SINGS: THE SIRTF NEARBY GALAXIES SURVEY

Physics of the Star-Forming ISM and Galaxy Evolution

SIRTF Legacy Science Project Proposal

Principal Investigator: Robert C. Kennicutt, Jr.
Steward Observatory
University of Arizona
Tucson, AZ 85721
Co-Investigators: Lee Armus SIRTF Science Center
Daniela Calzetti Space Telescope Science Institute
Daniel Dale Caltech
Bruce Draine Princeton University
Chad Engelbracht University of Arizona
Karl Gordon University of Arizona
George Helou Caltech
David Hollenbach NASA Ames
Claus Leitherer Space Telescope Science Institute
Sangeeta Malhotra Space Telescope Science Institute
Michael Regan Space Telescope Science Institute
George Rieke University of Arizona
Marcia Rieke University of Arizona
Michele Thornley NRAO/Bucknell University


ABSTRACT

We propose a comprehensive Legacy survey to characterize the infrared emission across the entire range of galaxy properties and star formation environments, including regions that until now have been inaccessible at infrared wavelengths. SINGS will provide: 1) new insights into the physical processes connecting star formation to the ISM properties of galaxies; 2) a vital foundation of data, diagnostic tools, and astrophysical inputs for understanding SIRTF observations of the distant universe and ultraluminous and active galaxies; and 3) an archive that integrates visible/UV and IR/submillimeter studies into a coherent self-consistent whole, and enables many follow-up investigations of star formation and the ISM.

SINGS will characterize the large-scale infrared properties of galaxies and their principal infrared-emitting components through SIRTF imaging and low-resolution spectroscopy of 75 nearby galaxies (d<30 Mpc), and targeted high-resolution spectroscopy of their centers and a representative set of extranuclear IR-emitting regions in the galaxies. These data will be combined with an extensive library of ground- and space-based data at other wavelengths.

1. OVERVIEW: LAYING THE FOUNDATION


Nearly half of the bolometric luminosity of the local universe is channeled through the mid- and far-infrared emission of galaxies; this spectral region directly probes the youngest star forming regions and their associated interstellar gas and dust. However, existing infrared instruments have only begun to probe the millionfold ranges in star formation rates, ISM environments, and dynamical environments found in external galaxies.

We propose a comprehensive Legacy survey to characterize the infrared emission across the entire range of galaxy properties and star formation environments, including regions that until now have been inaccessible at infrared wavelengths. SINGS will provide: 1) new insights into the physical processes connecting star formation to the ISM properties of galaxies; 2) a vital foundation of data, diagnostic tools, and astrophysical inputs for understanding SIRTF observations of the distant universe and ultraluminous and active galaxies; and 3) an archive that integrates visible/UV and IR/submillimeter studies into a coherent self-consistent whole, and enables many follow-up investigations of star formation and the ISM.

SINGS will characterize the large-scale infrared properties of galaxies and their principal infrared-emitting components through SIRTF imaging and low-resolution spectroscopy of 75 nearby galaxies (d < 30 Mpc), and targeted high-resolution spectroscopy of their centers and a representative set of extranuclear IR-emitting regions in the galaxies. These data will be combined with an extensive library of ground- and space-based data at other wavelengths. The primary products will be:

 $"ullet$"  A spectrophotometric library of IR-emitting galaxy components (nuclei, star forming regions, IR-selected objects) spanning the full range of accessible metallicities, densities, luminosities, and radiation fields present in the normal galaxies, as derived from spatially-sampled IRS spectra;

 $"ullet$"  A complete pixel-resolved SED library of 75 representative galaxies, extending from the visible to 160 m (and in many cases from the UV to submillimeter), and covering the full range of physical properties found in the local universe, as derived from full-area IRAC and MIPS imaging, and large-area spectral maps with MIPS and IRS.


The scientific core of SINGS is a comprehensive study of the physics of the star-forming ISM in galaxies. We will use the above products to derive:

 $"ullet$"  Complete, unbiased maps of massive star formation, through combination of our new data on dust-obscured star forming regions with existing UV and H$"lpha$" measurements

 $"ullet$"  Spectral diagnostics and physical properties of the set of discrete IR-emitting regions

 $"ullet$"  A robust model for the radiative transfer of stellar radiation through dust in galaxies, and the heating and infrared emission of the dust

 $"ullet$"  A critical comparison and integration of UV, H$"lpha$", and IR-based star formation rates (SFRs) in galaxies

 $"ullet$"  Definitive insights into the physical relationships between the large-scale SFR and the physical properties of the galaxies and their ISM.


