Keck DEIMOS Spectroscopy of a GALEX UV-Selected Sample from the Medium Imaging Survey

Abstract

We report results from a pilot program to obtain spectroscopy for objects detected in the Galaxy Evolution Explorer (GALEX) Medium Imaging Survey (MIS). Our study examines the properties of galaxies detected by GALEX fainter than the Sloan Digital Sky Survey (SDSS) spectroscopic survey. This is the first study to extend the techniques of Salim and coworkers to estimate stellar masses, star formation rates (SFRs), and the b (star formation history) parameter for star-forming galaxies out to z ~ 0.7. We obtain redshifts for 50 GALEX MIS sources reaching NUV = 23.9 (AB mag) having counterparts in the SDSS Data Release 4 (DR4). Of our sample, 43 are star-forming galaxies with z < 0.7, 3 have emission-line ratios indicative of active galactic nuclei with z < 0.7, and 4 objects with z > 1 are QSOs, 3 of which are not previously cataloged. We compare our sample to a much larger sample of ~50,000 matched GALEX/SDSS galaxies with SDSS spectroscopy; while our survey is shallow, the optical counterparts to our sources reach ~3 mag fainter in SDSS r than the SDSS spectroscopic sample. We use emission-line diagnostics for the galaxies to determine that the sample contains mostly star-forming galaxies. The galaxies in the sample populate the blue sequence in the NUV − r versus Mr color-magnitude diagram. The derived stellar masses of the galaxies range from 108 to 1011 M☉, and derived SFRs are between 10−1 and 102 M☉ yr−1. Our sample has SFRs, luminosities, and velocity dispersions that are similar to the samples of faint compact blue galaxies studied previously in the same redshift range by Koo and collaborators, Guzmán and collaborators, and Phillips and collaborators. However, our sample is ~2 mag fainter in surface brightness than the compact blue galaxies. We find that the star formation histories for a majority of the galaxies are consistent with a recent starburst within the last 100 Myr.

Full text

Page 1

arXiv:0706.3424v1 [astro-ph] 23 Jun 2007
Draft version February 5, 2008
Preprint typeset using LATEX style emulateapj v. 10/09/06
KECK/DEIMOS SPECTROSCOPY OF A GALEX UV SELECTED SAMPLE FROM THE MEDIUM IMAGING
SURVEY∗
Ryan P. Mallery1, R. Michael Rich1, Samir Salim1, Todd Small2, Stephane Charlot3,4, Mark Seibert2, Ted
Wyder2, Tom A. Barlow2, Karl Forster2, Peter G. Friedman2, D. Christopher Martin2, Patrick Morrissey2,
Susan G. Neff5, David Schiminovich6, Luciana Bianchi7, Jose Donas8, Timothy M. Heckman9, Young-Wook
Lee10, Barry F. Madore11, Bruno Milliard8, Alex S. Szalay9, Barry Y. Welsh12, Suk Young Yi10
Draft version February 5, 2008
ABSTRACT
We report results from a pilot program to obtain spectroscopy for objects detected in the Galaxy
Evolution Explorer (GALEX) Medium Imaging Survey (MIS). Our study examines the properties of
galaxies detected by GALEX fainter than the Sloan Digital Sky Survey (SDSS) spectroscopic survey.
This is the first study to extend the techinques of (Salim et al. 2005) to estimate stellar masses, star
formation rates (SFR) and the b (star formation history) parameter for star-forming galaxies out to
z ∼ 0.7. We obtain redshifts for 50 GALEX MIS sources reaching NUV = 23.9 (AB mag), having
counterparts in the SDSS Data Release 4 (DR4). Of our sample, 43 are starforming galaxies with
z < 0.7, 3 have emission line ratios indicative of AGN with z < 0.7, and 4 objects with z > 1 are
QSOs, 3 of which are not previously cataloged. We compare our sample to a much larger sample of
∼50,000 matched GALEX/SDSS galaxies with SDSS spectroscopy; while our survey is shallow, the
optical counterparts to our sources reach ∼3 magnitudes fainter in SDSS r than the SDSS spectroscopic
sample. We use emission line diagnostics for the galaxies to determine that the sample contains mostly
