., & Carlberg, R. G. 1996, ApJS, 102, 269Yi, S. K., Yoon, S.-J., Kaviraj, S., et al. 2005, ApJ, 619, L111Zahid, H. J., Geller, M. J., Kewley, L. J., et al. 2013, ApJ, 771, L19Zahid,H. J., Dima, G., Kudritzki, R., et al. 2014, arXiv:1404.7526
This preprint was prepared with the AAS L A TEX macros v5.2.
23 –Table 1. List of Masked Regions and Radii a ID R.A.
Decl.
Radius(deg) (deg) (arcsec)1 138.7028360 30.2579880 39.12 138.7108755 29.7457428 38.53 138.7169266 29.6362076 53.54 138.7175417 29.6406803 35.05 138.7219620 30.1846695 10.86 138.7447071 30.8056984 50.47 138.7479687 29.2265320 32.98 138.7577963 29.5055408 30.89 138.7579107 30.9256687 32.910 138.7674952 29.8776798 10.8 a This table is available in its entirety in amachine-readable form in the online journal.A portion is shown here for guidance regard-ing its form and content.
24 –Table 2.
SHELS F2 Redshift Survey PropertiesParameter ValueSurvey Area (deg ) 3.98N phot, . z, . phot, . z, . z med, . z,mask c z,R> . a Number of photometric objects in the complete survey (unmasked)region brighter than the quoted limit. b Number of galaxies with a measured redshift brighter than the speci-fied limit in the complete survey region. c Number of galaxies within the masked region of any apparent magni-tude (photometry is unreliable in these regions). d Number of objects with a redshift in the complete survey region butwith R > . Table 3. SHELS Redshifts with R ≤ . a SHELS ID SDSS ObjID b R z z
Mask d D n M ⋆ /M ⊙ ) 12(mag) Source c +log(O/H)138.7003546+30.7364270 1237664093976986160 18 . ± .
041 0 . ± . . ± .
003 0 . ± . . +0 . − . ...138.7015818+30.4485196 1237664668965601390 18 . ± .
019 0 . ± . . +0 . − . ...138.7018711+30.8165660 1237664669502538244 20 . ± .
004 0 . ± . . +0 . − . ...138.7027976+30.5422030 1237664668965601456 18 . ± .
002 0 . ± . . ± .
002 0 . ± . . +0 . − . ...138.7056609+29.9143189 1237664668428600041 20 . ± .
005 0 . ± . . +0 . − . . ± .
002 0 . ± . . +0 . − . . ± .
004 0 . ± . . +0 . − . ...138.7061703+30.1344974 1237664093439984063 19 . ± .
002 0 . ± . a This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding its formand content. The full table contains 13325 galaxies with a measured redshift; 13300 are MMT Hectospec redshifts and 25 are from the SDSS. Thereare 12705 galaxies in the complete survey region (12680 MMT redshfiifts and 25 SDSS redshifts all denoted 0 in column 6) b SDSS ObjID from DR10. If DR10 ObjID is not available, we present SDSS DR7 ObjID that starts with ‘58’. c (1) This study; (2) SDSS. d (0) Outside masked regions; (1) Inside masked regions.
26 –Table 4. SHELS Redshifts with
R > . aSHELS ID SDSS ObjID b R z c Mask d D n M ⋆ /M ⊙ ) 12(mag) +log(O/H)138.7021866+30.7739149 1237664093976986206 20 . ± .
005 0 . ± . . +0 . − . ...138.7045840+30.5183923 1237664668965601819 20 . ± .
005 0 . ± . . +0 . − . . ± .
006 0 . ± . . +0 . − . ...138.7072475+29.4259415 1237664667891663313 20 . ± .
004 0 . ± . . +0 . − . ...138.7072691+30.5814852 1237664668965601912 20 . ± .
005 0 . ± . . +0 . − . ...138.7073207+30.5341867 1237664668965601842 20 . ± .
005 0 . ± . . +0 . − . ...138.7099088+29.9916361 1237664668428600135 20 . ± .
005 0 . ± . . +0 . − . ...138.7104152+29.4288178 1237664667891663325 20 . ± .
005 0 . ± . . +0 . − . ...138.7114904+29.5393724 1237664667891663438 20 . ± .
005 0 . ± . . +0 . − . ...138.7126167+30.2366575 1237664093439984562 20 . ± .