In addition to fulfilling these scientific objectives, SINGS is designed to maximize the long-term scientific return for the broader astronomical community, and to serve as a foundation for GO projects. Key elements in this design include:

 $"ullet$"  Unbiased, physically-based target samples that cover the full range of physical properties of local galaxies and IR-emitting sources

 $"ullet$"  Homogeneous and complete SIRTF data sets, including uniform wavelength coverage, sensitivity limits, and spatial sampling for the spectroscopic targets

 $"ullet$"  A comprehensive set of coordinated BVRIJHK, H$"lpha$", UV, far-infrared, CO, and HI data, fully integrated into the Legacy data products

 $"ullet$"  A customized web-based interface with tools to access the SIRTF and ancillary data, and data releases timed to support GO proposal planning

 $"ullet$"  Consideration of approved GTO programs in the project design, to minimize duplication and maximize the long-term value of the combined SIRTF database.

As a result, SINGS will serve dozens of archival investigations, GO projects, and other Legacy projects. For example, our multi-wavelength SED and spectral data products will be applicable to long-wavelength studies of the distant universe and ultraluminous infrared galaxies. Our observations of the centers of normal galaxies will provide insight into the link between AGN activity and star formation. And our observations of discrete star-forming regions and interstellar clouds in nearby galaxies will provide a reference for more detailed studies of our own Galaxy. SINGS will also have broad applicability for interpreting future observations of galaxies with such diverse instruments as ALMA, FIRST, and NGST.

Our approach and project design minimize the risk of missing crucial observations across the range of galaxy types over the limited lifetime of SIRTF, and provides a critical mass of homogeneous data in the unwelcome event of a premature mission failure. It also avoids the inevitable redundancy in a combined set of many disconnected GO projects. The standardized set of observations, calibrations, data processing, and data access tools will be invaluable for interpreting observations spanning more than one galaxy or class of objects.

2. SCIENTIFIC JUSTIFICATION

Information on the large-scale distribution of star formation in galaxies is critical to a host of larger astrophysical problems: the physical nature and origin of the Hubble sequence, the structure and phase balance of the ISM, the interpretation of observations of the high-redshift universe, the physical nature and triggering of starbursts, and as a lynchpin for the physical understanding and modeling of galaxy formation and evolution.

In contrast to the study of the formation of individual stars, which has advanced dramatically in the past two decades, our understanding of star formation on galactic scales remains relatively immature. The limitations begin at the very foundation, with our limited ability to measure and map the large-scale star formation rates (SFRs) in galaxies. SIRTF's unique contribution derives from its ability to probe the flow of energy from stars into the ISM on all scales, ranging from the immediate environs of star-forming molecular clouds to the radiative transfer of diffuse radiation on kiloparsec scales.

