A Near-Infrared Spectroscopic Survey of K-selected Galaxies at z~2.3: Redshifts and Implications for Broadband Photometric Studies

3 MB·English

by  Mariska Kriek

Accepted forpublicationin ApJ
PreprinttypesetusingLATEXstyleemulateapjv.10/09/06
A NEAR-INFRARED SPECTROSCOPIC SURVEY OF K-SELECTED GALAXIES AT z ∼2.3: REDSHIFTS
AND IMPLICATIONS FOR BROADBAND PHOTOMETRIC STUDIES 1,2
Mariska Kriek3,4,5, Pieter G. van Dokkum4, Marijn Franx3, Garth D. Illingworth6, Danilo Marchesini4, Ryan
Quadri4,3, Gregory Rudnick7, Edward N. Taylor3, Natascha M. Fo¨rster Schreiber8, Eric Gawiser4,9, Ivo
Labb´e10, Paulina Lira11 and Stijn Wuyts3,12
Accepted for publication inApJ
ABSTRACT
Using the Gemini Near-InfraRed Spectrograph (GNIRS), we have completed a near-infrared spec-
8 troscopic survey for K-bright galaxies at z ∼ 2.3, selected from the MUSYC survey. We derived
0 spectroscopic redshifts from emission lines or from continuum features and shapes for all 36 observed
0 galaxies. The continuum redshifts are driven by the Balmer/4000˚A break, and have an uncertainty
2 in ∆z/(1+z) of <0.019. We use this unique sample to determine, for the first time, how accurately
redshifts and other properties of massive high-redshift galaxies can be determined from broadband
n
a photometric data alone. We find that the photometric redshifts of the galaxies in our sample have a
J systematicerrorof0.08andarandomerrorof0.13in∆z/(1+z). Thesystematicerrorcanbereduced
7 by using optimal templates and deep photometry; the random error, however,will be hard to reduce
below 5%. The spectra lead to significantly improved constraints for stellar population parameters.

] For most quantities this improvement is about equally driven by the higher spectral resolution and
h by the much reduced redshift uncertainty. Properties such as the age, A , current star formation
V
p
rate,andthestarformationhistoryaregenerallyverypoorlyconstrainedwithbroadbanddataalone.
-
o Interestinglystellarmassesandmass-to-lightratiosareamongthemoststableparametersfrombroad-
r band data. Nevertheless, photometric studies may overestimate the number of massive galaxies at
st 2<z <3,andthus underestimate the evolutionofthe stellarmass density. Finally,the spectroscopy
a supports our previous finding that red galaxies dominate the high-mass end of the galaxy population
[ at z =2−3.

Subject headings: galaxies: high redshift — galaxies: distances and redshifts — galaxies: evolution
1
v
0
1. INTRODUCTION galaxies, photometric redshifts allow the study of
1
apparently unbiased samples. Our current under-
1 The extensive use of photometric redshifts has
standing of the evolution of the mass-density (e.g.,
1 greatly enhanced our knowledge of the z = 2 −
. 3 universe. While color criteria, such as the Ly- Rudnick et al. 2001, 2003, 2006; Dickinson et al. 2003;
1
Drory et al. 2005) and the luminosity function (e.g.,
0 man break technique (Steidel et al. 1996a,b), dis-
Dahlen et al.2005;Saracco et al.2006;Marchesini et al.
8 tant red galaxy selection (DRGs, Franx et al. 2003;
2007), the nature of massive high-redshift galaxies
0 van Dokkum et al.2003)andBzKselection(Daddi et al.
(e.g., F¨orster Schreiber et al. 2004; Labb´e et al. 2005;
: 2004) provide an easy identification of high-redshift
v van Dokkum et al. 2006; Papovichet al. 2006), and
Xi Electronicaddress: [email protected] galaxy clustering (e.g., Daddi et al. 2003; Quadri et al.
1Based on observations obtained at the Gemini Observatory, 2007a;Foucaud et al.2007)essentiallyallrelyonphoto-
r which is operated by the Association of Universities for Research metric redshifts.
a in Astronomy, Inc., under a cooperative agreement with the NSF
Ideally, all these studies would have been based on
onbehalfoftheGeminipartnership.
2Fullyreducedspectraandcorrespondingdataproductswillbe spectroscopic redshifts. However, obtaining spectro-
madepublicathttp://www.astro.yale.edu/MUSYConJanuary15, scopic redshifts for the required samples is hampered
2008 by several obstacles. Due to their faintness, obtaining
3LeidenObservatory,LeidenUniversity,POBox9513,2300RA
redshifts of high-redshift galaxies requires long integra-
Leiden,Netherlands
4Department of Astronomy, Yale University, New Haven, CT tions on 8-10m class telescopes. The largest samples of
06520 spectroscopically confirmed high redshift galaxies num-
5Current address: Department of Astrophysical Sciences,
ber in the 1000s(e.g., Steidel et al.2003), severalorders
PrincetonUniversity,Princeton,NJ08544
6UCO/Lick Observatory, University of California, Santa Cruz, of magnitude short of the largest photometric samples.
CA95064 Furthermore, and more fundamentally, current spectro-
7Goldberg Fellow, National Optical Astronomy Observatory, scopic samples are strongly biased towards blue, star
950NorthCherryAvenue, Tucson,AZ85719
forming galaxies which are bright in the observer’s opti-
8Max-Planck-Institut fu¨r extraterrestrische Physik, Giessen-
bachstrasse,Postfach1312,D-85748Garching,Germany cal (Lyman break galaxies, or LBGs). However, it has
9Department of Physics and Astronomy, Rutgers University, becomeclearthatthetypicalmassivegalaxyatthis red-
Piscataway, NJ08854 shift range is red in the rest-frame optical and faint in
10Hubble Fellow, Carnegie Observatories, 813 Santa Barbara
the rest-frame ultra-violet (UV), and blue LBGs con-
Street,Pasadena, CA91101
11Departamento de Astronom´ıa, Universidad de Chile, Casilla stitute only ∼20% of massive galaxies at z = 2 − 3
36-D,Santiago, Chile (van Dokkum et al. 2006).
