Spitzer Survey of the Large Magellanic Cloud, Surveying the Agents of a Galaxy's Evolution (SAGE) I: Overview and Initial Results

TL;DR: In this paper, the agents of a galaxy's evolution (SAGE), the interstellar medium (ISM) and stars in the Large Magellanic Cloud (LMC) were surveyed using the Spitzer Space Telescope.

Abstract: We are performing a uniform and unbiased, ~7x7 degrees imaging survey of the Large Magellanic Cloud (LMC), using the IRAC and MIPS instruments on board the Spitzer Space Telescope in order to survey the agents of a galaxy's evolution (SAGE), the interstellar medium (ISM) and stars in the LMC. The detection of diffuse ISM with column densities >1.2x10^21 H cm^-2 permits detailed studies of dust processes in the ISM. SAGE's point source sensitivity enables a complete census of newly formed stars with masses >3 solar masses that will determine the current star formation rate in the LMC. SAGE's detection of evolved stars with mass loss rates >1x10^-8 solar masses per year will quantify the rate at which evolved stars inject mass into the ISM of the LMC. The observing strategy includes two epochs in 2005, separated by three months, that both mitigate instrumental artifacts and constrain source variability. The SAGE data are non-proprietary. The data processing includes IRAC and MIPS pipelines and a database for mining the point source catalogs, which will be released to the community in support of Spitzer proposal cycles 4 and 5. We present initial results on the epoch 1 data with a special focus on the N79 and N83 region. The SAGE epoch 1 point source catalog has ~4 million sources. The point source counts are highest for the IRAC 3.6 microns band and decrease dramatically towards longer wavelengths consistent with the fact that stars dominate the point source catalogs and that the dusty objects, e.g. young stellar objects and dusty evolved stars that detected at the longer wavelengths, are rare in comparison. We outline a strategy for identifying foreground MW stars, that may comprise as much as 18% of the source list, and background galaxies, that may comprise ~12% of the source list.

Summary

1. INTRODUCTION

• The interstellar medium (ISM) plays a central role in the evolution of galaxies as the birth site of new stars and the repository of old stellar ejecta.
• To understand this recycling, the authors have to study the physical processes of the ISM, the formation of new stars, the injection of mass by evolved stars, and their interrelationship on a galaxy-wide scale.
• Both of these previous far-IR surveys revealed the most luminous dusty inhabitants of the LMC—supergiants, AGB stars, H ii regions, and planetary nebulae—but lacked the angular resolution and corresponding point-source sensitivity to detect the more populous, less luminous sources.
• Section 2 describes SAGE, including the science drivers and observing strategy.

2. SAGE OBSERVING PROGRAM

• In order to obtain an unbiased and complete census of star formation throughout the LMC, SAGE is required to be sensitive to all star formation activity from the massive star formation traced by H ii regions to the lower mass star formation traced by Taurus-like complexes (Fig. 2).
• The analysis of the IRAS data on the LMC indicates a lower 12 mdiffuse emission in comparison to the MWand suggests a deficit of very small dust grains, possibly due to the intense UV radiation of the LMC (Schwering 1989).
• Each position in the SAGE IRAC survey has at least four frames of coverage, resulting in an exposure time per pixel of at least 43.2 s in all IRACbands for the complete survey and a quarter of that, 11 s, for the single-frame photometry of each epoch.
• The 3month time baseline between epochs is ideal for the removal of the systematic artifacts, because it provides a 90 roll angle in the orientation of the detectors, which optimally removes the ‘‘striping’’ artifacts in MIPS and IRAC image data.

3. DATA-PROCESSING APPROACH

• The full LMC mosaics of the IRAC and MIPS (Figs. 3 and 4) data show the coverage of the SAGE survey with these two instruments over the two epochs.
• The removal of this background from the photometry measurements is handled with the same iterative approach as the IRAC photometry, albeit implemented with different programs.
• The SAGE IRAC andMIPS point-source catalogs are ingested into the database system.

4. PRELIMINARY EPOCH 1 RESULTS FROM A REGION NEAR N79 AND N83

• The 4.5 m diffuse emission is a combination of Brackett , bound-free continuum, and possibly very small dust grain emission.
• On the ½3:6 ½8:0 versus ½8:0 ½24 color-color diagram the authors plot these MW templates in a simplified manner by grouping them into three broad categories: stars without dust (Fig. 10, asterisks), dusty evolved stars (Fig. 10, triangles), and YSOs (Fig. 10, squares).
• The 1175 point sources detected at 3.6, 8.0, and 24 m in the N79/N83 region are plotted on this color-color diagram and classified into one of the three categories as follows; using theMW templates as a guideline, MEIXNER ET AL.2282 Vol.
• The percentage of such background galaxy candidates in the southwest quarter of the N79/N83 region is 14% of the total (=219/1576), scaling the number source density to the entire N79/N83 region suggests that 876 of the 7595 sources detected at 8 m are background galaxies, or 12% of the total sample.

5. SUMMARY

• The SAGE data are nonproprietary, and the SAGE team is committed to delivering point-source lists and improved images to the SSC for community access in support of proposal cycles 4 and 5 of Spitzer.
• The authors present initial results on the epoch 1 SAGE data for a region near N79 and N83 that provide a verification of the survey’s goals and a start at interpreting the results.
• Using MW templates as a guide, the authors adopt a simplified point-source classification to identify three candidate groups—stars without dust, dusty evolved stars, andYSOs—on the ½3:6 ½8:0 versus ½8:0 ½24 color-color diagram.

Figures

TABLE 2 Epoch 1: Initial Results for Point-Source Sensitivities and Source Counts

Fig. 6.—Plots demonstrating the quality of the SAGE flux and magnitude measurements in the epoch 1 IRAC point-source catalogs. For each IRAC band, noted in the top left of the plots, the measured SAGE magnitudes are plotted against the predicted magnitudes for the 139 calibration stars in the SAGE field. The size of the crosses represents the error bars on both axes.