Within star-forming regions, stellar radiation shining on the surrounding gas produces a series of ISM components, all of which are probed by SIRTF (Figure 1). Ionizing radiation produces an HII region that is traced by a suite of mid-infrared fine structure lines from species covering a wide range of ionization states. Outside of the ionization front the gas becomes optically thick to hydrogen-ionizing photons, but ultraviolet photons in the range 6 eV <hnu.gif (857
 bytes) 13.6 eV dissociate molecules, ionize atoms like C, Si, and Fe, and heat the gas via grain photoelectric heating to form a photodissociation region or PDR (Hollenbach & Tielens 1999). Deeper within the clouds, the warm molecular interface is directly observable via the pure rotational transitions of H2. Dust grains are present throughout these phases and provide diagnostics of the physical conditions in the star-forming regions, through the thermal continuum emission and the aromatic emission bands. Much of the infrared continuum emission from a galaxy arises from dust grains in HII regions and PDRs, but dust in ambient clouds illuminated by the general interstellar radiation field (ISRF) can also be important (Lonsdale & Helou 1987). Another energy channel - mechanical energy transfer - can be traced via shock-sensitive features such as H2, SiII, and FeII, which emerge with different ratios than from the HII regions and PDRs.These diagnostic capabilities are exploited most effectively in nearby galaxies (d = 1-30 Mpc), where the angular resolutions of the SIRTF instruments are well-matched to astrophysical scales of interest. For a galaxy at a fiducial distance of 3.5 Mpc (the distance of the M81 group, a primary component of our sample), the spatial resolution of the telescope projects to linear dimensions ranging from 40 pc (2 pixels, 2.5-8 m) to 100, 300, and 700 pc ($"ambda$"/D respectively at 24, 70, and 160 m). At this distance a 10 sec set of MIPS integrations at 24, 70, and 160 m will detect with 10-$"sigma$" significance a star forming region with SFR = 2x10-4 M$"_{odot}$"/yr (comparable to the Orion core), with a typical cloud mass (gas + dust) of order 105 M$"_{odot}$". Likewise a 10 sec integration set with IRAC can detect the dust emission from ~104 M$"_{odot}$" of ISM illuminated by the average ISRF. For high-resolution spectroscopy, IRS would measure the [NeII] 12.8 m emission line from a Orion-luminosity region at 10-$"sigma$" significance in a 30 sec integration, or the H2 emission from the S(0) and S(1) lines at 17 and 28 m of a 5x105 M$"_{odot}$" cloud at 100 K in 500 sec (again at 10$"sigma$").

spectrum11.gif (795220 bytes)

Top: Infrared spectrum a nearby galaxy (M82 from ISO SWS), smoothed to the resolution of the SINGS spectroscopy (Sec 2.1.1). Bottom: A schematic SED of the same region extending from the ionizing UV to submillimeter, smoothed to IRS low-resolution mode at 10-40 m, illustrating the diagnostic power of the SINGS imaging and SED strips (Sec2.1.2).

2.1  Core Project: Global Patterns of Star Formation in Galaxies

Ultimately, the global patterns of star formation and the evolution of galaxies must be regulated by parameters such as gas density, phase, metallicity, and dust content. The goal of SINGS is to distinguish the influences of these parameters by examining a diverse set of galaxies and interstellar environments within them.

To contribute optimally to these investigations, SIRTF must be used in a manner that is best adapted to its angular resolution. We have therefore adopted a two-part strategy. In the first part, 75 galaxies will be imaged completely with IRAC and MIPS, and partially mapped with low-resolution spectroscopy, to characterize the global infrared and star-forming properties of nearby galaxies, and explore their dependencies on type and other integrated properties. In the second part, we will use complete spectroscopic observations to physically characterize a set of discrete IR-emitting sources. These include the nuclear regions of each of the 75 galaxies, and observations of 75 extranuclear regions in ~20 of the galaxies, divided between visible star-forming regions and infrared-bright objects discovered with our SIRTF images, chosen to contain the complete range of physical characteristics. The powerful combination of these two data sets will fully characterize the infrared properties of normal galaxies, and enable a broad set of astrophysical applications.

2.1.1  Components of the IR Emission: The Physics of the Star-Forming ISM

SIRTF allows us for the first time to apply the powerful suite of mid-infrared spectral diagnostics to galactic IR-emitting components which span the full range of metallicities, densities, extinctions, and radiation field properties found in normal galaxies. These observations provide the link between the observed IR spectra of galaxies and the physical conditions in the emitting gas and dust, and are crucial inputs for interpreting the integrated SEDs and spectra of high-redshift objects and more luminous starburst galaxies.

There is a dramatic physical dichotomy between IR-bright and UV-bright star-forming regions on the 100-500 pc scale (Kennicutt 1998a,b). The giant OB/HII associations in disks which dominate the visible star formation have mean emitting gas densities of order 10-100 cm-3, neutral + molecular column densities of order 1020 - 1022 cm-2 and visual extinctions of 0-3 magnitudes. By contrast the well-studied IR-selected star-forming regions are mainly circumnuclear with typical densities of order 103 - 106 cm-3, molecular column densities of 1022 - 1025 cm-2 and visual extinctions of 5-1000 magnitudes (Kennicutt 1998a). Because of this observational segregation neither set of data (or even the two together) provides a complete picture of the physics of star-forming galaxies and the IR-emitting source population. Furthermore the UV/visible and IR spectral diagnostics have not been placed yet on a common empirical foundation, so questions remain about whether we are measuring the physical properties of these regions in a consistent manner. The long-standing controversy over whether the IMF of young stars in IR-luminous starbursts is radically different from that observed in optically-selected star-forming regions is a prime example of this ambiguity (e.g. Rieke et al. 1993; Alonso-Herrero et al. 2000).