12Harvard-Smithsonian Center for Astrophysics, 60 Garden Obtaining spectroscopic redshifts for typical mas-
Street,Cambridge,MA02138
2 Kriek et al.
TABLE 1
SampleandObservations
Exptime
ID RA DEC Ks R J−K observationdates min IDKR06B+07a
1030-32 103041.5 51955 19.68 26.83 2.62 2006/12/19 +2007/03/12 230 -
1030-101 103010.1 52011 18.97 25.63 2.15 2006/02/23 120 -
1030-301 103050.8 52049 18.82 25.49 2.81 2006/01/20 90 -
1030-609 103049.6 52150 19.68 24.23 2.02 2006/02/24 130 -
1030-807 103020.0 52233 19.72 24.77 2.40 2006/02/23 120 -
1030-1531 103038.9 52452 19.38 22.92 2.23 2006/02/25 80 -
1030-1611 103048.4 52503 19.58 25.55 2.37 2006/02/24 120 -
1030-1813 103051.2 52536 19.01 24.97 2.93 2006/01/20 80 -
1030-1839 103045.4 53007 19.61 24.20 2.35 2006/12/16 80 -
1030-2026 103022.7 52826 19.48 25.22 2.93 2006/02/22 120 -
1030-2329 103016.2 52732 19.72 25.24 2.47 2006/02/25 120 -
1030-2559 103040.1 52634 19.62 25.89 2.52 2006/02/22 110 -
1030-2728b 103018.4 52605 19.52 25.09 2.69 2006/01/21 120 -
1030-2927 103043.3 52934 19.48 24.52 2.23 2006/12/18 +2007/03/13 230 -
1256-0 125459.6 11130 19.26 24.98 2.26 2005/05/19+27+30 +2006/02/24 305 151
1256-142 125502.7 10732 19.45 25.99 2.54 2006/02/23 120 465
1256-519 125508.4 10614 18.99 25.51 2.55 2006/02/25 80 -
1256-1207 125519.7 11246 19.25 25.34 2.04 2006/02/25 80 -
1256-1967 125525.8 10325 18.71 23.55 2.02 2005/05/18 +2006/01/18 240 2889
HDFS1-259 223311.2 -604047 19.42 24.00 2.11 2006/12/17-18 140 -
HDFS1-1849 223337.9 -603315 19.30 25.18 2.51 2004/09/06 115 -
HDFS2-509 223123.1 -603908 18.57 23.20 2.50 2005/05/16+19 235 -
HDFS2-1099 223203.2 -603613 19.26 24.94 2.55 2006/12/19 120 -
HDFS2-2046 223230.8 -603244 19.38 24.24 2.21 2005/05/20 125 -
ECDFS-4454 33211.5 -275523 19.24 24.28 2.89 2006/01/18 100 3662
ECDFS-4511 33243.2 -275515 18.77 23.36 2.62 2006/01/21 +20006/02/25 190 3694
ECDFS-4713 33152.5 -275448 18.68 22.94 2.13 2006/02/22 60 3896
ECDFS-5856 33213.3 -275226 19.42 25.42 3.21 2006/01/19 120 4937
ECDFS-6842 33151.3 -275056 19.09 24.47 2.47 2006/12/19 +2007/03/11+14 210 -
ECDFS-6956 33202.5 -275046 19.16 23.40 2.43 2006/01/20 +2006/02/24 150 5754
ECDFS-9822 33133.9 -274603 19.14 24.13 3.04 2006/12/17 120 -
ECDFS-11490 33245.0 -274309 19.25 24.01 2.49 2006/01/20+21 190 9510
ECDFS-12514 33139.5 -274120 19.11 22.79 1.72 2006/02/23 90 10525
ECDFS-13532 33154.8 -273923 19.52 24.98 3.51 2006/12/18 160 -
ECDFS-16671 33158.9 -273516 18.99 22.08 1.74 2006/12/16 60 -
CDFS-6202 33231.5 -274623 19.04 23.62 2.28 2004/09/02+03 90 6036
a IDnumbersinKrieketal.(2006b,2007)b ThespectroscopicredshiftofthisgalaxyhasfirstbeenconfirmedusingK-bandspectroscopy
withNIRSPEConKeck,in2005January.
sive galaxies requires deep spectroscopy at near-infrared NIRspectroscopyonasubstantial,unbiasedsampleof
(NIR)wavelengths. Unfortunately,NIRobservationsare massive,high-redshiftgalaxiesis neededto testourpho-
complicated by the combination of the high sky bright- tometric studies, and obtain insights regarding possible
ness, numerous bright and variable night sky lines and systematic effects. The Gemini Near-InfraRed Spectro-
strong atmospheric absorption bands, and the limited graph(GNIRS,Elias et al.2006)isespeciallywell-suited
field of view of current and planned NIR spectrographs. for spectroscopy of z ∼ 2.3 galaxies, due to the large
Thus, obtaining spectroscopic redshifts for hundreds or wavelength coverage offered by the cross-disperser (0.9-
thousands of galaxies with K ∼ 21 (the typical bright- 2.5 µm). This coverageoffers two advantages: there is a
ness of galaxies with M > 1011M⊙ at z ∼ 2.5) will large probability of finding emission lines, and it allows
not be feasible in the foreseeable future. Until the next the characterization of the continuum emission, which
generation of space missions and > 20m ground-based is particularly important for galaxies without detected
telescopes we remain largely dependent on photometric lines. In Kriek et al. (2006b) we found that a substan-
redshifts for studies of large and faint galaxy samples tial fraction of the massive galaxies at z ∼ 2.3 have no
beyond z >1.5. detected emission lines. For these galaxies we derived
Our provisional dependency on broadband photomet- spectroscopic redshifts by modeling the stellar contin-
ric studies requires a more accurate calibration and un- uum, driven by the Balmer/4000 ˚A break (Kriek et al.
derstanding of the involved systematics. The current 2006a).
spectroscopic samples used for calibration of photomet- In this paper we present our full survey of K-selected
ric high-redshift studies are based primarily on optical galaxies at z ∼ 2.3, conducted with GNIRS on Gemini-
spectroscopy. As these samples are biased towards un- South between September 2004 and March 2007. In to-
obscuredstar-forminggalaxies,theircalibrationmaynot tal we integrated more than ∼80 hours, divided over 6
be representative for the total sample of massive galax- observing runs, on a sample of 36 galaxies. In previ-
ies. Photometric properties of red, massive galaxies at ous papers based on preliminary results of this survey
high redshift are poorly calibrated, and since red galax- we discussedthe stellar populations (Kriek et al. 2006b)
iesdominatethehighmassendat2<z <3,systematics andtheoriginofthelineemission(Kriek et al.2007). In
may have large effects on the final results. this paper we will give an overview of the total survey
A NIR Spectroscopic Survey of z ∼2.3 Galaxies 3
Fig. 2.—ComparisonofJ−K andR−K colorsasafunctionof
Fig. 1.—Comparisonbetweenthephotometricpropertiesofthe zphot betweenaphotometricmass-limited(M >1011M⊙)sample
and our spectroscopic sample. The gray diamonds and dots rep-
GNIRSsampleat2<zphot<3andamass-limitedsample(M > resentall massivegalaxies inthe deep MUSYC fields (SDSS1030,
1011M⊙)at2<zphot<3. Theprobabilities(P)thattheGNIRS 1256 and HDF-S)with K <19.7 and K >19.7 respectively. The
sampleandthefullmass-selectedsamplehavesimilardistributions, blacksymbolsrepresentthe 36galaxies oftheGNIRS sample,se-
asderivedusingaMann-Whitney(MW)andaKolmorov-Smirnov lectedfromboththeMUSYCdeep(diamonds)andwide(ECDFS,
(KS)test,aregiveninthepanels. Additionally,wedividethemass- crosses)surveys.
selected sample into its K-bright (K < 19.7) and K-faint (K >
19.7)members. TheGNIRSsamplemaybelessrepresentativefor
aK-brightsample,astheredshiftdistributionisdifferent. work is described by S. Wuyts et al. (2007, in prepara-
tion).