star-forming galaxies. The galaxies in the sample populate the blue sequence in the NUV − r vs Mr
color-magnitude diagram. The derived stellar masses of the galaxies range from 108to 1011M⊙and
derived SFRs are between 10−1and 102M⊙yr−1. Our sample has SFRs, luminosities, and velocity
dispersions that are similar to the samples of faint compact blue galaxies studied previously in the
same redshift range by Koo et al (1995), Guzm´ an et al. (1996), & Phillips et al. (1997). However,
our sample is ∼ 2 mag fainter in surface brightness than the compact blue galaxies. We find that the
star-formation histories for a majority of the galaxies are consistent with a recent starburst within
the last 100 Myr.
Subject headings: galaxies: active galaxies:redshifts galaxies: starburst ultraviolet: galaxies
∗SOME OF THE DATA PRESENTED HEREIN WERE OB-
TAINED AT THE W.M. KECK OBSERVATORY, WHICH IS
OPERATED AS A SCIENTIFIC PARTNERSHIP AMONG THE
CALIFORNIA INSTITUTE OF TECHNOLOGY, THE UNIVER-
SITY OF CALIFORNIA AND THE NATIONAL AERONAU-
TICS AND SPACE ADMINISTRATION. THE OBSERVATORY
WAS MADE POSSIBLE BY THE GENEROUS FINANCIAL
SUPPORT OF THE W.M. KECK FOUNDATION.
1Department of Physics and Astronomy, University of Califor-
nia, Los Angeles, CA 90095-1562
2California Institute of Technology,MC 405-47, 1200 East Cali-
fornia Boulevard, Pasadena, CA 91125
3Max-Planck Institut f¨ ur Astrophysik, D-85748 Garching, Ger-
many
4Institut d’Astrophysique de Paris, CNRS, 98 bis boulevard
Arago, F-75014 Paris, France
5Laboratory for Astronomy and Solar Physics, NASA Goddard
Space Flight Center, Greenbelt, MD 20771
6Department of Astronomy, Columbia University, New York,
NY 10027
7Center for Astrophysical Sciences, The Johns Hopkins Univer-
sity, 3400 N. Charles St., Baltimore, MD 21218
8Laboratoire d’Astrophysique de Marseille, BP 8, Traverse du
Siphon, 13376 Marseille Cedex 12, France
9Department of Physics and Astronomy, The Johns Hopkins
University, Homewood Campus, Baltimore, MD 21218
10Center for Space Astrophysics, Yonsei University, Seoul 120-
749, Korea
11Space Sciences Laboratory, University of California at Berke-
ley, 601 Campbell Hall, Berkeley, CA 94720
11Observatories of the Carnegie Institution of Washington, 813
Santa Barbara St., Pasadena, CA 91101
12Space Sciences Laboratory, University of California at Berke-
ley, 601 Campbell Hall, Berkeley, CA 94720
1. INTRODUCTION
Following the successful launch and early operations
of the Galaxy Evolution Explorer satellite in 2003, we
decided to undertake an initial spectroscopic assay of
sources detected by GALEX using the DEIMOS mul-
tiobject spectrograph (Faber et al. 2003) on the Keck II
telescope. Although GALEX data have been matched
to relatively deep optical surveys (Schiminovich et al.
2005) our study is the first to extend the modeling tech-
nique of Salim et al. (2005) from z = 0.25 to z ∼ 0.5.
The power of the GALEX Medium Imaging Survey to
select interesting objects is noteworthy. We recall that
the MIS observations are 1500 sec in duration, imaging
FUV and NUV simultaneously (Martin et al. (2005a) &
Morrissey et al. (2005)). The GALEX satellite has in-
vestigated the ultraviolet universe to z ∼ 1 with the goal
of determining among other things, the star formation
rate (SFR) of galaxies in the local universe, z ? 0.25
(Salim et al. 2005; Martin et al. 2005b; Wyder et al.
2005; Treyer et al. 2005) and the evolution of the global
star formation density out to z ∼ 1 (Schiminovich et al.
2005). The ultimate goal of such work along with other
surveys and observations of galaxies at similar and higher
redshifts is to constrain the baryonic physics of galaxy
formation and evolution. Questions still remain as to
how the SFR depends on different factors such as envi-
ronment, mass, morphology etc. (Martin et al. 2005a).
Previous work deriving SFRs with a UV-selected sam-