Fig. 3a

Fig. 5.—Locations of the SAGE absolute photometric standards calibration network listed in Table 3. The trapezoid box is the approximate location of the SAGE survey area adequate to cover SAGE’s IRAC and MIPS observations. The two dotted circles represent stars within roughly 1 and 2 of the south ecliptic pole. Any such stars will be of gratuitous value to other missions observing the ecliptic poles, such as the Wide-Field Infrared Survey Explorer and the James Webb Space Telescope. Both of these are subsets of the short-dashed box, which encompasses the entire field of view, that indicates the area searched for appropriate calibration stars using SIMBAD. Symbols indicate the quality of each stellar calibrator, with the best being the filled circles, followed by filled squares and then filled triangles. These show consistency between optical and near-IR (2MASS) photometry and the spectral type and extinction to within 0.010, 0.030, and 0.060 mag, respectively. Open squares, open triangles, and open stars represent lower ranked calibrators with consistency at the levels of 0.090, 0.12, and 0.15 mag, respectively.

TABLE 3 SAGE Calibration Stars for the IRAC Photometry

Fig. 9.—MIPS epoch 1 SAGE data of the N79/N83 region in comparison with ISM gas tracers. Top left: MIPS 24 m with a square-root gray scale ranging from 0 to 20 MJy sr 1. Middle left: MIPS 70 m with a square-root gray scale ranging from 0 to 100 MJy sr 1. Bottom left: MIPS 160 m with a square-root gray scale ranging from 0 to 300 MJy sr 1. Top right: H SHASSA data in a log scale (Gaustad et al. 2001). Middle right: H i 21 cm line data in a square-root gray scale (Staveley-Smith et al. 2003; Kim et al. 2003). Bottom right: CO J ¼ 1 0 line data in a square-root scale where the contour marks the 3 level (0.7 K km s 1) and the data are absent from the lower right (southwest) part of the image (Fukui et al. 2001; Y. Fukui et al. 2006, in preparation).

TABLE 1 Principal Characteristics of the SAGE Survey, Spitzer Program ID 20203

Fig. 1.—Examples of the many existing LMC surveys: H i (Staveley-Smith et al. 2003; Kim et al. 2003), CO (Fukui et al. 2001; Y. Fukui et al. 2006, in preparation), IRAS 100 m, H (Gaustad et al. 2001) with the SAGEMIPS coverage overlaid, UV (Smith et al. 1987), and stellar density (Zaritsky et al. 2004) with the SAGE IRAC coverage overlaid. The SAGE survey maps the LMC twice. Epoch 1 coverage is outlined in green, and epoch 2 coverage is outlined in black. The location of the N79/N83 region, discussed in x 4, is outlined by the square box in the IRAS 100 m image.

Fig. 13.—The [8.0] vs. ½3:6 ½8 CMD for all sources detected in the N79/N83 region at these two wavelengths. Overplotted with cyan circles is the subset of sources in the southwest quarter of the N79/N83 region that is outside of the H ii regions.

Fig. 8.—IRAC epoch 1 SAGE data of the region surrounding N79/N83. Top left: IRAC 3.6 m. Top right: IRAC 4.5 m. Bottom left: IRAC 5.8 m. Bottom right: IRAC 8.0 m.

Fig. 3a

Fig. 2.—Spitzer CMDs showing the discovery space for SAGE. Square brackets denote the brightness in Vega magnitudes at the wavelength enclosed in the brackets. For example, [8.0] means the Vega magnitude at 8.0 m. The symbols identify populations of key sources throughout the LMC: YSOs (1–30 M ), H ii regions, Taurus-like clusters, O-rich and C-rich AGB stars, RSGs, and main-sequence O stars. Symbols, defined in the figure, represent the template/model photometry of Cohen (1993) and Whitney et al. (2004). SAGE’s sensitivity limit (solid line) falls 1000 times below theMSX limit (dashed line) and the lower limit to AGB mass loss, >10 8 M yr 1 (dotted line). The circle with a star represents a subregion of Taurus containing 12 stars, placed at the distance of the LMC. [See the electronic edition of the Journal for a color version of this figure.]

Fig. 14.—Near-IR (2MASS) and IRAC CMDs for sources detected in all seven bands in the N79/N83 region (top left). The southwest region (top right) contains sources detected in the N79/N83 southwest quarter and has been corrected for background galaxies based on the cuts of Fig. 11; see text. The [3.6] vs. J ½3:6 diagram (bottom left) is effective in discriminating between foregroundMWobjects and LMC giants and AGB stars that should generally lie below and to the right of the dashed line (see text). The ‘‘corrected southwest’’ diagram (bottom right) shows the remaining sources, primarily LMC AGB stars, after removing foreground MW stars and background galaxy contributions. [See the electronic edition of the Journal for a color version of this figure.]

Fig. 11.—Color-color plots. Left: IRAC-only plot. Right: IRAC and 2MASS plot, showing how the source classification holds at other wavelengths for point sources detected in the SAGEN79/N83 region. The black circles are plotted first and include all the sources detected in the bands plotted. The colored circles are overlaid on the black circles and include the same circles as plotted in the source classification template (Fig. 10).

Fig. 12.—Four CMDs of SAGE data showing in black all the point sources from the N79/N83 region that have been detected at the wavelengths plotted in the CMD. The candidate classified sources from the source classification are overlaid in the same color used in Fig. 10. The black circles are sources detected in the bands of the CMD but not in all of the bands of the source classification diagram and hence tend to be fainter sources not detected at the longer wavelengths, e.g., 24 m. The dashed line at ½8:0 ¼ 11 mag marks the mass-loss rate of 10 8 M yr 1, above which the AGB stars appear to have dusty winds.

Fig. 7.—Three-color SAGE images of the N79/N83 region near the southwest end of the LMC bar. Top: IRAC three-color image, with IRAC-1 3.6 m in blue, IRAC-3 4.5 m in green, and IRAC-4 8 m in red.Middle: IRAC 3.6 m in blue, IRAC 8 m in green, and MIPS 24 m in red. Bottom: MIPS three-color image, with 160 m in red, 70 m in green, and 24 m in blue. In all panels white is a combination of all three colors.