(§ 2), compare photometric and spectroscopic redshifts
We selected galaxies with 2.0.z .2.7 (see § 3.2)
(§3)andstellarpopulationsproperties(§4),anddiscuss phot
and K . 19.7. This redshift interval is chosen as the
the implications for photometric studies (§ 4).
bright emission lines Hβ, [Oiii], Hα, and [Nii] fall in
Throughoutthe paper we assume a ΛCDM cosmology
the H and K atmospheric windows. A few galaxies had
with Ω = 0.3, Ω = 0.7, and H = 70 km s−1 Mpc−1.
m Λ 0 2.7 . z . 3.0 at the time of selection, but were
phot
The broadband magnitudes are given in the Vega-based
observed because a large part of their confidence inter-
photometric system unless stated otherwise. Further-
vals fell in the targeted redshift range, or because the
more,wewillmeasurethe scatterandthe offsetbetween
galaxy had a known z between 2.0 and 2.7 (CDFS-
spec
various properties using the normalized biweight mean
6202, van Dokkum et al. 2005; Daddi et al. 2005). Due
absolute deviation and the biweight mean, respectively
to catalog updates of the ECDFS field, the final z of
phot
(Beers, Flynn & Gebhardt 1990). As biweight statistics
several more galaxies scattered out of the selected red-
arelesssensitivetowardsoutliersthanthenormalmean,
shift range. The fact that the ECDFS catalog was still
and more efficient than the median, they are most ap-
in a preliminary stage at the time of selection may com-
propriate for the small sample sizes in this work.
plicate some of the interpretation in this work. For this
2. DATA reasonandthefactthattheECDFSistheonlyfieldwith
shallow NIR photometry, the analysis in this paper will
2.1. Sample Selection
also focus on the subsample excluding the ECDFS.
The galaxies studied in this paper are selected from In total we obtained usable NIR spectra for a sample
the MUlti-wavelength Survey by Yale-Chile (MUSYC, of 36 galaxies. For ∼4 additional galaxies we obtained
Gawiser et al. 2006; Quadri et al. 2007b). This survey empty spectra due to mis-alignment or extremely bad
consist of optical imaging (UBVRIz) of four 30′ × 30′ weather conditions.
fields, shallow NIR imaging (JHK) over the same area, Itis importantto establishwhether our sample is rep-
and deeper NIR imaging over four 10′ × 10′ fields. The resentative for the galaxy population at z ∼2.5. In Fig-
depth of the deep and wide NIR photometry is K ∼ 21 ure1wecomparethedistributionsofJ−K,R−K,rest-
and K ∼20 (5σ) respectively. The spectroscopic follow- frame U−V color andzphot ofour spectroscopicsample
up presented in this paper is selected from the deep with a photometric mass-limited sample (>1011M⊙) at
fields HDF-South, 1030,and1256(Quadri et al. 2007b), 2 < z < 3. For the latter we use the deep MUSYC
and the shallow extended Chandra Deep Field South fields, as the wide NIR data are too shallow to extracta
(ECDFS, E.N. Taylor et al. 2007, in preparation). One mass-limited sample (van Dokkum et al. 2006). Accord-
galaxy is selected from the Great Observatories Ori- ing to a Mann-Whitney and a Kolmorov-Smirnov test,
ginsDeep Survey(GOODS;Giavalisco et al.2004). The the photometric properties J − K, R − K, rest-frame
optical-to-NIR photometry that we used as part of this U −V color (as derived from the photometry) and z
phot
4 Kriek et al.
Fig. 3.— The differences between the low resolution spectra Fig. 4.—ComparisonofphotometricNIR colors,andthose de-
of two observing sequences for the same galaxy (left: 1030-301, rived from the spectra, for galaxies in the deep (solid) and wide
right: 1256-0)areusedtoestimateasystematicuncertaintyonthe (open) MUSYC fields. Both colors arenot corrected forflux con-
spectra. Inordertoimprovetheconsistencybetweenthedifferent tributions by emission lines. For a few galaxies the spectra are
observing sequences, we increase the original uncertainty per bin too noisy to measure the spectroscopic NIR colors. The fraction
(blackerrorbars)byasystematicerrorof10%(grayerrorbars)of ofgalaxiesforwhichthephotometricandspectroscopiccolorsare
theaveragefluxinthebinnedspectrum. Thefractionsofbinsthat consistentwithin1σisgiveninthetopleftcorner. Asthesefactors
areconsistentwithin1σfortheoriginalandincreaseduncertainties arelessthan0.68,theerrorsmaybeslightlyunderestimated.
aregiveninthepanelsinblackandgrayrespectively.
of the GNIRS sample are representativefor a photomet- lines mm−1 grating and the 0′.′675 slit. This configu-
ric mass-limited sample at 2 < z < 3 (see probabilities ration resulted in a spectral resolution of R ∼ 1000.
in panels of Fig. 1). The Mann-Whitney test assesses The galaxies were observed during six observing runs in
whether the two sample populations are consistent with 2004 September (program GS-2004B-Q-38), 2005 May
the same mean of distribution, while the K-S test ex- (program GS-2005A-Q-20),2006 January (program GS-
amines whether the two samples could have been drawn 2005B-C-12)and 2006 February (programGS-2006A-C-
from the same parent distribution. For both tests the 6), 2006 December (program GS-2006B-C-5) and 2007
probability should be greater than the 0.05 significance March (program GS-2007A-C-9). During the first two
level. runs most time was lost due to bad weather, and only a
The galaxies targeted with GNIRS are all bright in K handfulofgalaxieswasobservedundermediocreweather
(< 19.7). In order to examine if bright galaxies are a condition (seeing ∼ 1′′). The weather was excellent
biased sub-sample of the total mass-limited sample, we throughout the full 3rd and 4th run, and we reached
split the sample in bright and faint members. Figure 1 a median seeing of ∼0′.′5. The conditions were slightly
shows that the bright and faint members have the same worse during the last two runs, with a median seeing of
distribution of rest-frame U − V and observed R − K ∼0′.′7, and some time was lost due to clouds.
colors. The maindifference betweenthe brightandfaint We observed the galaxies following an ABA′B′ on-
membersistheredshiftdistribution: almostallK-bright sourceditherpattern,suchthatwecanusetheaverageof
galaxies have zphot< 2.3. This also causes their bluer the previous and following exposures as sky frame. This
J − K colors: in contrast to the R-band, the J-band cancels sky variation and reduces the noise in the final
does not fall entirely bluewards of the optical break for frame. All targets were acquired by blind offsets from
z < 2.3. The redshift dependence of J −K is clearly nearbystars. Theindividualexposuresare5minutesfor
visible in Figure 2. However, except for the difference the galaxies observed during the first two runs, and 10
in redshift and presumably stellar mass, we see no hints minutes for the remaining runs13. The total integration
thatthebrightandfaintmembersofamass-limitedsam- times for all galaxies are listed in Table 1. Before and
pleat2<z <3havedifferentstellarpopulations. Thus, after every observing sequence we observean A V0 star,
althoughthespectroscopicsamplemayhavesimilarstel- forthe purposeofcorrectingfortelluricabsorption. The
lar population properties as K-bright galaxies, it is less final spectra of the two stars were combined to match
representative for a K-bright sample at 2 < z < 3, as the target’s airmass.
the medianredshiftanditscorrespondingdistributionis Adetaileddescriptionofthereductionprocedureofthe
substantially different. GNIRS cross-dispersed spectra is given in Kriek et al.