Page 2

2
ple was carried out by Sullivan et al. (2000) who de-
rive SFRs both from UV luminosities at 2000˚ A and Hα
luminosities from optical spectra for a sample of galax-
ies at redshifts of 0 < z < 0.4 detected in the UV by
the balloon-borne telescope FOCA (Milliard, B. et al.
1992). Estimates of local star formation rates before this
were only possible through either optical emission line lu-
minosities of the recombination lines of Hydrogen, mostly
Hα, and [OII]λ3727 or the far infrared luminosity from
10-100µm (Kennicutt 1998).
In this paper we present the observations and results
from the pilot program for a sample of objects detected
by GALEX and SDSS with spectra obtained from the
DEIMOS spectrograph at the Keck II telescope. The
analysis of this sample is enhanced greatly by the tech-
niques of Salim et al.(2005) that have been expanded
to include the derivation of galaxy physical parameters
such as stellar mass, from GALEX + SDSS photometry
and redshifts. The main goal of this paper is to charac-
terize the galaxies detected by GALEX that are fainter
than SDSS Spectroscopic Galaxy Sample, in terms of
mass, SFR, and UV-optical color.
nantly (about two-thirds of the sample) seeing the faint
blue galaxy population (Ellis 1997) and are also sensi-
tive to QSOs. While GALEX observations have been
undertaken in many deep fields with much higher me-
dian redshifts (Schiminovich et al. 2005) we have the
advantage in this sample of obtaining spectroscopy with
Keck/DEIMOS, which gives wavelength coverage suffi-
cient to derive some detailed physical parameters from
the spectra. In §2 we describe the Sample and the pho-
tometric and spectroscopic data. Redshift and emission
line flux measurements from the spectroscopy are pre-
sented in §3. The distributions of UV color, UV-optical
color, and the SDSS r magnitude for the sample are de-
scribed in §4. The derived galaxy parameters from SED
fits are described in §5. We give a discussion and sum-
mary of our findings in §6 and 7. Throughout this paper
we assume Ho=70 km s−1Mpc−1, Ωm=0.3, and ΩΛ=0.7.
We are predomi-
2. SAMPLE AND OBSERVATIONS
2.1. GALEX/SDSS Data
GALEX is a NASA Small Explorer Mission aimed to
survey the UV emission from Galactic and extragalactic
sources from 700km circular orbit (Martin et al. 2005a;
Morrissey et al. 2005). GALEX images the sky simul-
taneously in two bands, the far-UV (FUV 1344-1786˚ A)
and the near-UV (NUV 1771-2831˚ A). Each GALEX cir-
cular field is 1.25 deg. in diameter. We use FUV and
NUV magnitudes and magnitude errors derived in ellip-
tical apertures14. The photometry is taken from the
GALEX Internal Release 1.1 (IR1.1). The GALEX MIS
photometry for tile number 10273 has a limiting magni-
tude, mlim(AB)= 23.
We use optical photometry for our objects obtained
from SDSS Data Release 4 (DR4) (Abazajian et al.
2004). The SDSS photometric data are taken with the
2.5m telescope at Apache Point Observatory.
ing is obtained in ugriz bands (Fukugita et al. 1996;
Smith et al. 2002). The imaging data are photometri-
cally (Hogg et al. 2001) and astrometrically (Pier et al.
Imag-
14GALEX source detection and measurement is obtained from
SExtractor (Bertin, & Arnouts 1996)
2003) calibrated. An overview of the SDSS data pipelines
and products can be found in Stoughton et al. (2002).
2.2. Sample
Our sample is derived from objects with detections
in the GALEX Medium Imaging Survey (MIS) tile
10273, centered at a right ascension/declination of
17h40m32.07s/ 57o10′45.15′′.
The GALEX field being much larger than the DEIMOS
slit mask, we selected objects from the GALEX field to
populate two DEIMOS slit masks. The two DEIMOS
fields were initially chosen to maximize the number of
MIS objects with optical counterparts in the SDSS clas-
sified as galaxies with FUV −NUV < 0 photometry from
an early version of the GALEX data reduction pipeline.
The number of galaxies matching this criteria was much
smaller than the number of available slits, and so to take
full advantage of the DEIMOS field of view, the slits
were populated by any galaxy detected by both GALEX
and SDSS. In addition, the remainder of available slits
in both fields was populated with blue stellar-like ob-
jects (QSOs and white dwarfs), main sequence outlier
stars (u − g < 1 or g − r < 0), and bright alignment
stars. The first DEIMOS field contains 26 objects classi-
fied as galaxies in SDSS, and 10 blue stellar objects and
main sequence outlier stars. The second DEIMOS field
contains 19 objects classified as galaxies in SDSS, and 9
blue stellar objects and main sequence outlier stars.
2.3. Spectroscopic observations
The spectroscopic data were obtained at the Keck II
telescope with the DEIMOS multi-object spectrometer
on October 1, 2003 (Faber et al. 2003). Spectra were ob-
tained using two slit masks, for a total 64 spectra. The
830 grooves mm−1grating was used with a slit width of
0′′.73 giving a resolution of ∼ 2.5˚ A FWHM. The signal-
to-noise ratio obtained over the continuum of each spec-
trum varied between sources from 1 to ∼ 20, depending
on the brightness of the source. The integration time
for both slit masks was 30 minutes.
of the spectra was not performed as no flux standards
were observed. The spectra cover a wavelength range
of ∼ 5000 − 9000˚ A. Figure 1 shows a panel of several
one dimensional spectra for star-forming galaxies in the
sample at a range of redshifts, plotted in the rest wave-
length, and Figure 2 shows the spectra for objects with
measured z > 1.
Flux calibration
3. REDSHIFT DETERMINATION AND MEASUREMENT OF
EMISSION LINES
In all, we measured 50 redshifts, 45 for sources classi-
fied as galaxies in SDSS and 5 for sources classified as
stars in SDSS. Of those five, three are QSO’s at z > 1
and the other two are star-forming galaxies.
maining 14 of the sources classified as stars by the SDSS
have stellar spectra. The redshifts for objects with de-
tected emission lines were measured by eye. For objects
with z < 1, multiple emission lines were detected includ-
ing either Hα and/or Hβ and [OII]λ3727, [OIII]λ4959
,[OIII]λ5007, [NII]λ6584,[SII]λ6717, and [SII]λ6731. For
all objects where the doublet [OII]λ3727 was detected,
Hα was redshifted beyond the wavelength range of the
detector. For one source, at z=1.028, we detected two
features that we assigned as [OII]λ3727 and MgIIλ2800
The re-