ISO already has made inroads into this problem, by producing mid-IR and far-IR spectra for a sample of IR-luminous galaxies and a few lower-extinction star-forming regions in disks (e.g. Sturm et al. 2000, Thornley et al. 2000, Helou et al. 2000, Malhotra et al. 1997, 2000). In addition, several SIRTF GTO programs will provide spectral information on selected regions, skewed toward IR-luminous objects, starbursts, and AGNs.

SINGS is designed to fulfill SIRTF's potential for characterizing the full range of IR-emitting regions found in normal galaxies, using its unprecedented sensitivity to measure an unbiased set of IR-emitting and optically-selected objects, including the centers of every SINGS galaxy and an equal number of bright extranuclear regions (see Sec 3.1 for details). The data set will also provide a comprehensive spectral library of IR-emitting components in local galaxies. Our own interpretation will focus on the following problems:

Physics of Interstellar Dust: The mid-IR emission from the ISM is dominated by Aromatic Features in Emission or AFEs (e.g., Beichman 1987, Boselli et al. 1997). The AFEs appear in two main groups, one in the 5.5 to 9 m region, with peaks at 6.2, 7.7 and 8.6 m, and the other one in the 11 to 12.5 m region. The IRS 5-14 m low-resolution spectra are well matched to study these features over a maximum range of metallicity, stellar radiation field intensity and local galaxy spectral properties. ISO observations have shown that the AFE strengths (relative to the FIR continuum) are sensitive to the local radiation field (e.g. Boselli et al. 1998, Helou et al. 2000), and AFE diagnostics will be used to constrain dust heating and radiative transfer models. The angular resolution of IRS (effectively ~3.6 " in this mode) will minimize the spatial-spectral confusion that hampers much of the lower-resolution ISO data.

Hardness Measures, Stellar Temperatures, and the IMF: The high resolution mode of IRS provides access to the complex of fine structure lines in the mid infrared that are ideal for studying the souces of UV continuum in heavily obscured regions. A particularly useful line set is [NeII] 12.8m, [NeIII] 15.6m and [NeV] 14.3m, but other important lines include [SIII] 18.7,33.5m, [SIV] 10.5m, and [OIV] 25.9m. The ratios of the high-ionization features will be used to constrain the hardness of the ionizing radiation fields and the effective temperatures of the ionizing stars. We will compare with measurements from optical spectra of the same objects, to test the consistency of the IR and optical Teff and IMF scales. The high-ionization species of SIV and OIV can probe for very hot stars (Teff > 60 kK) and shock-ionized gas in the star-forming regions.

Ionization Rates and SFRs from [NeII]: The bright [NeII] 12.8 m line provides a measure of the ionization rate (Roche et al. 1991). We will compare it with our observations of optically-visible HII regions to cross-check the calibration against extinction-corrected H$"lpha$" measurements and (when available) P$"lpha$" and Br$"gamma$" fluxes. Comparison of the [NeII]-based SFRs with the FIR luminosities in dusty regions will provide constraints on the absorption of ionizing radiation by dust and test for systematic errors in the respective SFR scales.

Physical Conditions in Circumnuclear vs Disk SF Regions: Optical, near-IR, and ISO observations hint at significant differences between the physical properties and stellar content of circumnuclear and disk star-forming regions (e.g. Kennicutt et al. 1989, Ho et al. 1997, Dale et al. 1999). Our IRS data will allow us to apply a homogeneous set of nebular diagnostics and photoionization models to both sets of objects, to determine whether there is a single physical sequence of star-forming regions, or distinct physical populations.