Fig. 4.—Full LMC mosaic of the SAGE data shown in three colors: epoch 1 IRAC-1 3.6 m in blue, IRAC-4 8 m in green, and MIPS 24 m in red.

Fig. 10.—The ½3:6 ½8:0 vs. ½8:0 ½24 color-color diagram, showing the broadest separation of sources while still retaining enough sources to derive a source classification. For all bands, square brackets mean the Vega magnitude at that wavelength, e.g., [8.0] is the Vega magnitude at 8.0 m. Black symbols represent MW templates from Cohen et al. (1993) andWhitney et al. (2004). Asterisks represent stars without dust, such as main-sequence stars or red giants. Triangles represent dusty evolved stars. Squares represent YSOs. The black dashed line marks the boundary for the source classification based on the location of the MW templates. The orange dashed line shows the color temperature for blackbodies with temperatures ranging from 10,000 to 400 K, marked as gray crosses on the line. The 1175 colored circles are point sources detected in IRAC 3.6 and 8.0 m andMIPS 24 mbands from theN79/N83 region. The red, green, and blue circles are candidate YSOs, dusty evolved stars, and stars without dust, respectively. These color-classified sources are plotted in the color-color diagrams and CMDs of Figs. 11 and 12, respectively, to provide guidance on the types of sources plotted.

Full text

SPITZER SURVEY OF THE LARGE MAGELLANIC CLOUD: SURVEYING THE AGENTS
OF A GALAXY’S EVOLUTION (SAGE). I. OVERVIEW AND INITIAL RESULTS
Margaret Meixner,
1
Karl D. Gordon,
2
Remy Indebetouw,
3
Joseph L. Hora,
4
Barbara Whitney,
5
Robert Blum,
6
William Reach,
7
Jean-Philippe Bernard,
8
Marilyn Meade,
9
Brian Babler,
9
Charles W. Engelbracht,
2
Bi-Qing For,
2
Karl Misselt,
2
Uma Vijh,
1
Claus Leitherer,
1
Martin Cohen,
10
Ed B. Churchwell,
10
Francois Boulanger,
11
Jay A. Frogel,
12
Yasuo Fukui,
13
Jay Gallagher,
9
Varoujan Gorjian,
14
Jason Harris,
2
Douglas Kelly,
2
Akiko Kawamura,
13
SoYoung Kim,
15
William B. Latter,
7
Suzanne Madden,
16
Ciska Markwick-Kemper,
3
Akira Mizuno,
13
Norikazu Mizuno,
13
Jeremy Mould,
17
Antonella Nota,
1
M.S. Oey,
18
Knut Olsen,
2
Toshikazu Onishi,
13
Roberta Paladini,
7
Nino Panagia,
1
Pablo Perez-Gonzalez,
2
Hiroshi Shibai,
13
Shuji Sato,
13
Linda Smith,
1,19
Lister Staveley-Smith,
20
A. G. G. M. Tielens,
21
Toshiya Ueta,
22
Schuyler Van Dyk,
7
Kevin Volk,
23
Michael Werner,
24
and Dennis Zaritsky
2
Received 2006 February 10; accepted 2006 June 9
ABSTRACT
We are performing a uniform and unbiased imaging survey of the Large Magellanic Cloud (LMC; 7

; 7

) using
the IRAC (3.6, 4.5, 5.8, and 8 m) and MIPS (24, 70, and 160 m) instruments on board the Spitzer Space Telescope
in the Surveying the Agents of a Galaxy’s Evolution (SAGE) survey, these agents being the interstellar medium
(ISM) and stars in the LMC. This paper provides an overview of the SAGE Legacy project, including observing
strategy, data processing, and initial results. Three key science goals determined the coverag e and depth of the survey.
The detection of diffuse ISM with column densities >1:2 ; 10
21
Hcm
2
permits detailed studies of dust processes in
the ISM. SAGE’s point-source sensitivity enables a complete census of newly formed stars with masses >3 M

that
will determine the current star formation rate in the LMC. SAGE’s detection of evolved stars with mass-loss rates
>1 ; 10
8
M