Page 3

3
emission. For the other three sources with z > 1 we find
only 1 feature that we identify as MgIIλ2800, based on
the width of the line and the absence of other features
in the spectrum. Figure 3 shows the redshift distribu-
tion of the sample. The mean redshift of the sample
is zmean = 0.421. Two of the sources in the DEIMOS
sample have corresponding SDSS spectra. The redshifts
obtained for these two objects agree with the SDSS red-
shifts. One of the objects with SDSS spectra is the QSO
at z = 1.028 and the other is a galaxy at z = 0.176.
Both the DEIMOS spectra and the SDSS spectra for the
latter object have significant detections of only Hα and
[NII]6584.
We measure spectral emission line fluxes and equiva-
lent widths for [OII]λ3727, Hβ, [OIII]λ4959, [OIII]λ5007,
Hα, [NII]λ6584, [SII]λ6717, and [SII]λ6731.
equivalent widths and errors for the emission lines were
measured in IDL using the MPFIT function. The con-
tinuum of each spectrum was first fit by a polynomial,
then the emission lines were simultaneously fit with gaus-
sians. Table 1 lists the objects, their equivalent widths
and flux ratios of [OIII]λ5007/Hβ and [NII]λ6584/Hα
for objects with a 3σ measurement of at least one of
the above lines. We did not perform a reddening cor-
rection to the fluxes, since the standard procedure re-
quires either detections of both Hα and Hβ or a mea-
surement of the radio continuum (Osterbrock 1989).
The effect of this on the flux ratios of [OIII]λ5007/Hβ
and [NII]λ6584/Hα is thought to be negligible due to
the small wavelength separation of the emission lines.
In the DEIMOS sample we only have detections of
both Hα and Hβ for 18 objects, all at z < 0.3, and
chose for consistency not to perform corrections on any
of the spectra.We note that the foreground Galac-
tic extinction, E(B-V), calculated from the dust maps
of D.J. Schlegel, D.P. Finkbeiner, & M. Davis (1998) is
∼ .05 magnitudes for the field.
Fourteenobjectshave
Hβ,[OIII]λ5007,Hα,
ure4 showsthe emission
Baldwin, Phillips, & Terlevich
tinguish star-forming galaxies and Type II AGN. The
solid curve is the boundary between star-forming galax-
ies and AGN determined through modeling of starburst
galaxy spectra by Kewley et al. (2001). A star forming
galaxy with only ∼ 20% of the optical emission line flux
due to an AGN would lie above the star-forming/AGN
boundary (Kewley et al. 2001). The dashed curve is the
more stringent demarcation used by Kauffmann et al.
(2003) to distinguish between AGN and star-forming
galaxies in SDSS. The shaded histogram shows the
distribution of emission line ratios for a sample of
51,000 SDSS/GALEX objects (see §4).
Kauffmann et al. (2003) distinguishes between the two
different sources of emission line flux in this large sample
with the AGN occupying the parameter space to the
right of the distribution of star-forming galaxies.
the DEIMOS sample only three galaxies of the fourteen
have emission line flux ratios above this curve, implying
that some portion of the flux is due to an AGN, though
it is still within the errors that the emission line flux for
two of these objects results entirely from star-forming
regions inside the host galaxies.
AGN if log [OIII]/Hβ > 1 or log [NII]/Hα > 0.3.
Fluxes,
measured
[NII]λ6584.
line
(1981) used to dis-
fluxesof
andFig-
diagnosticof
The curve of
In
Objects can also be
Another 13 objects also have only detections of Hβ and
[OIII]λ5007. Of the thirteen, one has a flux ratio of log
[OIII]/Hβ > 1, indicative of an AGN. Sullivan et al.
(2000) find a similar ratio of galaxies classified as
star-forming/AGN from examination of emission lines
in their sample of UV selected galaxies.
4. MAGNITUDE AND COLOR DISTRIBUTION
The DEIMOS sample is ∼ 3 magnitudes deeper in r
than the SDSS spectroscopic sample. Figure 5 shows
a histogram of the SDSS r magnitudes for the sample
plotted with the SDSS r magnitude for GALEX MIS
sources in IR1.1 with matches in the SDSS DR2 spec-
troscopic sample.The IR1.1/DR2 sample consists of
∼ 51,000 galaxies at redshifts 0.005 < z < 0.25 with de-
rived star-formation histories (SFHs) by SED fitting (see
§5). The IR1.1/DR2 histogram is normalized to the size
of the DEIMOS galaxy sample. The IR1.1/DR2 sample
is mostly bounded in r by the SDSS spectroscopic survey
limits, r < 17.7 (Strauss et al. 2002).
Using the redshifts determined for the DEIMOS sam-
ple we calculate the absolute r magnitude, Mr, of galax-
ies in the sample, k-corrected to the mean redshift of the
IR1.1/DR2 sample, z = 0.1. K-corrections for all band-
passes were calculated using the publicly available code
of Blanton et al. (2003) version 4 1 4. In Figure 6 we
construct a color magnitude diagram (CMD) of NUV −r
versus Mr. We overplot our sample onto the IR1.1/DR2
sample, plotted as a shaded contour plot. In the figure
the diamond symbols correspond to DEIMOS objects
with spectroscopically measured redshifts at z < 0.25,
and the crosses correspond to objects with spectroscop-
ically measured redshifts at z > 0.25.
enclose 40% and 80% of the objects in the IR1.1/DR2
sample.
The CMD of the IR1.1/DR2 sample clearly shows the
bimodality of galaxies seen by GALEX in the nearby
universe, z < 0.25, with distribution peaks at blue
NUV −r colors of ∼ 3 and at red NUV −r colors of ∼ 6
(Wyder et al. 2005). We henceforth call these the blue
and red sequences. The galaxies in the blue sequence
are generally late-type in morphology and have spec-
tra with emission line-ratios indicative of star-formation
(Salim et al. 2005; Brinchmann et al. 2004), while the
galaxies in the red sequence typically have absorption
line spectra and lower star-formation rates than blue
sequence galaxies. The majority of the objects in the
DEIMOS sample lie in the region of the diagram occu-
pied by blue star-forming galaxies in the IR1.1/DR2 sam-
ple, mostly along the blue edge of the distribution, with
only one having a NUV − r color greater than 4. The
spectrum for this object, at z = 0.077, is shown in Fig-
ure 1. The spectrum shows absorption features: NaD,
and MgIb, but also shows weak emission features of Hα,
[NII]6584, and [SII]6717,6731.
In the bottom-left panel of Figure 7 we again plot the
color magnitude diagram for the sample, highlighting the
objects with FUV detections (filled circles). The other
3 panels of this figure show the UV color distribution
for the 24 galaxies in the DEIMOS sample with FUV
detections. Plotted for reference are galaxies from the
IR1.1/DR2 sample with FUV detections shown as the
shaded countour plots or shaded histogram in the respec-
tive panels. Besides the 3 galaxies in the DEIMOS sam-
The contours