2.2. NIR spectra 13 Aninstrumentupgradeafterthefirsttworunsimprovedthe
throughput and thus the quality of the spectra, and eliminated
Weobservedthefullsampleof36galaxieswithGNIRS
“radiation events” caused by radioactive coatings. This allowed
in cross-dispersed mode, in combination with the 32 longerintegrations
A NIR Spectroscopic Survey of z ∼2.3 Galaxies 5
Fig. 5.— GNIRS spectra (black squares) and MUSYC broadband photometry (open circles) for all 36 galaxies, sorted for their total
K-bandmagnitudestartingwiththebrightestgalaxy(seeTable1). Fluxesaregivenin10−19ergss−1cm−2˚A−1. Emissionlinefluxesare
removedfromthebinnedspectra,usingthebest-fitmodelstothelines. Thephotometry isnotcorrectedforemissionlinecontamination.
Thebest-fit stellar population models tothe spectra andphotometry areshown ingray. For emission-linegalaxies the redshiftwas fixed
atzline duringfitting,whilefortheremaininggalaxiesz wasafreeparameter.
6 Kriek et al.
Fig. 5.—Continued
(2006a). In summary, we subtract the sky, mask cos- photometric colors are consistent within 1σ are listed in
mic raysandbadpixels,straightenthe spectra,combine thepanels. Forallcolorsthesefractionsareslightlylower
the individual exposures, stitch the orders and finally than expected, suggesting that the uncertainties may be
correct for the response function. 1D spectra are ex- slightlyunderestimated. Thismaybepartlyduetocolor
tracted by summing all adjacent lines (along the spatial gradientsinthegalaxies,asweusedifferentaperturesto
direction) with a mean flux greater than 0.25 times the determine the colors. The spectroscopic apertures are
fluxinthecentralrow,usingoptimalweighting. Wealso rectangular, and depend on the extraction radius and
constructed“lowresolution”binnedspectrafromthe2D method, which differs per galaxy. On average the spec-
spectraforeachgalaxyfollowingthemethodasdescribed troscopic apertures are 0′.′675 by 1′.′2, slightly smaller
inKriek et al.(2006a). Eachbincontains80“good”pix- than the circular photometric apertures which have a
els(i.e.,wavelengthregionswithhighatmospherictrans- diameter of ∼1′.′4.
mission and low sky emission), corresponding to 400 ˚A The final low resolution spectra and broadband pho-
per bin. Sky, transmission, and noise spectra were con- tometry for the full sample are presented in Figure 5.
structed for each galaxy as well. The galaxies are ordered by total K magnitude (see Ta-
Weassesstheuncertaintiesonthelowresolutionspec- ble 1), starting with the brightest galaxy. Remarkably,
tra by splitting the data in two sequences for several there is not a strong trend between quality of the spec-
objects and comparing the results. We find that the er- trumandthetotalK-bandmagnitudeforthesameinte-
rors as derived from the photon noise underestimate the gration time. We suspect that the surface brightness of
true uncertainty for most galaxies. In Figure 3 we show theobjectplaysanimportantrole,asbrightobjectswith
two examples. In order to obtain a better consistency lowqualityspectraandtypicalintegrationtimes,suchas
between the observing sequences, we increase the uncer- ECDFS-4511andECDFS-9822,areextendedeveninthe
tainties for all bins by quadratically adding 10% of the MUSYC imaging, which have a image quality of ∼1′′.
average flux in the spectrum.
WeusethebroadbandNIRphotometrytoperformthe 3. SPECTROSCOPICVERSUSPHOTOMETRICREDSHIFTS
absolute flux calibration. For each galaxy we integrate 3.1. Spectroscopic Redshifts and Galaxy Properties
the spectrum over the same J, H and K filter curves
For 19 of the galaxies in the sample we detected one
as the photometry. We determine one scaling factor per
or more emission lines, and thus for these galaxies we
galaxy,fromthedifferencebetweenthespectroscopicand
could determine exact spectroscopic redshifts. The re-
photometricNIRmagnitudes,andusethisfactortoscale
maining galaxies may have no or very faint emission
the NIR spectrum. We note that at this stage both the
lines,orthelinesareexpectedinatmosphericwavelength
spectra and the photometry may contain flux contribu-
regions with low transmission or strong sky emission.
tions by emission lines. We extend our wavelength cov-
Fortunately, we can derive fairly precise redshifts from
erage by attaching the optical photometry to the scaled
the continuum emission alone, mainly due to the pres-
spectra. For the emission-line galaxies we subsequently
ence of the Balmer/4000 ˚A break in the NIR spectra
remove the line fluxes from the affected bins, using the
(Kriek et al.2006a). Fornone ofthe galaxiesabsorption
best-fit to the emission lines as derived in Kriek et al.
lines are detected.
(2007).
We fit the low resolutionbinned spectra togetherwith
As a quality check we compare the photometric NIR
the optical photometry by Bruzual & Charlot (2003)
colorsJ−H,J−K andH−K tothosederivedfromthe
stellarpopulationmodels. Weallowagridof24different
spectra. The direct comparisonfor the individual galax-
ages (not allowing the galaxy to be older than the age
ies is presented in Figure 4. As expected, the scatter
of the universe), and 31 different exponentially declin-
is larger for the galaxies with shallow NIR photometry.
ing starformationhistories (SFHs) with the characteris-
The fraction of galaxies for which the spectroscopic and
tic timescale (τ) varying between 10 Myr and 10 Gyr.
A NIR Spectroscopic Survey of z ∼2.3 Galaxies 7
Fig. 6.— In this figure weillustratethe accuracy of continuum Fig. 7.—Inthisdiagramweexaminethecauseoferrorsincon-
redshifts,byderivingzcont forthe19emission-linegalaxiesinthe tinuumredshifts. Asthemodelingismainlydrivenbytheoptical
sample. The scatter and the systematic offset in ∆z/(1+z) are break,weexpectandindeedfindlessaccuratecontinuumredshifts
listed in the figure. The continuum redshifts are ∼ 4−7 times forgalaxiesforwhichalargepartofthebreakfallsbetweenatmo-
moreaccuratethanphotometricredshifts,andshownosignificant sphericwindowsoroutsidethespectrum(panela,shadedregions).
systematic offset. Galaxies without emission lines generally have Thesegalaxiesareindicatedbygraysymbolsinallpanels. Inpan-
largerbreaks,sotheirzcont mayevenbemoreaccurate. els b and c we show that for galaxies with bluer SEDs, and thus
weaker optical breaks, the continuum redshifts are less accurate.