yr
1
will quantify the rate at which evolved stars inject mass into the ISM of the LMC. The observing
strategy includes two epochs in 2005, separated by 3 months, that both mitigate instrumental artifacts and constrain source
variability. The SAGE data are nonproprietary. The data processing includes IRAC and MIPS pipelines and a database
for mining the point-source catalogs, which will be released to the community in support of Spitzer proposal cycles 4 and
5. We present initial results on the epoch 1 data for a region near N79 and N83. The MIPS 70 and 160 m images of the
diffuse dust emission of the N79/ N83 region reveal a similar distribution to the gas emissions, especially the H i 21 cm
emission. The measured point-source sensitivity for the epoch 1 data is consistent with expectations for the survey. The
point-source counts are highest for the IRAC 3.6 m band and decrease dramatically toward longer wavelengths,
A
1
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218; meixner@stsci.edu, leitherer@stsci.edu, vijh@stsci.edu, nota@stsci.edu,
panagia@stsci.edu.
2
Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721; kgordon@as.arizona.edu, cengelbracht@as.arizona.edu, biqing@
email.arizona.edu, kmisselt@as.arizona.edu, pgperez@as.arizona.edu, jharris@as.arizona.edu, dkelly@as.arizona.edu, dennis@fishingholes.as.arizona.edu.
3
Department of Astronomy, University of Virginia, P.O. Box 3818, Charlottesville, VA 22903-0818; remy@virginia.edu, f k2n@virginia.edu.
4
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, MS 65, Cambridge, MA 02138-1516; jhora@cfa.harvard.edu.
5
Space Science Institute, 4750 Walnut Street, Suite 205, Boulder, CO 80301; bwhitney@spacescience.org.
6
Cerro Tololo Inter-American Observatory, Casilla 603, La Serena, Chile; rblum@ctio.noao.edu, kolsen@ctio.noao.edu.
7
Spitzer Science Center, Jet Propulsion Laboratory, California Institute of Technology, MS 220-6, Pasadena, CA 91125; reach@ipac.caltech.edu, vandyk@ipac
.caltech.edu, paladini@ipac.caltech.edu.
8
Direction de la Recherche, Centre d’Etude Spatiale des Rayonnements, 18 Avenue Edouard Belin, Toulouse Cedex F-31055, France; jean-philippe.bernard@cesr.fr.
9
Department of Astronomy, University of Wisconsin, Madison, 475 North Charter Street, Madison, WI 53706-1582; meade@sal.wisc.edu, brian@sal.wisc.edu,
ebc@astro.wisc.edu, jsg@astro.wisc.edu.
10
Radio Astronomy Laboratory, University of California at Berkeley, 601 Campbell Hall, Berkeley, CA 94720-3411; mcohen@astro.berkeley.edu.
11
Institut d’Astrophysique de Paris, CNRS UPR 341, 98bis, Boulevard Arago, Paris F-75014, France; francois.boulanger@ias.u-psud.fr.
12
Association of Universities for Research in Astronomy, Inc., Suite 350, 1200 New York Avenue NW, Washington, DC 20005; jfrogel@aura-astronomy.org.
13
Department of Astrophysics, Nagoya University, Chikusa-Ku, Nagoya 464- 01, Japan; fukui@a.phys.nagoya-u.ac.jp, kawamura@a.phys.nagoya-u.ac.jp, mizuno@
a.phys.nagoya-u.ac.jp, norikazu@a.phys.nagoya, ohnishi@a.phys.nagoya-u.ac.jp, shibai@nagoya-u.jp, ssato@z.phys.nagoya-u.ac.jp.
14
Jet Propulsion Laboratory, 4800 Oak Grove Boulevard, MS 169-327, Pasadena, CA 91109; varoujan.gorjian@jpl.nasa.gov.
15
Department of Physics and Astronomy, Johns Hopkins University, Homewood Campus, Baltimore, MD 21218; sykim@pha.jhu.edu.
16
Service d’Astrophysique, Commissart a
`
L’Energie Atomique de Saclay, 91191 Gif-sur-Yvette, France; smadden@cea.fr.
17
National Optical Astronomy Observatory, 950 North Cherry Avenue, Tucson, AZ 85726-6732; jmould@noao.edu.
18
Department of Astronomy, University of Michigan, 830 Dennison Building, Ann Arbor, MI 48109-1042; msoey@umich.edu.
19
Department of Physics and Astronomy, University College London, Gower Street, London WC1E 6BT, UK; ljs@zuaxp0.star.ucl.ac.uk.
20
Commonwealth Science and Industrial Research Organization, Head Office, GPO Box 4908, Melbourne, VIC 3001, Australia; lister.staveley-smith@
csiro.au.
21
NASA Ames Research Center, SOFIA Office, MS 211-3, Moffett Field, CA 94035; tielens@astro.rug.nl.
22
Current address: Department of Physics and Astronomy, University of Denver, Denver, CO 80208; and NASA Ames Research Center, URSA SOFIA Office, MS
211-3, Moffett Field, CA 94035; tueta@mail.sofia.usra.edu.
23
Gemini Observatory, Northern Operations Center, 670 North A’ohuku Place, Hilo, HI 96720; kvolk@gemini.edu.
24
Jet Propulsion Laboratory, 4800 Oak Grove Boulevard, MS 264-767, Pasadena, CA 91109; mwerner@sirtfweb.jpl.nasa.gov.
2268
The Astronomical Journal, 132:2268–2288, 2006 December
# 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A.

consistent with the fact that stars dominate the point-source catalogs and the dusty objects detected at the longer
wavelengths are rare in comparison. The SAGE epoch 1 point-source catalog has 4 ; 10
6
sources, and more are antici-
pated when the epoch 1 and 2 data are combined. Using Milky Way ( MW ) templates as a guide, we adopt a simplified
point-source classification to identify three candidate groups—stars without dust, dusty evolved stars, and young stellar
objects—that offer a starting point for this work. We outline a strategy for identifying foreground MW stars, which may
comprise as much as 18% of the source list, and background galaxies, which may comprise 12% of the source list.
Key words: du st, extinction — ISM: general — Magellanic Clouds — stars: AGB and post-AGB —
stars: formation — stars: mass loss — supergiants — surveys
Online material: color fi gures, ma chine -readable table
1. INTRODUCTION
The interstellar medium (ISM) plays a central role in the evo-
lution of galaxies as the birth site of new stars and the repository
of old stellar ejecta. The formation of new stars slowly consumes
the ISM, locking it up for millions to billions of years. As these
stars age, the winds from low-mass, asymptotic giant branch
(AGB) stars, high-mass, red supergiants (RSGs), and supernova
explosions inject the nucleosynthetic products of stellar interiors
into the ISM, progressively increasing its metallicity. This con-
stant recycling and the associated enrichment drive the evolution
of a galaxy’s baryonic matter and change its emi ssion character-
istics. To understand this recycling, we have to study the physical
processes of the ISM, the formation of new stars, the injection of
mass by evolved stars, and their interrelationship on a galaxy-wide
scale.
Among the nearby galaxies, the Large Magellanic Cloud ( LMC)
is the best astrophysical laboratory for studies of the life cycle of
baryonic matter, because its proximity (50 kpc; Feast 1999) and
its favorable viewing angle (35

; van der Marel & Cioni 2001)
permit studies of the resolved stellar populations and ISM clouds.
The ISM in the Milky Way (MW) and in the Small Magellanic
Cloud (SMC) is confused in infrared ( IR) images due to crowding
along the line of sight. In contrast, all LMC features are at approx-
imately the same distance from the Sun, and there is typically only
one substantial cloud along a given line of sight, so their relative
masses and luminosities are directly measurable. The LMC also
offers a rare glimpse into the physical processes in an environment
with spatially varying subsolar metallicity (Z  0:30:5 Z