Interpretation of PDR, Molecular, and Shock Features: The IRS spectra are expected to detect a number of lines excited in photodissociation regions, in particular the H$"_2$" rotational lines (S(1) - S(7)), [SiII]34.8m, and [FeII]26.0m. The H2 emission constrains the temperature and mass of the warm molecular gas, and the heating rates needed to maintain this component. On galactic scales, this emission is probably dominated by emission from far-ultraviolet heated gas in PDRs, rather than by shock-heated gas (Luhman et al. 1994, Hollenbach & Tielens 1999). Here the H$"_2$" line fluxes help to determine the fraction of the dust IR continuum which arises from UV heating, and the proximity of the OB stars to their natal clouds. The emission lines from PDRs will be calculated by coupling the Starburst99 models with the PDR codes of Kaufman et al. (1999) and Draine & Bertoldi (2000). We will compare the HII region and PDR spectra with emission expected from shocks as well (e.g., Draine et al. 1983, Hollenbach & McKee 1989).

                                n6946.gif (608251 bytes)

H$"lpha$" image of SINGS galaxy NGC 6946 with ISOCAM 15 m contours overlaid, contrasting the distributions and populations of star-forming regions. The H$"lpha$" image is from Ferguson et al. 1998, ApJ, 506, L19.

2.1.2  Large-Scale IR Properties of Galaxies: The Physics of Galaxy Evolution

Apart from a handful of nearby galaxies with resolved stellar populations, most of what is known about the star formation properties of normal galaxies is based on integrated light measurements in the UV and visible (Kennicutt 1983, 1998a). The largest systematic error in these SFR scales is interstellar extinction. The average extinction integrated over a typical star-forming galaxy is ~1 magnitude for both UV (1500-2500 ) and H$"lpha$" (e.g., Calzetti et al. 1994; Bell & Kennicutt 2000), consistent with approximately half of this radiation being re-emitted in the infrared. However this extinction varies enormously within and among galaxies, with the highest extinction occuring in the regions with the highest column densities and SFR per unit area. These regions are all but invisible at UV-visible wavelengths; this completely obscured component represents at least 20-30% of all star formation in the local universe (Heckman 1999), and may be even more important at high redshift (e.g. Lilly et al. 1999). Figure 2 illustrates how different the star formation distributions can be.

The infrared emission from dust offers another route for measuring large-scale SFRs. In galaxies dominated by young stars and with high dust optical depths, the IR continuum provides a near-bolometric measurement of the stellar radiation output and SFR (Kennicutt 1998a), but interpreting this emission in normal galaxies is problematic. There the dust can also be heated by older stars, and the fractional contribution of this "cirrus" component to the total FIR emission is a strong function of galaxy properties and the SFR itself (Helou 1986; Lonsdale & Helou 1987, Rowan-Robinson & Crawford 1989). Moreover, the infrared emission may under-represent the output of regions with low dust content. The modeling of such effects is made difficult by the complex geometry of the dust and heating sources, and by the absence of high-quality spatially-resolved FIR data.

SIRTF promises a breakthrough in this problem. It can produce spatially-resolved IR SED maps of galaxies, with sufficient angular and spectral resolution to separate individual dust emission and heating components. The SIRTF images and low-resolution SEDs can easily reach the sensitivity limits needed to map the emission everywhere in the disks, not just in the dense optically-obscured regions, allowing us to apply a consistent set of diagnostics across the entire population of galaxies and star-forming regions.

To address these goals we will obtain full-area IRAC and MIPS imaging for 75 nearby galaxies, selected to span the parameter space of type, luminosity, metallicity, and IR/optical properties, and inclinations (see Sec 3.1 and Fig 3). Our sample ranges from galaxies with luminosities and SFRs bordering on the Luminous Infrared Galaxy (LIG) class (LIR>1011  L$"_{odot}$"), to elliptical and S0 galaxies with no detectable star formation (to characterize how a quiescent stellar population heats the dust and influences the global IR emission properties).

The coarse spectral sampling of the broadband MIPS and IRAC maps by themselves is insufficient to constrain the infrared SED shapes, much less diagnose the temperature distributions and properties of the emitting grains, as illustrated in Figure 4. Therefore we will supplement our imaging with IRS 14-40 m spectral maps and MIPS 52-99 m SED mode strip scans in all 75 galaxies. These will permit robust modeling of the dust continuum and heating over a several square arcminute slice of each galaxy, and with the images help constrain the spectral shapes throughout the disks. Our resolved SED maps, when combined with CO, HI, and visible/near-IR maps will probe the physical processes controlling cloud formation, gas flows, spiral structure, bars, and the large scale triggers and inhibitors of star formation. Our scientific investigation will focus on the following applications:

figure2a.gif (795220 bytes) figure2b.gif (795220 bytes) figure2c.gif (795220 bytes)

This set of model SED's for normal galaxies from Dale et al. (2000b) illustrates the expected variations in the emission, and the necessity of adding low-resolution spectra in the 14-99 m region to enable the interpretation of the IRAC and MIPS broadband colors (inset).