Page 4

4
ple with FUV −NUV < 0 the UV color for this sample
shows a similar distribution with the IR1.1/DR2 sam-
ple. In the FUV CMD the objects with Mr< 20 show
a range of UV color not seen in the local blue sequence
galaxies of similar optical luminosity. Both figures 6 and
7 show that this sample probes the type of galaxies that
one obtains when looking at objects detected by GALEX
several magnitudes deeper than the SDSS spectroscopic
limit.
5. DERIVED GALAXY PARAMETERS
We derive the following galaxy parameters according
to the approach of Salim et al. (2005): the V-band dust
attenuations, AV in magnitudes, stellar metallicity Z, the
current star formation rate, SFR, averaged over the past
100 Myr in M⊙yr−1, the present-day stellar mass, M∗,
of the galaxy in M⊙, the fraction of stellar mass formed
in bursts over the last 100 Myr, Fburst, and the Scalo
(1986) b parameter, defined as the ratio of the current
SFR to the past time-averaged SFR (averaged over the
estimated age, not Hubble time). The galaxy parameters
are derived from model libraries of galaxies at redshifts
between 0.1 and 1.6 at redshift increments of 0.1 for the
DEIMOS sample, and at redshifts of 0.05,0.1,0.15,0.2,
and 0.25 for the IR1.1/DR2 sample. Each library con-
sists of up to ∼105models. Each model is parameterized
according to galaxy age, optical depth, star formation
history (SFH), and metallicity. The SFH of each model
is parameterized according to Kauffmann et al. (2003),
with an underlying, continuous, exponentially declining
SFR upon which bursts of star formation, random in time
and amplitude, are superimposed. Dust attenuation in
each model is parameterized using the prescription of
Charlot & Fall (2000) using an effective V -band optical
depth τV and absorption curve, τ ∝ λ0.7resulting from
both giant molecular clouds and the diffuse ISM, with
the fraction µ of τV contributed only by the diffuse ISM.
The V -band optical depth from giant molecular clouds
is taken to only affect stars younger than 10 Myr. A
description of the prior distributions of the model pa-
rameters is discussed in Salim et al. (2005).
Spectral energy distributions (SEDs) are created for
each galaxy in the library using the population synthe-
sis code of Bruzual & Charlot (2003). The model SEDs
are convolved with the GALEX and SDSS filter response
curves. Statistical estimates of physical galaxy param-
eters are derived by comparing the observed 7 band
GALEX/SDSS fluxes of each galaxy to all the convolved
model SEDs in the nearest redshift library. Probabil-
ity density functions (PDFs) for each physical parame-
ter are created by assigning weights to the parameters.
The χ2goodness of fit of the model determines the weight
(∝ exp[−χ2/2]) that is assigned to the parameters of that
model. The median (or most typical parameter value) of
the PDF is taken as the estimate of the galaxy parame-
ter.
We perform SED fits for galaxies not identified as
QSOs. In Table 2 we list the galaxies, their redshifts,
NUV − r non k-corrected colors, and their derived pa-
rameters. The parameters derived from the SED fits are
not used unless the reduced χ2fit of at least one model
is below 10. Only two of the 46 galaxies do not meet
this criterion. Figures 8 and 9 show the derived param-
eters for the DEIMOS sample of galaxies overplotted on
the IR1.1/DR2 sample plotted as shaded contour plots.
The contours are labeled and encompass 52%, 84%, and
97% of the data. The diamonds as before are objects
with z < 0.25, and the crosses are objects with z > 0.25.
The objects span the range of derived parameters of the
IR1.1/DR2 sample for Z, and AV.
The derived metallicities, Z = [Fe/H], for the sam-
ple are not very well constrained. The typical error, 1/4
of the difference between the 97.5 percentile and the 2.5
percentile of the PDF, of the derived metallicities are
±43% Z⊙. The derived M∗range from 108to 1011M⊙
with an average error of ∼ ±0.2dex, but the majority
of the M∗ are at or below 1010M⊙, about an order of
magnitude below the mean of the IR1.1/DR2 sample.
The derived SFRs lie between 10−1and 102M⊙yr−1,
with five galaxies having derived SFRs greater than 10
M⊙yr−1. The average error on the derived SFRs is typ-
ically ±0.3dex.Comparing the SFRs between galax-
ies with similar NUV − r color or M∗, the galaxies at
z > 0.25 have SFRs about an order of magnitude greater
than the galaxies at z < 0.25. This is mostly a selec-
tion effect since the more distant galaxies are also more
luminous on average. A comparison of these SED de-
rived SFRs with Hα derived SFRs cannot be performed
since the spectra are not flux calibrated and many lack
Hα. The range of SFRs we find for this sample is simi-
lar to the range of SFRs for local UV galaxies found by
Sullivan et al. (2000) derived both from Hα luminosity
and from the UV luminosity, but is an order of magnitude
larger than the SFRs found by Kobulnicky & Kewley
(2004) in their sample of Goods-North Treasury Keck
Redshift Survey galaxies at 0.4 < z < .9 when compared
to galaxies in the DEIMOS sample with similar redshifts.
We find that the galaxies are largely starbursts with
logb ∼ −0.1.This is shown in Figure 10.
tribution of logb for objects in the DEIMOS sample
with SED fits are plotted with the distribution for the
IR1.1/DR2 sample normalized to the number of objects
in the DEIMOS sample with SED fits. The DEIMOS
sample shows a highly peaked value of logb compared to
the IR1.1/DR2 sample which has two small peaks near
logb ∼ −3 and ∼ −1, showing that most of these galax-
ies currently have less star-formation now than in the
past.The value logb ∼ −0.1 that describes most of
the galaxies in the DEIMOS sample indicates that while
these galaxies have less star formation than in the past,
they are currently or have recently gone through a burst
of star-formation. Two of the galaxies have logb > 0
indicating that these galaxies are going through major
starbursts in their histories. From a statistical stand-
point, the SED fits reveal with a 95% reliability that at
least three galaxies have not had a burst of starformation
in the last 100 Myr. Half of the galaxies in the sample
could have formed as much as ∼ 10% of their stellar mass
in a burst within the last 100 Myr, 10 could have formed
up to ∼ 25% and five of the less massive systems could
have had up to ∼ 50% of their stellar mass form in bursts
within the last 100Myr.
The dis-
6. DISCUSSION
In the plot of NUV − r vs Mr our sample lies along
a “blue sequence” of star forming galaxies and they
have among the bluest colors for star forming galax-
ies in the local Universe found by GALEX and SDSS.