;
Westerlund 1997, p. 234) that is similar to the mean metallicity
of the ISM during the epoch of peak star formation in the universe
(redshift of 1.5; Madau et al. 1996; Pei et al. 1999). The dust-to-
gas mass ratio has real spatial variations and is 2–4 times lower
than the value for the solar neighborhood (Gordon et al. 2003),
resulting in substantially higher ambient UV fields than in the so-
lar neighborhood. The LMC has been surveyed with many instru-
ments, revealing structures on all scales and a global asymmetry
that varies with wavelength (Fig. 1). The ISM gas that fuels star
formation (Fukui et al. 1999; Mizuno et al. 2001; Y. Fukui et al.
2006, in preparation; Staveley-Smith et al. 2003; Kim et al.
2003), the stellar components that trace the history of star
formation ( Zaritsky et al. 2004; Van Dyk et al. 1999; Nikolaev
& Weinberg 2000; Holtzman et al. 1999; Olsen 1999; J. Harris
& D. Zaritsky 2006, in preparation), and the dust (Schwering
1989; Egan et al. 2001; Zaritsky et al. 2004) have all been mapped
in the LMC ( Fig. 1). From the perspective of galaxy evolution,
the LMC is uniquely suited to study how the agents of evolu-
tion, the ISM and stars, interact as a whole in a galaxy that has
undergone tidal interactions with other galaxies, the MW and
the SMC (Zaritsky & Harris 2004; Harris & Zaritsky 2004; Bekki
& Chiba 2005).
The study of the life cycle of galaxies has been hampered by
the association of dust with the key objects driving this galactic
evolution—evolved stars and protostars—and the associated ex-
tinction of the light of these objects. However, the absorbed stellar
light is reradiated by the dust in the IR, and this emission provides
an effective tracer of stellar mass loss, star formation, and the ISM
in general. The launch of the Spitzer Space Telescope (Werner
et al. 2004) with its sensitive detector arrays provides the nec-
essary IR tools for the Surveying the A gents of a Galaxy’s Evo-
lution (SAGE) survey, which surveys the ISM and stars and
thereby traces the life cycle of baryonic matter. We are conduc-
ting a uniform 7

; 7

survey of the LMC in all the IRAC (3.6,
4.5, 5.8, and 8.0 m) and MIPS (24, 70, and 160 m) bands
(Fig. 1). The SAGE project builds on previous IR surveys of
the LMC. The all-sky IR survey by IRAS included 8N5 ; 8N5of
pointed observations of the LMC imaged at12, 25, 60, and 100 m
(Fig. 1) with an angular resolution of 1
0
and resulted in a point-
source list of 1823 objects (Schwering 1989). The 10

; 10

Mid-
course Space Experiment (MSX ) imaging survey of the LMC at
8.3 m with an angular resolution of 20
00
provided a source list
of 1806 objects with more precise positions than IRAS, enabling
cross-correlation with ground-based near-IR surveys such as the
Two Micron All Sky Survey (2MASS; Egan et al. 2001). Both of
these previous far-IR surveys revealed the most luminous dusty
inhabitants of the LMC—supergiants, AGB stars, H ii regions,
and planetary nebulae—but lacked the angular resolution and
corresponding point-source sensitivity to detect the more popu-
lous, less luminous sources. On the other hand, the ground-based
near-IR surveys of the LMC based on 2MASS at J (1.25 m),
H (1.65 m), and K
s
(2.15 m) (Nikolaev & Weinberg 2000) and
DENIS at I (0.8 m), J (1.25 m), H (1.65 m), and K
s
(2.15 m)
(Cioni et al. 2000) revealed 820,000 and 1.3 million sources,
respectively, consisting of red giants, AGB stars, and supergiants.
These near-IR surveys detected many more sources than the far-
IR surveys because of their better angular resolution and because
their wave bands are more sensitive to stellar photospheres, which
are more numerous than the dust-enshrouded objects. One of the
purposes of SAGE is to push this IR survey work to the fainter and
more numerous dusty sources in the LMC.
The remainder of this paper is organized as follows. Section 2
describes SAGE, including the science drivers and observing
strategy. Section 3 describes the data-processing approach and
the Legacy aspects of SAGE. Section 4 takes a preliminary look
at some of the initially processed SAGE data for a region sur-
rounding N83 and N79, near the wes tern end of the bar. Section 5
summarizes the paper.
2. SAGE OBSERVING PROGRAM
SAGE is a uniform and unbiased 7

; 7

survey of the LMC
in all the IRAC (3.6, 4.5, 5.8, and 8 m) and MIPS (24, 70, and
160 m) bands using 508 total hours (291 IRAC, 217 MIPS) of
SPITZER SAGE SURVEY OF THE LMC. I. 2269

Fig. 1.—Examples of the many existing LMC surveys: H i (Staveley-Smith et al. 2003; Kim et al. 2003), CO ( Fukui et al. 2001; Y. Fukui et al. 2006, in preparation),
IRAS 100 m, H (Gaustad et al. 2001) with the SAGE MIPS coverage overlaid, UV (Smith et al. 1987), and stellar density (Zaritsky et al. 2004) with the SAGE IRAC
coverage overlaid. The SAGE survey maps the LMC twice. Epoch 1 coverage is outlined in green, and epoch 2 coverage is outlined in black. The location of the
N79/ N83 region, discussed in x 4, is outlined by the square box in the IRAS 100 mimage.
TABLE 1
Principal Characteristics of the SAGE Survey, Spitzer Program ID 2020 3
Characteristic IRAC Value MIPS Value
Nominal center point
a
.................................................................  ¼ 05 18 48,  ¼68 34 12  ¼ 05 18 48,  ¼68 34 12
Survey area (deg)......................................................................... 7.1 ; 7.1 7.8 ; 7.8
AOR size...................................................................................... 1N1 ; 1N125
0
; 4