ThereisnoclearcorrelationwiththeS/Nofthespectruminpanel
We leave redshift as a free parameter for the galax-
d.
ies without emission lines. Furthermore, we adopt the
Calzetti et al. (2000) reddening law and allow 41 values
for A between 0 en 4 mag. We compute the χ2 sur-
V
faceas functionofallstellarpopulationparameters. For
all grid points we assume the Salpeter (1955) IMF, and
solar metallicity. A Chabrier (2003) or a Kroupa (2001)
IMF yield stellar masses and SFRs which are a factor of
∼ 2 lower. The mass differences when using the stellar
population library by Maraston (2005) are discussed in
Kannappan & Gawiser (2007) and Wuyts et al. (2007).
We derive 1σ confidence intervals on the redshifts and
stellarpopulationpropertiesusing200MonteCarlosim-
ulations. We vary all bins of the low-resolution spectra
according to their uncertainties, and fit the simulated
spectra using the same procedure as described above.
Next, we determine the contour in the original χ2 sur-
face that encompasses 68% of the Monte Carlo simula-
tions (see Papovich et al. 2003; Kriek et al. 2006a). The
1σ confidence intervals for all properties are the mini-
mum and maximum values that are allowed within this
χ2 contour.
For the emission-line galaxies, we removed the emis-
sion line fluxes from the spectra before fitting. This is
differenttoourpreviousmethodpresentedinKriek et al.
(2006a) in which we mask the bins that are contami- Fig. 8.— left: Galaxy templates as used by the original photo-
metricredshiftcode. right: Galaxytemplates asconstructedfrom
nated by emission lines. The difference in modeling re-
theGNIRSsample.
sults(althoughconsistentwithintheerrors)comparedto
Kriek et al. (2006b, 2007) are due to this improvement sionline redshifts with the continuumredshifts. We find
and updates of the broadband photometry catalogs. All ascatterof∆z/(1+z)=0.019andnosignificantsystem-
spectroscopicredshiftsandcorrespondingstellarpopula- atic offset. In Figure 7 we examine causes of the errors.
tion properties are listed in Table 2. First, as the modeling is driven by the optical break, we
In order to test the accuracy of the continuum red- expect this method to be less accurate if the break falls
shifts, we also fit the emission-line galaxieswith redshift between atmospheric windows. In the Figure 7a we in-
as a free parameter. In Figure 6 we compare the emis- deed find that galaxies for which the break falls outside
8 Kriek et al.
TABLE 2
Spectroscopic modelingresults
age τ AV M SFR M/LV
ID za Gyr Gyr mag 1011M⊙ M⊙yr−1 (M/LV)⊙ V U-V
1030-32 2.34+−00..0122 1.01+−00..1221 0.040+−00..008300 0.0+−00..40 1.40+−00..3189 0.0+−00..10 0.85+−00..2087 -23.29+−00..1073 0.70+−00..1017
1030-101 1.77+−00..0191 0.57+−00..7117 0.020+−00..103100 1.5+−00..58 3.16+−01..7154 0.0+−40..40 2.42+−00..6534 -23.03+−00..1248 1.18+−00..1101
1030-301 1.893 0.20+−00..6100 0.040+−00..201300 2.8+−00..25 4.48+−10..2943 87.0−+28954.3.8 3.25+−00..8650 -23.10+−00..0053 1.10+−00..1046
1030-609 1.800 0.20+−00..5120 0.120+−90..808905 1.8+−00..22 0.70+−00..2197 159.8−+114171..91 0.91+−00..3250 -22.46+−00..0052 0.35+−00..0075
1030-807 2.367 0.72+−00..0592 0.100+−00..002805 0.0+−10..40 0.97+−00..5051 1.0+−60..39 0.59+−00..3070 -23.29+−00..0041 0.46+−00..1011
1030-1531 2.613 0.57+−00..5177 0.650+−90..335500 0.8+−00..11 1.67+−00..6129 227.1−+7526..73 0.56+−00..1086 -23.92+−00..0045 0.06+−00..0072
1030-1611 1.93+−00..1144 0.57+−10..0347 0.120+−00..118100 1.7+−01..61 2.25+−00..5627 20.7−+6280..67 1.97+−00..9481 -22.85+−00..3167 0.96+−00..1152
1030-1813 2.56+−00..1042 0.51+−00..2010 0.020+−00..006100 0.8+−00..15 5.04+−01..9235 0.0+−10..70 0.96+−00..2105 -24.42+−00..0133 0.72+−00..0079
1030-1839 2.312 2.00+−01..7459 4.000+−63..080000 1.3+−00..34 2.83+−01..9243 143.2−+18299.7.0 1.83+−00..4892 -23.22+−00..0078 0.53+−00..0191
1030-2026 2.511 1.14+−00..2597 0.200+−00..005800 0.4+−00..91 2.68+−10..4059 6.0−+207.7.2 1.11+−00..5094 -23.71+−00..0011 0.78+−00..1005
1030-2329 2.236 0.81+−00..3630 0.150+−00..015400 0.7+−10..24 1.44+−00..4312 5.8−+354.8.9 1.15+−00..3276 -23.00+−00..0052 0.73+−00..0160
1030-2559 2.39+−00..0156 0.57+−10..0147 0.010+−00..109000 0.7+−00..57 2.10+−00..6687 0.0+−20..50 1.07+−00..6223 -23.32+−00..1220 0.83+−00..1151
1030-2728 2.504 0.51+−00..0262 0.080+−00..002700 0.9+−00..43 2.34+−00..2482 6.4+−46..04 1.04+−00..1128 -23.63+−00..0022 0.69+−00..0056