005 0 . ± . . +0 . − . ... a This table is available in its entirety in a machine-readable form in the online journal. A portion is shown here for guidance regarding itsform and content. The full table contains 2994 galaxies all with an MMT Hectospec redshift. Among these galaxies, 2942 are outside themasked region. b SDSS ObjID from DR10. If DR10 ObjID is not available, we present SDSS DR7 ObjID that starts with ‘58’. c All redshifts are from this study. d (0) Outside masked regions; (1) Inside masked regions.
27 –Table 5. Objects without redshifts at R ≤ . a SHELS ID SDSS ObjID b R (mag)138.7025353+30.9340239 1237661381695505433 20 . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . a This table is available in its entirety in a machine-readable formin the onli ne journal. A portion is shown here for guidance regardingits form and content.The full table contains 703 objects. b SDSS ObjID from DR10. If DR10 ObjID is not available, weinclude SDSS DR7 ObjID that starts with ‘58’.
Table 6. Data for the spectra in Figure 14 a Wavelength (˚ A ) S(11.00 − b S(11.00 − − − − z = 0 . − . c ( z = 0 . − .
4) ( z = 0 . − .
5) ( z = 0 . − .
6) ( z = 0 . − . a This table is available in its entirety in machine-readable form in the online journal. A portion is shownhere for guidance regarding its form and content. We show only 5 of the 22 columns of data. Each columncorresponds to a spectrum in a panel of Figure 14. b Normalized Flux. Numbers in parentheses indicate the mass range for summed spectra. c Numbers in parentheses indicate the redshift range for summed spectra.
Table 7. Data for the spectra in Figure 15 a Wavelength (˚ A ) S( 9.08 − b S( 9.32 − − − − a This table is available in its entirety in machine-readable form in the online journal. A portion is shownhere for guidance regarding its form and content. We show only 5 of the 10 columns of data. Each columncorresponds to a spectrum in Figure 15. b Normalized Flux. Numbers in parentheses indicate the mass range for summed spectra.
30 –Fig. 1.— Survey mask (left) along with the cumulative area covered by the masked regionsas a function of angular radius of the region (right). The red dots in the left-hand panelshow the size of the masked regions; the dots indicate galaxies in the survey with measuredredshifts. 31 –Fig. 2.— Cross-correlation r -value (Tonry & Davis 1979), a redshift quality indicator, as afunction of redshift (upper panel). The center panel shows apparent R-band magnitude asa function of redshift. Points scattered above the faint limit result from poor photometry.The lower panel shows a redshift histogram in bins of ∆ z = 0 .
01. 32 –Fig. 3.— Sample absorption-line (left) and emission-line spectra (right) demonstrating therange of quality (cross-correlation coefficient) at z ∼ .
29. Labels indicate major spectralfeatures; unlabeled spikes are badly subtracted night sky lines. 33 –Fig. 4.— Completeness of the SHELS redshift survey of the DLS F2 field. The upper panelshows the completeness as a function of DLS R-band magnitude. The color bar insert showsthe completeness fractions for the spatial completeness displays in the lower two panels.The lower left panel shows the completeness in 12 ×
12 arcminute bins for galaxies with R < .
3. The yellow points indicate galaxies in the photometric sample without a measuredredshift. The right hand plot shows the completeness in the interval 20.3 < R < . ≤ direction.The color coding indicated the value of D n n . n < . ≤ D n < . n ≥ g − r ) fiber, and ( r − i ) fiber, SDSS extinction-corrected colors for theSHELS sample (left panels). (Right) K-corrected colors for SHELS galaxies shifted to theapproximate median survey redshift, z = 0 .
35. We display only 30% of the data for clar-ity. The narrowing color distribution at the largest redshifts occurs because the magnitudelimited sample contains increasingly luminous galaxies that are generally somewhat redder. 37 –Fig. 8.— D n n n z/z max where z is the redshift of the galaxy and z max isthe redshift where the galaxy would fall out of the magnitude limited sample. The shapeof the contours in each interval reflects the proportion of galaxies with young/old stellarpopulations. The typical population is younger for less luminous galaxies as expected. Theright-hand panels show histograms of D n z = 0 .
35) R-bandabsolute magnitude as a function of redshift (right). In the both panels galaxies are color-coded by D n z/z max ; z is the galaxy redshift and z max is the redshift where the galaxy becomes fainter than the magnitude limit. The panelsshow bins in K-corrected absolute magnitude. Note that the contours have no significantslope with z/z max . Note also the peak of the M/L R distribution moves toward smaller M/L R for less luminous galaxies as expected. 41 –Fig. 12.— R-band mass-to light ratio M/L R as a function of D n n . n & n n n n < z << z <