AOR grid size.............................................................................. 7 ; 719; 2
Total time ( hr) ............................................................................. 290.65 216.84
k (m) .......................................................................................... 3.6, 4.5, 5.8, and 8 24, 70, and 160
Pixel size at k (arcsec)................................................................. 1.2, 1.2, 1.2, and 1.2 2.5, 9.8, and 15.9
Angular resolution at k (arcsec) .................................................. 1.7, 1.7, 1.9, and 2 6, 18, and 40
Exposure time per pixel at k (s).................................................. 43, 43, 43, and 43 60, 30, and 6
Predicted point-source sensitivity, 5  at k (mJy) ...................... 0.0051, 0.0072, 0.041, and 0.044 0.5, 30, and 275
Predicted point-source sensitivity, 5  at k (mag)...................... 19.3, 18.5, 16.1, and 15.4 10.4, 3.5, and 0.6
Saturation limits at k (Jy) ............................................................ 1.1, 1.1, 7.4, and 4.0 4.1, 23, and 3
Saturation limits at k (mag)......................................................... 6, 5.5, 3.0, and 3.0 0.60, 3.7, and 3.2
Surface brightness limits 5  at k ( MJy sr
1
)............................ ... , ... , 0.5, and 1 1, 5, and 10
Epoch 1 ........................................................................................ 2005 Jul 15–26 2005 Jul 27–Aug 3
Epoch 2 ........................................................................................ 2005 Oct 26–Nov 2 2005 Nov 2–9
a
Units of right ascension are hours, minutes, and seconds, and units of decl ination are degrees, arcminutes, and arcse conds.

Spitzer. The principal characteristics of SAGE are summarized
in Table 1. The spatial coverage shown in Figure 1 extends be-
yond the IR edge of the LMC’s star formation activity and pro-
vides adequate background for calibration and measurement of
non-LMC populations. The point-source sensitivity estimates for
the SAGE survey, listed in Table 1, improve on previous IR sur-
veys with MSX (Egan et al. 2001) and IRAS (Schwering 1989)
by a factor of 1000 and have better wavelength coverage. This
sensitivity improvement is due to the more sensitive detectors
and improved angular resolution. The angular resolution is 2
00
(0.5 pc at the distance to the LMC) in the IRAC bands and 6
00
(1.5 pc),
18
00
(4.5 pc), and 40
00
(10 pc) in the MIPS 24, 70, and 160 m
bands, respectively. The IRAC 8 m and MIPS 24 m images
have a 100 times better areal resolution, which is proportional to
angular resolution squared, than the MSX (8 m) and IRAS 25 m
surveys. The MIPS 70 m band has an 11 times better areal res-
olution compared to the IRAS 60 msurvey.TheMIPS160m
band has a 2.3 times better areal resolution compared to the IRAS
100 m survey. Below we describe the science drivers for the sur-
vey’s characteristics and the observing strategy implemented to
meet these goals.
2.1. Scienc e Dri
vers
SAGE is designed to detect the population of IR point sources
down to the confusion limit imposed by Spitzer’s spatial resolu-
tion and to map, with high signal-to-noise ratio (S/N), the dust
emission from diffuse and molecular clouds, photodissociation
regions (PDRs), and H ii regions. SAGE’s coverage and sensi-
tivity limits are driven by the science goals of the SAGE survey
in three areas: star formation, evolved stars, and ISM.
2.1.1. Star Formation
Star formation in the LMC appears to be a stochastic process,
in which stars form in clumps, clusters, and supershells (e.g.,
Panagia et al. 2000; Walborn et al. 1999). Star formation may be
self-propagating through the energetic feedback of stellar winds
and supernovae (e.g., Oey & Massey 1995; Efremov & Elmegreen
1998), but this stellar feedback also acts to eventually squelch star
formation by dissipating the local ISM (Yamaguchi et al. 2001;
Israel et al. 2003). The CO survey of the LMC (Fukui et al. 2001;
Y. Fukui et al. 2006, in preparation) has uncovered 272 giant mo-
lecular clouds (GMCs) with masses and radii similar to MW GMCs.
Current optical and near-IR observations reveal that one-third of
the LMC GMCs are forming compact young massive star clusters,
e.g., R136 in 30 Dor, while an equally large fraction exhibit no
massive star formation (Blitz et al. 2006; Y. Fukui et al. 2006, in
preparation). This contrast in star formation activity may indicate
a phase of deeply embedded or lower mass star formation in the
latter. To date, searches for IR young stellar objects (YSOs) have
been targeted near current star formation. The first detected YSO
(or protostar), N159-P1 (Gatley et al. 1982), and its surroundings
have been the subject of several follow-ups, the most recent of
which is a detailed Sp i tz er study of the region by Jones et al.
(2005). Studies of the Henize 206 region by Gorjian et al. (2004)
and the LMC superbubble, N51D, by Chu et al. (2005) show the
potential of Spitzer imaging in discovering and characterizing new
YSOs in the LMC. In order to obtain an unbiased and complete
census of star formation throughout the LMC, SAGE is required
to be sensitive to all star formation activity from the massive star
formation traced by H ii regions to the lower mass star formation
traced by Taurus-like c omplexes ( Fig. 2). Variability of some
YSOs, e.g., FU Ori systems, may be detectable in the two epochs
of photometry.
2.1.2. Evolved Stars
High mass loss during the AGB and RSG phases leads to the
formation of circumstellar envelopes that are observable via their
dust emission in all IRAC and MIPS bands. Stellar mass loss can
drive the late stages of stellar evolution, yet the mechanism for
mass loss remains poorly understood. Moreover, this mass loss is
Fig. 2.—Spitzer CMDs showing the discovery space for SAGE. Square brackets denote the brightness in Vega magnitudes at the wavelength enclosed in the
brackets. For example, [8.0] means the Vega magnitude at 8.0 m. The symbols identify populations of key sources throughout the LMC: YSOs (1–30 M