1030-2927 1.82+−00..2106 0.51+−00..7471 0.150+−00..115400 1.3+−01..53 1.08+−00..3318 31.7−+5301..17 1.23+−00..3456 -22.61+−00..3302 0.63+−00..1125
1256-0 2.31+−00..0057 0.57+−00..4248 0.080+−00..007700 1.2+−00..66 4.06+−10..0778 5.2−+354.2.2 1.51+−00..4267 -23.82+−00..1018 0.90+−00..1008
1256-142 2.37+−00..0155 0.40+−20..3152 0.010+−00..309000 1.3+−01..43 3.22+−11..0008 0.0−+106.0.1 1.46+−00..7468 -23.60+−00..2017 0.89+−00..1151
1256-519 1.857 0.40+−20..6300 0.250+−90..725400 3.1+−01..30 4.27+−31..4683 523.4−+851271..59 4.32+−31..6630 -22.74+−00..0054 1.14+−00..1196
1256-1207 1.84+−00..0054 0.40+−00..7142 0.025+−00..102155 1.5+−01..30 2.06+−00..3542 0.0+−90..80 1.70+−00..2491 -22.96+−00..0191 0.96+−00..0088
1256-1967 2.02+−00..0079 0.20+−00..0100 0.025+−00..001155 1.4+−00..41 2.48+−00..4303 3.7−+438.7.1 0.99+−00..0097 -23.75+−00..1162 0.53+−00..0053
HDFS1-259 2.249 0.81+−00..6532 5.000+−54..080000 1.5+−00..22 2.00+−00..4680 287.1−+116068..43 1.14+−00..2372 -23.36+−00..0043 0.37+−00..0180
HDFS1-1849 2.31+−00..0098 0.40+−00..7240 0.080+−00..102700 1.6+−01..50 3.47+−10..0806 35.0−+9315..40 1.59+−00..6407 -23.59+−00..1133 0.84+−00..1116
HDFS2-509 2.918 0.57+−00..0000 0.100+−00..000000 0.0+−00..10 3.54+−00..7000 15.0+−30..00 0.52+−00..0004 -25.02+−00..1000 0.34+−00..0005
HDFS2-1099 2.73+−00..0187 1.01+−00..0300 0.150+−00..000500 0.0+−00..40 2.83+−00..8494 2.8+−41..95 0.76+−00..2025 -24.18+−00..1180 0.59+−00..1003
HDFS2-2046 2.24+−00..0094 0.29+−00..2088 0.012+−00..006082 0.8+−00..56 1.41+−00..5205 0.0+−60..50 0.71+−00..2111 -23.49+−00..0174 0.51+−00..0047
ECDFS-4454 2.351 0.72+−00..0391 0.120+−00..003200 0.1+−10..21 1.20+−00..9126 3.2−+609.4.7 0.62+−00..5005 -23.46+−00..0063 0.45+−00..1045
ECDFS-4511 2.122 2.75+−00..2555 10.000+−07..000000 1.1+−00..01 3.95+−00..4487 165.9+−13.45.9 1.80+−00..1270 -23.60+−00..0023 0.46+−00..0042
ECDFS-4713 2.309 0.51+−00..2010 0.150+−00..105300 0.7+−00..32 2.73+−00..9134 80.0−+14102.1.0 0.71+−00..2037 -24.21+−00..0025 0.35+−00..0060
ECDFS-5856 2.56+−00..1024 0.40+−00..5102 0.030+−00..009200 1.0+−00..48 2.74+−00..7678 0.0+−90..60 1.06+−00..3216 -23.78+−00..0163 0.73+−00..0191
ECDFS-6842 2.40+−00..0180 0.57+−00..2248 0.150+−00..015000 1.5+−00..35 5.35+−01..9407 103.1−+17327.4.7 1.66+−00..3415 -24.02+−00..1151 0.81+−00..0163
ECDFS-6956 2.037 0.05+−00..4060 0.010+−90..909000 1.4+−00..21 0.56+−00..6020 43.9−+400.02.3 0.41+−00..3080 -23.10+−00..0009 0.08+−00..1030
ECDFS-9822 1.612 0.10+−00..3000 0.020+−10..908005 1.9+−00..31 0.68+−00..2097 25.2−+21790.0.1 0.97+−00..4053 -22.36+−00..1002 0.50+−00..0055
ECDFS-11490 2.34+−00..0072 0.51+−00..0160 0.065+−00..001555 0.1+−00..21 0.95+−00..0181 0.7+−00..67 0.52+−00..0036 -23.41+−00..0047 0.37+−00..0045
ECDFS-12514 2.024 0.40+−00..1270 10.000+−09..080000 1.3+−00..11 1.25+−00..2228 369.1−+9110.57.7 0.62+−00..0192 -23.51+−00..0073 0.10+−00..0057
ECDFS-13532 2.336 0.90+−00..5530 0.250+−00..715000 1.0+−00..95 1.88+−00..6496 26.8−+21187.4.2 1.42+−00..5395 -23.05+−00..0052 0.72+−00..0182
ECDFS-16671 2.61+−00..0021 0.05+−00..0000 0.010+−00..003000 0.9+−00..40 1.08+−00..1015 83.9−+110.081.2 0.26+−00..0011 -24.30+−00..0015 -0.16+−00..0015
CDFS-6202 2.225 0.20+−00..0185 0.800+−90..270900 1.7+−00..12 1.26+−00..1461 649.8−+154853..16 0.64+−00..0179 -23.48+−00..0043 0.16+−00..0034
Note. — The stellar population propertiesare derived by fitting the low resolution spectra together with the optical photometry by
Bruzual&Charlot(2003)stellarpopulationmodels. Fortheemission-linegalaxiestheredshiftwasfixedtozspec. Theerrorsrepresentthe
68%confidenceintervalsderivedusing200MonteCarlosimulations.
a Thecontinuumredshiftsarethoseforwhichwegivetheconfidenceintervals.
the spectrum or in between the J and H band have less these galaxies generally have larger breaks (Kriek et al.
accurate z . 2006b).
cont
Furthermore, we expect the continuum redshifts to be Finally, in Figure 7d we examine the correlation be-
moreaccurateforgalaxieswithstrongopticalbreaks. In tween(z −z )/(z +1)and the S/Nper bin. Re-
cont line line
the Figures 7b and c we show the correlation with rest- markably, we do not find an obvious trend. However,
frame U −V color and observed R−K color. If we ex- the different causes for errors and the small size of the
cludethegalaxiesforwhichthebreakfallsbetweenatmo- sample may have affected a possible correlation.