Page 5

5
While all the galaxies show blue NUV − r color this
is not a homogenous sample.
ies show disk structure in the SDSS images. The re-
maining two-thirds of the sample fit into the heteroge-
neous class of faint blue galaxies at intermediate redshifts
previously studied by Koo et al (1995); Guzm´ an et al.
(1996, 1998); Phillips et al. (1997) and Guzm´ an et al.
(1997) in the Hubble Deep Field and adjoining fields,
and by Hammer et al. (2000) and Mallen-Ornelas et al.
(1999) in the Canada-France Redshift Survey.
The faint blue subset of the DEIMOS sample has
similar luminosities, SFRs and optical colors to the
compact galaxies found in the Hubble Deep Field by
Phillips et al. (1997), though none are “compact” (op-
tical half-light radii < 0′′.5.) This is a selection effect
in our sample. Galaxies with half-light radii this small
are classified as stars in SDSS and would not have been
selected for spectroscopy in our sample. In Figure 11 we
plot, as filled diamonds, the rest frame absolute B magni-
tude versus the B magnitude surface brightness and SFR
versus velocity dispersion. The absolute B magnitudes
are calculated from the observed bandpasses using the k-
correction code of Blanton et al. (2003), and the surface
brightness are calculated with the half-light SDSS r pet-
rosian radii. The velocity dispersions are derived from
the measured linewidths of [OIII]5007 for objects with
≥ 3σ detections of the emission line. For comparison,
we also plot, as stars and squares, the faint blue galaxy
samples of Koo et al (1995); Guzm´ an et al. (1996) and
Phillips et al. (1997) making the necessary corrections
to our adopted cosmology. While our sample of galaxies
have luminosities, SFRs, and velocity dispersions compa-
rable with the previous samples, the size selection effect
separates our sample from the previous samples in sur-
face brightness.
Koo et al (1995) and Guzm´ an et al. (1998) propose
that the subset of the faint blue population of galaxies
with compact geometry (half light radii < 0′′.5) and nar-
row emission lines (σ < 65kms−1) will fade to become
dwarf spheroidals by z=0, while Hammer et al. (2000)
claims that the most luminous of these are too massive to
become dwarf spheroidals, and instead will become the
bulges of spiral galaxies. The most massive intermediate
redshift galaxies (M∗? 1010) in our sample could likely
follow this latter evolution path. Indeed, we see that the
galaxies in our sample that show extended structure at
low redshifts (z < 0.3) are among the the most massive
of the sample.
Theconclusionsof
Guzm´ an et al. (1998) rest partially on the assumption
that galactic winds from the last starburst event will
remove the remaining gas from these systems and halt
star formation causing them to fade several magnitudes
by z=0. Through modeling the UV and optical broad-
band colors we find that about half the galaxies in our
present sample could have formed at most ∼ 10% of
their stellar mass in bursts within the past 100 Myr,
and only fifteen (all having M∗ < 109) could have
formed over 20% of their stellar mass within the last 100
Myr. If the last starburst event removes all remaining
gas from these systems, as Koo et al (1995) propose,
then why did the previous starforming events in these
galaxies that produced the majority of the stars in
these systems not halt star formation?
Fifteen of the galax-
Koo et al(1995) and
Our derived
starformation histories of these galaxies argues that the
current star formation events will not halt starfromation
in every galaxy. But what percentage of these will
continue to experience further bursts of star formation
is unknown. However, even in the low mass systems of
the present sample, residual star formation or another
burst of star formation is ultimately unlikely to change
the evolutionary outcome proposed by Koo et al (1995)
and Guzm´ an et al. (1998) for the most compact low
mass galaxies in our sample.
more extended sources of our sample are more likely to
become present-day dwarf irregulars rather than dwarf
spheroidals.
We suggest that the
7. SUMMARY
We have presented DEIMOS spectra for objects de-
tected by GALEX in the MIS survey with imaged coun-
terparts in SDSS; a total exposure time of 30 min per
slitmask was used. GALEX has proven to be a sensi-
tive instrument for wide field galaxy surveys. We have
shown that the GALEX Medim Imaging Survey followed
up with a 30 m integration with Keck/DEIMOS yields
redshifts and line measurements for star forming galax-
ies to z∼ 0.7, and has yielded 4 QSO’s 3 of which are
previously un cataloged. The matched sample reaches
approximately 3 magnitudes fainter in r than the SDSS
spectroscopic survey limits. The sample is not a homoge-
nous sample, but is indicative of the types of galaxies
forming stars out to z∼ 0.7. We have derived physical
parameters for these galaxies from the SEDS and com-
pared this sample to a sample of 50,000 SDSS galaxies
with GALEX detections.
1. We find that roughly one-third of the galaxies are
starforming late type disk galaxies, four are QSOs at
z > 1, and the remaining galaxies are faint blue low
mass starbursts.
2.
ies show emission line ratios indicative of an AGN. A
similar fraction was found by Sullivan et al. (2000) in
their UV selected sample.
Approximately 3 out of 14 star formimg galax-
3.
than what is found locally in the SDSS spectroscopic
sample. The range of M∗for the DEIMOS sample spans
from 108to 1011M⊙, whereas the median of the SDSS
spectroscopic sample is ∼ 5 × 1010.
The masses of the galaxies are typically lower
4. The SFRs of the galaxies at z > 0.25 are roughly
an order of magnitude greater than the SFRs for the
galaxies at z < 0.25.
5.
in NUV -r color, the remaining galaxies show evidence
of a starburst in the last 100 Myr, with logb ∼ −0.1.
Fifteen of the galaxies in the lower mass range of the
sample could have formed more than 20% of their stellar
mass in bursts of star formation within the last 100 Myr.
Besides three of the most massive, and reddest
6.
SFRs, and B luminosities to previous samples of faint
blue galaxies, though the galaxies in our present study
are 2 mag fainter in surface brightess.
Our sample has similar velocity dispersions,