Comparison and Calibration of Star Formation Rates: The Full Picture: Coordinated H$"lpha$" and UV imaging will be combined with the IR SED maps, providing a detailed inventory of the whole spectrum of star-forming regions, from the completely dust-obscured to completely exposed. The obscured, IR-bright regions will preferentially trace the youngest star-forming cores, whereas the UV-bright regions are mostly those which have already disrupted and dissipated their natal molecular cores. The comparison of the SFR distributions derived at the three wavelengths will test current extinction correction schemes for the UV and H$"lpha$" fluxes, and will be used to develop more robust methods. Our data will enable testing and refining of other potential SFR tracers, e.g. the non-thermal radio continuum emission (Condon 1992, Mobasher et al. 1999). They will also reveal any systematic patterns in obscured vs unobscured star formation, as functions of radial location, metallicity, galaxy type, or dynamical environment (e.g., Kennicutt 1998a, Roussel et al. 1999).

The Star Formation Law: High-resolution CO and HI maps will be correlated on a point-by-point basis with the SFR distributions, to quantify the empirical form of SFR vs gas surface density law (Kennicutt 1998b). The combined data will allow us to trace the SFR over dynamic ranges of >10-5 in gas and SFR surface densities.

Dust Mapping and Modeling: The 14-40 m and 52-99 m radial-strip SEDs are powerful diagnostics of the temperature mix of the emitting dust, because this wavelength region falls on the Wien side of the dust continuum (Dale et al. 2000). Combined with IRAC ratio maps and with the 5-14 m IRS spectra, the SEDs will constrain variations in dust properties and the ISRF as well. They can also be inverted to recover or improve constraints on the temperature distribution of the dust (Li et al. 1999, Dale et al. 2000).

Radiative Transfer Models: The infrared maps, when combined with UV-visible imaging, will provide much more rigorous constraints on dust radiative transfer models of disks than has been possible to date. In addition to building on existing modeling (e.g., Xu & Helou 1996), we will apply Monte Carlo techniques to compute the radiative transfer of photons from arbitrary distributions of stars through arbitrary distributions of dust, using the DIRTY code (Gordon et al. 2000). The model self-consistently computes the dust re-emission including equilibrium thermal emission (large grains), non-equilibrium thermal emission (small grains), and PAH-like molecules.

Dust in Early-Type Galaxies: Observations of elliptical and S0 galaxies will establish the IR-emission properties of galaxies without active star formation, and address several other outstanding questions. Whether the dust in early-type galaxies results from internal processes, such as stellar mass loss, or from mergers has obvious implications for the evolutionary history of early-type systems. Goudfrooij & de Jong (1995) and Merluzzi (1998) found surprisingly that the dust mass required to produce the IRAS FIR emission is much greater than that required to explain the optical extinction and exceeds expectations from stellar mass loss alone. Our survey should resolve this issue by testing whether the distribution of warm dust follows the stellar light, establishing the presence (or not) of PAH and silicate features, and better constraining the dust temperature with the MIPS 160 m data.

Combining the Constraints: The IR SED and multi-wavelength analysis described in this section and the spectral line analysis in the previous section provide complementary constraints on the properties of the dust and the star-forming regions. For instance the dust temperature distribution translates into a heating intensity distribution, constraining the incident heating flux and the dust density distribution of the medium, whereas the spectral line ratios constrain the electron density, spectral hardness and ionizing flux. This will yield constraints on the geometry of the star-forming regions, for instance in terms of the clumping of the molecular cloudlets or the penetration depth of ionizing radiation within clouds.

2.2  Ancillary Data and Coordinated Observations

The scientific and archival value of the Legacy data in this project will be greatly enhanced with a coordinated set of groundbased images and spectra of the program galaxies and targeted sources. In addition to obtaining the data listed below and including it in our Legacy database, we will compile and/or link to archival data from other surveys (e.g., 2MASS, GALEX, HST).