sphericwindows(gray symbols), weindeedfindthatblue
galaxieswithweakopticalbreaks,havelessaccuratecon- 3.2. Photometric Redshifts and Galaxy Properties
tinuum redshifts. The continuum redshifts for galaxies
Photometricredshiftsarederivedusingthemethodde-
without emission lines might even be more accurate, as
scribed by Rudnick et al. (2001, 2003). The code fits a
A NIR Spectroscopic Survey of z ∼2.3 Galaxies 9
TABLE 3
Photometric modeling results
age τ AV M SFR M/LV
ID zphota Gyr Gyr mag 1011M⊙ M⊙yr−1 (M/LV)⊙ Va U-Va
1030-32 2.26+−00..2248 2.20+−01..5850 0.250+−00..125400 0.0+−20..10 2.35+−11..8019 0.2−+100.2.0 1.49+−31..8021 -23.25+−00..4365 0.91+−00..7218
1030-101 2.18+−00..0582 0.90+−10..5602 0.010+−00..209000 0.2+−10..72 2.70+−30..5716 0.0−+105.0.8 0.91+−40..4386 -23.92+−00..9059 0.72+−10..3195
1030-301 2.94+−00..0668 1.01+−00..5390 0.120+−00..018100 0.0+−00..90 5.21+−41..3095 1.2−+111.2.0 0.71+−20..3188 -24.91+−00..9126 0.53+−00..3010
1030-609 2.22+−00..4922 0.51+−20..2540 0.300+−90..720900 1.4+−11..34 1.59+−21..2169 148.7−+4164281.3.8 0.99−+105.7.015 -23.26+−20..1409 0.36+−00..2274
1030-807 2.44+−00..3584 0.10+−20..5009 0.030+−90..907200 2.2+−02..80 1.90+−41..1293 262.5−+6246525.5.7 1.02−+204.7.167 -23.42+−00..8467 0.42+−00..1065
1030-1531 2.74+−00..1006 1.28+−00..9827 1.000+−90..090000 0.5+−00..35 2.64+−11..7517 131.9−+114097..19 0.70+−00..2374 -24.19+−00..1127 0.17+−00..1179
1030-1611 2.32+−00..0368 1.01+−00..8881 0.150+−00..115400 0.3+−10..63 1.85+−10..2515 1.9−+216.9.7 0.93+−00..8494 -23.50+−00..6079 0.46+−00..5090
1030-1813 2.32+−00..1166 0.57+−10..4437 0.030+−90..907200 0.7+−20..37 3.29−+111.5.092 0.0−+204.017.3 0.91+−70..6090 -24.14+−00..9781 0.17+−10..0060
1030-1839 2.58+−00..2308 1.61+−01..7392 0.800+−90..270200 1.0+−01..70 3.78+−12..8579 96.2−+39901.5.7 1.53+−21..6195 -23.73+−00..6301 0.28+−00..3013
1030-2026 2.68+−00..1300 1.28+−00..6727 0.200+−00..110000 0.0+−10..60 2.44+−30..4397 2.7−+908.6.9 0.79+−10..9376 -23.97+−00..3153 0.68+−00..6099
1030-2329 2.30+−00..4380 0.20+−20..5155 0.050+−90..905400 2.1+−02..91 2.25+−31..9183 96.9−+29063.90.0 1.48−+110.1.701 -23.20+−00..6627 0.63+−00..1126
1030-2559 2.18+−00..3240 0.57+−20..1487 0.100+−00..400900 1.6+−11..46 2.72+−21..3572 11.5−+41618.5.0 1.96+−31..7534 -23.10+−00..3496 0.89+−00..3279
1030-2728 2.60+−00..3322 0.20+−10..2135 0.025+−90..907155 1.7+−11..33 3.19+−51..4366 4.8−+345.851.5 1.17+−20..5668 -23.84+−00..4389 0.57+−00..1141
1030-2927 2.26+−00..2542 0.29+−20..1214 0.080+−90..902700 1.6+−11..06 2.10+−20..9845 93.1−+19637.13.3 1.06+−60..6611 -23.49+−00..8324 0.50+−00..1142
1256-0 2.26+−00..0048 1.01+−00..1521 0.120+−00..013100 0.0+−00..80 1.90+−00..7186 0.4+−20..24 0.82+−00..4212 -23.67+−00..1137 0.65+−00..0012
1256-142 2.18+−00..1142 2.20+−02..5050 0.300+−00..120900 0.1+−20..31 2.76+−21..0333 0.8−+806.8.7 1.61+−20..7986 -23.33+−00..3240 0.92+−00..4158
1256-519 2.00+−00..2066 0.20+−00..9140 0.030+−00..202200 2.7+−01..35 4.94+−11..5489 23.6−+52439.6.0 2.92+−11..5171 -23.32+−00..1445 1.34+−00..3614
1256-1207 2.10+−00..0424 0.51+−00..7272 0.030+−00..102200 1.0+−01..80 2.37+−00..9980 0.0−+103.0.6 1.22+−20..2416 -23.47+−00..7073 1.00+−10..5029
1256-1967 2.16+−00..0762 0.10+−00..6020 0.012+−00..008082 1.9+−01..29 3.51+−01..4886 7.4−+176.44.4 1.03+−00..4665 -24.08+−10..3049 0.32+−00..1004
HDFS1-259 2.24+−00..7168 0.72+−20..0731 0.650+−90..365400 1.4+−11..34 2.21+−31..4888 212.2−+6201229.2.4 1.12+−60..1901 -23.49+−00..3915 0.35+−00..0273
HDFS1-1849 2.24+−00..2144 0.20+−10..5105 0.040+−90..906300 2.1+−01..99 3.22+−31..6358 62.6−+26621.60.0 1.59+−20..7853 -23.52+−00..2337 0.71+−00..0176
HDFS2-509 2.78+−00..4004 0.57+−10..6337 0.120+−90..808505 0.5+−10..55 5.62−+127.5.963 51.7−+34035.11.7 0.70+−00..8406 -25.02+−00..2208 0.30+−00..0257
HDFS2-1099 2.40+−00..1102 0.81+−00..2410 0.120+−00..003700 0.4+−00..94 2.47+−10..2755 3.2−+226.5.0 0.87+−00..8418 -23.88+−00..2134 0.67+−00..0161
HDFS2-2046 2.36+−00..1140 0.20+−00..5120 0.010+−00..009000 1.2+−01..82 2.05+−00..8896 0.0−+204.02.2 0.88+−00..1466 -23.67+−00..1255 0.33+−00..1066
ECDFS-4454 2.50+−00..1028 0.72+−00..1090 0.120+−00..003000 0.2+−00..52 1.94+−10..2388 5.2−+126.4.6 0.68+−20..3265 -23.89+−00..3129 0.83+−00..6593
ECDFS-4511 2.70+−00..1026 1.14+−00..1243 0.250+−00..005500 0.1+−00..11 4.67+−01..8109 26.0+−87..53 0.82+−00..4245 -24.64+−00..2143 0.52+−00..2107
ECDFS-4713 2.30+−00..3064 1.01+−10..7743 0.500+−90..540200 1.0+−01..50 5.21+−73..2881 203.0−+510984..40 1.26+−00..8992 -24.29+−00..1448 0.32+−00..0244
ECDFS-5856 3.32+−00..4102 1.61+−00..0890 0.300+−00..210800 0.7+−00..87 12.20−+180.3.700 25.6−+12717.6.8 2.75+−02..2235 -24.37+−00..0537 -0.12+−00..7126
ECDFS-6842 2.96+−00..0644 0.90+−10..1109 0.150+−00..605300 0.2+−10..32 4.55−+116.4.524 9.5−+341.80.5 0.77+−40..6320 -24.68+−00..7162 0.40+−00..4163
ECDFS-6956 2.86+−00..0142 0.90+−10..1100 0.200+−00..300000 0.0+−00..20 3.14+−31..4167 22.3−+200.0.9 0.63+−00..8156 -24.49+−00..4142 0.36+−00..7482
ECDFS-9822 2.60+−00..1082 0.81+−10..5099 0.150+−40..805300 0.3+−10..33 2.93+−81..2177 11.8−+377.07.4 0.76+−20..4555 -24.22+−00..1259 0.98+−10..0589
ECDFS-11490 3.58+−00..1648 0.81+−00..8294 0.150+−00..305500 0.0+−00..80 4.43−+125.6.601 17.8−+145.18.2 0.61+−00..8481 -24.91+−00..5457 0.08+−10..3314
ECDFS-12514 2.40+−00..3208 0.51+−20..0590 0.120+−90..818100 0.2+−10..42 1.87+−81..1717 28.7−+21274.26.3 0.58+−20..8313 -24.02+−00..6398 -0.13+−00..3202
ECDFS-13532 2.86+−00..1264 1.28+−00..9327 0.250+−00..215000 0.4+−00..54 3.83+−51..1949 12.3−+384.4.4 1.07+−40..7765 -24.14+−00..4250 1.29+−00..7634
ECDFS-16671 3.02+−00..0142 0.72+−10..2281 0.250+−10..715000 0.0+−00..20 2.68+−40..9635 81.9−+7161..44 0.41+−00..4190 -24.79+−00..1147 -0.11+−00..3370
CDFS-6202 3.04+−00..0664 0.72+−00..1090 0.150+−00..005000 0.0+−00..20 3.19+−00..6080 22.9−+202.0.1 1.45+−01..0002 -23.60+−00..0036 0.36+−00..0088
Note. —ThestellarpopulationpropertiesarederivedbyfittingthephotometrybyBruzual&Charlot(2003)models,fixingtheredshiftto
zphot. Thezphotconfidenceintervalsarederivedusing100MonteCarlosimulations(Rudnicketal.2001,2003). Weusethesamesimulations
incombinationwiththebest-fitzphot foreachsimulationtoderivethe1σ (68%)confidenceintervalsofthestellarpopulationproperties.