Page 6

6
GALEX is a NASA Small Explorer, launched in April
2003. We gratefully acknowledge NASA’s support for
construction, operation, and science analysis for the
GALEX mission, developed in cooperation with the Cen-
tre National d’Etudes Spatiales of France and the Korean
Ministry of Science and Technology.
The authors wish to recognize and acknowledge the
very significant cultural role and reverence that the sum-
mit of Mauna Kea has always had within the indigenous
Hawaiian community. We are most fortunate to have the
opportunity to conduct observations from this mountain.
The analysis pipeline used to reduce the DEIMOS data
was developed at UC Berkeley with support from NSF
grant AST-0071048.
The authors thank an anonymous referee for extremely
helpful comments.
Facilities: GALEX, KECK:II(DEIMOS)
REFERENCES
Abazajian, K. et al. 2004, AJ, 128, 502
Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5
Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393
Blanton, M. et al. 2003, AJ, 125, 2276
Brinchmann, J., Charlot, S., White, S. D. M., Tremonti, C.,
Kauffmann, G., Heckman, T., & Brinkmann, J. 2004, MNRAS,
351, 1151
Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000
Charlot, S. & Longhetti 2000, MNRAS, 323 887
Charlot, S. & Fall, S. M. 2000, ApJ 539, 718
Davis M. et al. in Discoveries and Research Prospects from 6- to
10-Meter-Class Telescopes II. Edited by Guhathakurta, Puragra.
Proceedings of the SPIE, Volume 4834, pp. 161-172 (2003)
Ellis, R., S. 1007, ARAA, 35, 389
Faber, S. M., et al. 2003, Proc. SPIE, 4841, 1657
Fukugita, M., Ichikawa, T., Gunn, J. E., Doi, M., Shimasaku, K.,
& Schneider, D. P. 1996, AJ, 111, 1748
Guzm´ an, R., Koo, D. C., Faber, S. M., Illingworth, G. D.,
Takamiya, M., Kron, R. G., & Bershady, M. A. 1996, ApJ, 460,
5
Guzm´ an et al. 1997, Apj, 489, 559
Guzm´ an, R., Jangren, A., Koo, D. C., Bershady, M. A., & Simard,
L. 1998 ApJ 495, 13
Hammer, F., Gruel, N., Thuan, T. X., Flores, H., & Infante, L.
2000, ApJ, 550, 570
Hogg, D. W., Finkbeiner, D. P., Schlegel, D. J., and Gunn, J. E.
2001, AJ, 122, 2129
Kauffmann, F. et al. 2003, MNRAS, 341,33
Kennicutt, R. C., ARA&A, 36, 189
Kennicutt R. C., Bresolin, F. & Garnett, D. R. 2003, ApJ, 591,
801
Kewley, L. J., Dopita, M. A., Sutherland, R. S., Heisler, C. A., &
Trevena, J. 2001, ApJ, 556,121
Kobulnicky, H. A. & Kewley, L. 2004, ApJ, 617, 240
Koo, D. C., Guzm´ an, R., Faber, S. M., Illingworth, G. D.,
Bershady, M. A., Kron, R. G., & Takamiya, M. 1995, ApJ, 440,
49
Martin, C. D. et al. 2005, ApJ, 619, 1
Martin, C. D. et al. 2005, ApJ, 619, 59
Milliard, B., Donas, J., Laget, M., Armand, C., Vuillemin, A. 1992
A&A, 257, 24
Morrissey, P. et al. 2005, ApJ 619, 7
Mallen-Ornelas, G., Lilly, S. J., Crampton, D., & Schade, D. 1999,
ApJ, 518, 83
Osterbrock, D. E. 1989, Astrophysics of Gaseous Nebulae and
Active Edmunds, M. G., Blackwell, D. E., Chun, M. S., & Smith,
G. 1979, MNRAS, 189, 95
Phillips, A. C. et al. 1997, ApJ, 489, 543
Pier, J. R., Munn, J. A., Hindsley, R. B., Hennessy, G. S., Kent,
S. M., Lupton, R. H., and Ivezic, Z. 2003, 125, 1559
Salim, S. et al. 2005, ApJ, 619, 39
Salim, S. et al 2007, arXiv:0704.3611
Scalo, J. M. 1986. Fundam. Cosm. Phys. 11:1-278
Schiminovich, D. et al. 2005, ApJ, 619,47
D.J. Schlegel, D.P. Finkbeiner,& M. Davis, ApJ, 500, 525 (20 June
1998)
Smith, J. A. et al. 2002, AJ, 123, 2121
Stoughton, C. et al. 2002, AJ, 123, 485
Strauss, M. A. et al. 2002, AJ, 124, 181
Sullivan, M., Treyer, M. A., Ellis, S. R., Bridges, T. J., Milliard,
B., Donas, J. 2000, MNRAS, 312, 442
Tremonti, C. A. et al. 2004, ApJ, 613, 898
Treyer, M. A. et al. 2005, ApJ, 619, 15
Wyder, T. K et al. 2005, ApJ, 619, 15
Wyder, T. K. 2006, in preparation