), H ii re-
gions, Taurus-like clusters, O- rich and C-rich AGB stars, RSGs, and main-sequence O sta rs. Symbols, defined in the figure, repre sent the template/model phot ometry
of Cohen (1993) and Whitney et al. (2004). SAGE’s sensit ivity limit (solid line) falls 1000 times below the MSX limit (dashed line) and the lower limit to AGB mass
loss, >10
8
M

yr
1
(dotted line). The circle with a star represents a subregion of Taurus containing 12 stars, placed at the distance of the LMC. [See the electronic
edition of the Journal for a color version of this figure.]
SPITZER SAGE SURVEY OF THE LMC. I. 2271

a dominant source of dust and gas return to the ISM. However,
present estimates disagree on the relative contributions from these
different stellar classes to the injected mass budget of a galaxy
(Tielens 2001). Measuring the mass-loss rates of the entire popu-
lation of evolved stars will help constrain stellar evolution model-
ing and the total returned mass to the LMC’s ISM.
Studies based on previous IR surveys have delved into these
evolved star topics. The IRAS catalog was used by Loup et al.
(1997) to select 198 mass-losing AGB star s. Trams et al. (1999)
followed up on this IRAS-selected sample with the Infrared Space
Observatory (ISO), deriving colors and chemical compositions.
Van Loon et al. (1999) derived mass-loss rates for these ISO
sources, finding a trend of increasing m ass-loss rate with lumi-
nosity. However, this IRAS-selected sample was limited to only
the most luminous AGB stars (L > 10
4
L

) with high mass-loss
rates (5 ; 10
6
M

yr
1
). Using ISOCAM, Loup et al. (1999)
detected 300 mass-losing AGB stars at significantly lower
luminosities (10 mag at 8 m) and mass-loss rates but over a
limited area, 0.5 deg
2
in the LMC bar.
In order to obtain a complete picture of mass loss among the
LMC’s evolved stars, each epoch of SAGE photometry is required
to be sensitive to all evolved stars with mass loss that produces
dust at significant rates; i.e., m ass-loss rates >10
8
M

yr
1
(Fig. 2). In addition, we will be able to constrain the variability
of evolved stars by comparing photometry derived by analyz-
ing the two epochs separately. The 3 month separation of the
epochs is well-suited to detecting evolved star variability,
which typically has a period of approximately 1 yr ( Wood et al.
1999).
2.1.3. ISM
The dust proper ties in the different phases of the ISM pro-
vide insight into the evolution of the dust between phases, as
well as the relationship of the dust components to stellar sources
of UV radiation and kinetic energy. UVextinction measurements
have indicated that the dust properties in the LMC vary spa-
tially (Gordon et al. 2003). Most of the dust mass is in the larg-
est grains and is traced by far-IR dust emission, such as in the
MIPS 70–160 m images. Comparison of these images with the
H i (Staveley-Smith et al. 2003; Kim et al. 2003) and CO data
(Fukui et al. 1999) can be used to map out the dust-to-gas ratio
across the LMC to search for variations. In addition to the amount
of dust, the grain size distribution can be measured using the color
ratios, where IRAC 3.6, 5.8, and 8 m trace polycyclic aromatic
hydrocarbon (PAH) emission, MIPS 24 m traces small grains,
and the MIPS 70 + 160 m traces larger grains. In particular, var-
iations in the properties of the smallest grains, as traced by PAH
emission, are of fundamental importance to the thermodynamics
of the ISM because small grains are very efficient in heating the
gas through the photoelectric effect ( Bakes & Tielens 1994). The
analysis of the IRAS data on the LMC indicates a lower 12 mdif-
fuse emission in comparison to the MW and suggests a deficit of
very small dust grains, possibly due to the intense UV radiation of
the LMC (Schwering 1989). SAGE’s complete IRAC mapping
of the lower metallicity LMC will yield high-resolution insight into
recent work on the paucity of PAH emission in low-metallicity
galaxies (Madden 2000, 2006; Houck et al. 2004; Engelbracht
et al. 2005; Galliano et al. 2005; Dale et al. 2005; Wu et al. 2006;
O’Halloran et al. 2006). The absence of PAH and small grains
will have a profound influence on t he gas heating and the exis-
tence of cold and warm phases in the ISM (Wolfire et al. 1995).
In order to carry out these ISM studies, SAGE must be sensi-
tive to diff use dust emission corresponding to column densities
>1:2 ; 10
21
Hcm
2
(A
V
¼ 0: 2 mag). Residual images, which
are images with the point sources subtracted and smoothed to
improve the S/Ns, will be used for the studies of the diffuse
ISM. The angular resolution achieved by SAGE is sufficient to
separate the stars from the ISM and to distinguish the major cloud
populations: H ii regions, photodissociation regions, molecular
clouds, atomic clouds, and diffuse medium.
2.2. Obser
ving Strategy
2.2.1. Mapping Strategy
To achieve the science and sensitivity goals, the LMC was
mapped at two different epochs separated by 3 months, as de-
tailed in Table 1. The region is many times too large to map with
one Astronomical Observing Request (AOR), so the survey area
was divided into smaller regions that could be efficiently planned
and scheduled. For IRAC, the area was divided into 7 ; 7 tiles of
1N1 ; 1N1 each, composed of 14 ; 28 pointings of high dynamic
range (HDR) 0.6 and 12 s frames with half-array steps with a to-
tal duration of 10,687 s per AOR. The HDR 0.6 and 12 s frames
have corresponding exp osure times of 0.4 and 10.4 s. We refer to
these frames as the 0.6 and 12 s frames throughout the paper.
Mapping steps were done instead of dithers to minimize the time
required to cover the desired area. This IRAC mapping tech-
nique has been used with good success on the Galactic Legacy
Infrared Mid-Plane Survey Extraordinaire (GLIMPSE) Spitzer
project (Benjamin et al. 2003). Each position in the SAGE IRAC
survey has at least four frames of coverage, resulting in an expo-
sure time per pixel of at least 43.2 s in all IRAC bands for the com-
plete survey and a quarter of that, 11 s, for the single-frame
photometry of each epoch. The LMC was mapped in 145.5 hr
per epoch, for a total of 291 hr of IRAC observing time.
For MIPS, the approximately 7N8 ; 7N8 region ce ntered on the
LMC was covered by 38 AORs, each covering approximately
25
00
; 4