a DerivedusingmethodbyRudnicketal.(2001,2003)
linearnon-negativesuperpositionofthe8templatespre- metric uncertainties.
sented in the left panel of Figure 8 to the broadband We determine photometric stellar populations proper-
photometry using χ2 minimization. This template set ties by fitting Bruzual & Charlot (2003) stellar popula-
consist of the empirical E, Sbc, Scd and Im templates tion models to the broadband SEDs, with the redshift
from Coleman, Wu & Weedman (1980), the two least fixed to the best-fit z . We apply the same grid and
phot
reddened starburst templates from Kinney et al. (1996), reddening law (Calzetti et al. 2000) as for the spectro-
and a 1 Gyr and 10 Myr Bruzual & Charlot (2003) sin- scopic fitting (§ 3.1). The 100 simulations of the pho-
gle stellarpopulationwith a Salpeter (1955)initial mass tometryincombinationwiththeir best-fitz areused
phot
function (IMF). Confidence intervals on z are deter- to derive the confidence intervals on the stellar popula-
phot
mined using 100 Monte-Carlo simulations, in which the tion properties (see procedure as described in § 3.1).
broadbandphotometry is variedaccordingto the photo- Theerrorsonthephotometricredshiftsandrest-frame
10 Kriek et al.
Fig. 9.— Spectroscopic vs. photometric redshifts. The scatter
andoffsetin∆z/(1+z)forthefullsampleare0.13and0.08respec-
tively. Thesquaresandcirclesrepresentthecontinuum and emis-
sionlineredshiftsrespectively,forgalaxiesinthedeep(filledsym-
bols)andwide(opensymbols)MUSYCfields. DRGs(J−K>2.3,
Franxetal. 2003) are indicated in red and non-DRGs are shown
inblue. Correspondingsystematicandrandomerrorsaregivenin
Table4.
TABLE 4
Comparison between photometricandspectroscopic
redshifts (∆z/(1+z))
Fig. 10.— Spectroscopic (black solid histograms) and photo-
metric (diagonally hatched histograms) redshift distributions (top
All Deep Wide panels)andtheirsummedprobabilitydistributions(middlepanels)
scat offs scat offs scat offs forthefullspectroscopicsample(leftpanels),andforthosegalaxies
withdeepNIRphotometry(rightpanels). Thescatterandsystem-
All 0.13 0.08 0.08 0.03 0.24 0.17 aticoffsetin∆z/(1+z)aregiveninthetoprightcornerinthetop
twopanels. Thelowerpanels show thedistributionof ∆z/(1+z)
zcont 0.13 0.06 0.10 0.01 0.31 0.21 forthefull(left)anddeep(right)samplerespectively.
zlines 0.10 0.07 0.06 0.03 0.24 0.16
DRGs 0.11 0.08 0.07 -0.00 0.32 0.22
and a systematic offset of ∆z/(1+z)= 0.08, such that
DRGs,zcont 0.15 0.01 0.09 -0.05 0.35 0.23
DRGs,zlines 0.07 0.04 0.06 0.03 0.30 0.21 ourphotometricredshiftsareonaveragetoohigh. These
non-DRGs 0.13 0.08 0.13 0.08 0.18 0.12 values are 0.08 and 0.03 respectively for galaxies in the
non-DRGs,zcont 0.13 0.08 0.12 0.08 - - MUSYCdeepfieldsonly(seeTable4). Thedistributions
non-DRGs,zlines 0.14 0.09 0.16 0.11 0.09 0.06 of the spectroscopic and photometric redshifts for the
full and the deep sample, are shown in Figure 10. Using
Note. — In this table we give the scatter and systematic optical spectroscopy and the same photometric redshift
offset (∆z/(1+z))betweenthephotometricandspectroscopic codeandtemplates,Wuyts et al.(2008)findascatterof
redshiftsforthefullsample,anddifferentsubsamples. ∆z/(1+z)∼0.06andamedianoffsetof∆z/(1+z)∼0.02
for galaxies with deep NIR photometry in the same red-
luminositiesarederivedindependently usingthe method
shiftrange. Thisresultissimilartoourswhencomparing
by Rudnick et al. (2001, 2003). Due to the independent
only to the galaxies with deep NIR photometry.
fitting procedures, we do not have a combined χ2 distri-
The broadband photometric redshifts of galaxies with
bution for M and L . For the confidence intervals on
V emission lines have smaller errors than those of galaxies
M/L we take the minimum and maximum M/L cor-
V V without emission lines, especially when we exclude the
responding to the 68% best stellar population fits, i.e.,
galaxies with shallow NIR photometry. However, our
with the lowest χ2. All photometric redshifts, stellar
sample is small, and studies of largerspectroscopic sam-
population properties and corresponding confidence in-
ples are needed to verify this result. If this result still
tervals are listed in Table 3.
holds for larger samples, it probably implies that the
qualityofphotometricredshifts,whicharecalibratedus-
3.3. Direct Comparison
ing just emission-line redshifts, may be overestimated.
In Figure 9 we compare z and z for all indi- The broadband photometric redshifts of galaxies with
spec phot
vidual galaxies. We find a scatter of ∆z/(1+z)= 0.13 emission lines have smaller errors than those of galax-