Page 7

7
Fig. 1.— Spectra of several galaxies from our sample at a range of different redshifts. All of the galaxies are blue sequence galaxies
except the galaxy at z = 0.77, which lies on the red sequence. The spectra have been boxcar smoothed by 5 pixels.

Page 8

8
Fig. 2.— Spectra of four QSOs from our sample. The spectra have been smoothed by 5 pixels.

Page 9

9
Fig. 3.— Redshift distribution of spectroscopic sample. The mean redshift for the sample is z = 0.421. Shaded boxes indicate the quasar
redshifts.

Page 10

10
Fig. 4.— Emission line diagnostic diagram first used by Baldwin, Phillips, & Terlevich
a to the GALEX IR1.1/SDSS DR2 spectroscopic sample. The dashed curves taken from Kewley et al. (2001); Kauffmann et al. (2003)
show the distinction between sources with emission line flux coming from AGN (above) and sources with emission line flux coming from
HII regions (below). Out of the 14 objects from our DEIMOS sample with detections of Hβ, [OIII], Hα, and [NII], only three have emission
line ratios consistent with emission due to an AGN.
(1981). The shaded 2D-histogram corresponds

Page 11

11
Fig. 5.— SDSS r magnitude distribution of the DEIMOS spectroscopic sample plotted with the SDSS r magnitude distribution of
the matched GALEX IR1.1/SDSS DR2 spectroscopic sample (shaded histogram). The DEIMOS matched spectroscopic sample mean r
magnitude ( 21) is 3 magnitudes fainter than the mean of the SDSS spectroscopic sample.

Page 12

12
Fig. 6.— CMD of Mr vs. NUV-r plotted for the sample of GALEX/SDSS objects with DEIMOS spectroscopy plotted with the entire
GALEX(IR1.1)/SDSS (DR2) matched sample having SDSS spectroscopy. The diamonds represent galaxies in the DEIMOS sample at
z < 0.25; crosses represent galaxies in the DEIMOS sample at z > 0.25. The IR1.1/DR2 sample is plotted as the shaded contour plot; the
darker regions correspond to a higher density of points and the contours encompass 40% and 80% of the of the objects in the sample.

Page 13

13
Fig. 7.— Panel 1: CMD of Mr vs. FUV − NUV plotted for the 24 galaxies in the DEIMOS sample with FUV detections, plotted with
objects in the IR1.1/DR2 matched sample with FUV detections shown as a shaded contour plot. PANEL 2: histogram of FUV − NUV
color for the 24 DEIMOS FUV galaxies plotted with distribution of the IR1.1/DR2 FUV sample (shaded histogram). PANEL 3: same
as figure 5. The filled (unfilled) circles correspond to objects with (without) FUV detections. Panel 4: Color-Color diagram of the FUV
DEIMOS galaxies again overplotted onto the FUV IR1.1/DEIMOS sample. The symbols used are the same as in figure 6 and the contours
encompass 40% and 80% of the of the objects in the IR1.1/DR2 sample.

Page 14

14
Fig.8.— Derived Galaxy parameters, metallicity and V-band attenuation plotted versus M∗ and NUV − r color. The DEIMOS
spectroscopic sample is plotted over the matched GALEX IR1.1/SDSS DR2 spectroscopic sample (shaded contour plot). The crosses
correspond to objects with s redshifts z > 0.25. The diamonds correspond to objects with redshifts z < 0.25. The contours for the plots of
stellar mass contain 57%, 84%, and 97% of the data. The contours of NUV − r color enclose 40% and 80% of the sample.

Page 15

15
Fig. 9.— Derived Galaxy parameters, log SFR and log b plotted versus M∗ and NUV − r color. The DEIMOS spectroscopic sample is
plotted over the matched GALEX IR1.1/SDSS DR2 spectroscopic sample (shaded contour plot). The crosses correspond to objects with
redshifts z > 0.25. The diamonds correspond to objects with redshifts z < 0.25. The contours for the plots of M∗ contain 56%, 84%, and
97% of the data. The contours of NUV − r color enclose 40% and 80% of the data. The SFRs for the high z galaxies are approximately
and order of magnitude higher than the SFRS of the low z galaxies.