.AMIPSAORconsistsof104

fast-scan legs with half-
array cross-scan steps, with a duration of 2.85 hr. The LMC is
mapped with a 19 ; 2 grid of these AORs, taking 108.5 hr per
epoch, or a tota l of 217 hr. Tight sequential constraints relative to
the roll angle rate of change have been invoked so that neighbor-
ing long strips have sufficient overlap. We have carefully designed
our MIPS strategy to allow for off-source measurements in every
scan leg, which allows for accurate self-calibration of the instru-
mental effects. While the MIPS fast-scan mode does not achieve
full coverage at 160 m, the combined epoch 1 and 2 map has
a good basket-weave pattern with small gaps less than a pixel in
size. The well-sampled 160 m point-spread function (PSF;
3 pixels per FWHM) means that interpolation can be used to
fill the gaps. Each position in the SAGE MIPS survey has 20, 10,
and 3 frames of coverage at 24, 70, and 160 m, respectively.
The exposure times per pixel are 60, 30, and 6 s at 24, 70, and
160 m, respectively, for the complete survey and half these val-
ues for each epoch’s photometry.
The mapping strategy maximizes observing efficiency while
minimizing artifacts that compromise data quality and limit the
scientific interpretation. The IRAC and MIPS artifacts fall into
two classes: random artifacts (e.g., cosmic rays and bad pixels)
and systematic artifacts that are tied to pixel location and usu-
ally systematically affect rows/columns. The random artifacts are
easily removed, since our mapping strategy provides four images
at each location (two overlapping images per epoch). The 3 month
time baseline between epochs is ideal for the removal of the
systematic artifacts, because it provides a 90

roll angle in the ori-
entation of the detectors, which optimally removes the ‘‘striping’’
artifacts in M IPS and IRAC image data. In addition, these two
MEIXNER ET AL.2272 Vol. 132

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Spitzer survey of the large magellanic cloud: surveying the agents of a galaxy’s evolution (sage). i. overview and initial results" ?

The authors are performing a uniform and unbiased imaging survey of the LargeMagellanic Cloud ( LMC ; 7 ; 7 ) using the IRAC ( 3. 6, 4. 5, 5. 8, and 8 m ) andMIPS ( 24, 70, and 160 m ) instruments on board the Spitzer Space Telescope in the Surveying the Agents of a Galaxy ’ s Evolution ( SAGE ) survey, these agents being the interstellar medium ( ISM ) and stars in the LMC. This paper provides an overview of the SAGE Legacy project, including observing strategy, data processing, and initial results. The authors present initial results on the epoch 1 data for a region near N79 and N83. 

Q2. What are the standard DAT reduction steps for the MIPS data?

The standard DAT reduction steps for the MIPS 70 and 160 m data are read rejection (autoreject, missing, and saturated reads), electronic nonlinearity correction, ramp jump detection (cosmic rays), ramp line fitting, correction for response variations using the frequent stimulator measurements, dark subtraction, correction of the illumination pattern of the simulator, flux calibration, distortion correction, residual cosmic-ray detection, and mosaicking. 

Q3. What are the sources that appear at the longer Spitzer wavelengths?

The point sources that appear at the longer Spitzer wavelengths have a significant amount of dust emission associated with them, such as YSOs, evolved stars, and background galaxies. 

Q4. What was used to do the processing and mosaicking of the individual images?

MIPS PipelineThe MIPS Data Analysis Tool version 3.02 (DAT; Gordon et al. 2005) was used to do the processing and mosaicking of the individual images. 

Q5. How is the final catalog expected to meet the predicted sensitivities?

The final catalog, which will combine epochs 1 and 2, is expected to meet the predicted sensitivities because the integration time for epoch 1 is only half of the final, and thus a factor of 1.4 (= ffiffiffi 2 p ) improved sensitivity is anticipated. 

Q6. What is the time baseline for the removal of systematic artifacts?

The 3month time baseline between epochs is ideal for the removal of the systematic artifacts, because it provides a 90 roll angle in the orientation of the detectors, which optimally removes the ‘‘striping’’ artifacts in MIPS and IRAC image data. 

Q7. What is the background in the high-background photometry iterations?

In these high-background photometry iterations, the sky plus nebular emission background is subtracted, and the photometry is done on the background-subtracted images. 

Q8. What is the importance of the dust properties in the different phases of the ISM?

In particular, variations in the properties of the smallest grains, as traced by PAH emission, are of fundamental importance to the thermodynamics of the ISM because small grains are very efficient in heating the gas through the photoelectric effect (Bakes & Tielens 1994). 

Q9. What is the role of the interstellar medium in the evolution of galaxies?

The interstellar medium (ISM) plays a central role in the evolution of galaxies as the birth site of new stars and the repository of old stellar ejecta. 

Q10. How did Van Loon et al. (1999) determine mass loss rates for these?

Van Loon et al. (1999) derived mass-loss rates for these ISO sources, finding a trend of increasing mass-loss rate with luminosity. 

Q11. What is the astrophysical laboratory for studying the life cycle of baryonic matter?

Among the nearby galaxies, the LargeMagellanicCloud (LMC) is the best astrophysical laboratory for studies of the life cycle of baryonic matter, because its proximity ( 50 kpc; Feast 1999) and its favorable viewing angle (35 ; van der Marel & Cioni 2001) permit studies of the resolved stellar populations and ISM clouds. 

Q12. How do you determine whether sources are over or undersubtracted?

By doing small-aperture photometry on the residual image at the location of every extracted source, one can assess whether the extracted sources are over or undersubtracted. 

Q13. How do the authors estimate the AV of the diffuse emission in the LMC?

From these diffuse emission sensitivity limits in the MIPS and IRAC 5.8 and 8 m bands, the authors estimate a minimum detectable column density of 1:2 ; 1021 H cm 2 (AV ¼ 0:2 mag) by assuming a solar neighborhood spectral energy distribution for the diffuse dust emission (Desert et al. 1990) and the LMC gas-to-dust ratio. 

Q14. How did Loup and his colleagues find the mass loss rates of the LMC?

Using ISOCAM, Loup et al. (1999) detected 300 mass-losing AGB stars at significantly lower luminosities ( 10 mag at 8 m) and mass-loss rates but over a limited area, 0.5 deg2 in the LMC bar.