BVRIJHK and H$"lpha$" Imaging: Multi-color broadband imaging in the 0.4-2 m region will trace the stellar mass distribution and provide the necessary inputs for modeling the dust extinction and heating. NOAO observing time is requested for optical imaging of the 60 northern galaxies (KPNO 2.1m) and for optical and IR imaging of the 15 targets that are only accessible from the south (CTIO 1.5m). JHK imaging of the northern subsample will be obtained with the PISCES camera on the Steward Observatory 1.5m and 2.3m telescopes, using up to 60 nights of time guaranteed by the SO Director. These data will be fully reduced and delivered as SINGS Legacy data products.

UV Images and Spectrophotometry: The GALEX mission, to be launched in the same time frame as SIRTF, will provide an all-sky survey in two bands over 1350-3000 , with angular resolution of $"e$" 5 arcsec, well matched to SIRTF and adequate to quantify the UV emission from our entire sample. Images in this wavelength region (1500-2800 ) are also available for 27 of the galaxies from the Ultraviolet Imaging Telescope (UIT) and/or HST, and IUE spectra of the central 10 " x 20 " regions are available for 25 galaxies. These data will be integrated into the database, and included in the final data products.

CO and HI Maps: Aperture synthesis maps are required to complement the SIRTF data by tracing the cold gaseous ISM in the 21-cm HI and millimeter CO lines. High-quality CO maps are available for 20 SINGS spiral galaxies from the BIMA Survey of Nearby Galaxies (SONG). Two of the SONG Co-PIs (Regan, Thornley) are members of our team, and will integrate these data (along with VLA HI maps for ~20 SINGS galaxies) into our data products. Follow-up BIMA observations are also planned for selected northern SINGS galaxies which are not in the original SONG sample.

Optical Spectrophotometry: Spectral drift-scans in the 3600-7000 region will be obtained over the same radial strips scanned with IRS and MIPS, using up to 40 nights of guaranteed time on the Steward Observatory 2.3 m telescope for northern targets, and the CTIO 1.5 m telescope for southern targets. Stellar feature measurements will be used for spectral synthesis, while the emission lines will provide the major nebular diagnostics ([OII], H$"eta$", [OIII], H$"lpha$", [NII], [SII]). The reduced spectra will be provided as Legacy data products.


2.3  Archival Science and Synergies with Other Projects

Although our target samples and observing strategy were designed to address the requirements of the core science program, our own project will tap only part of the rich scientific potential of this data set. Here we briefly highlight a few examples of archival applications of this survey, and the major benefits of this survey to other SIRTF programs (including other likely Legacy projects) and future NASA and ground-based projects.

The High-Redshift Universe: Our complete galaxy SED library covering 0.4-160 m (with more extended UV and submm coverage in many cases) will enable investigators to construct spectral models of any type of galaxy at any redshift, to predict colors, fluxes, and morphologies at any desired wavelength. Such data will be invaluable for other Legacy programs (e.g., cosmology, high-redshift surveys, deep surveys), for interpreting surveys of distant galaxies with SCUBA and other submillimeter facilities, and for planning future facilities such as ALMA, FIRST, Planck and SAFIR.

Ultraluminous Infrared Galaxies (ULIGs): Observations of ULIGs promise to be a major scientific focus of SIRTF, but most of these objects will be poorly resolved (or unresolved) at the wavelengths where most of their energy is emitted. Our spatially-resolved images and spectra will provide the empirical inputs needed to model the composite emission from the ULIGs and will also help to identify the truly unique physical attributes of these objects, and thus constrain their origins and evolutionary paths.

Are Disks Optically Thick?: The ongoing, decade-long controversy over whether face-on spiral disks are largely opaque at visible wavelengths (e.g. Valentijn 1994, Giovanelli et al. 1994, Burstein 1994) should be definitively settled by combining our IRAC/MIPS and archival broadband imaging for spirals covering a range of inclinations.

Cold Dust in Galaxy Halos and Extended Disks?: Nelson et al. (1998) and Alton et al. (1998) have presented tentative evidence for substantial dust halos in nearby galaxies. For galaxies located in low-cirrus parts of the sky our MIPS 70 and 160 $"mu$"m images offer unprecedented sensitivity to extended cold dust, providing a direct test of this result.

Stellar Structure of Disks: The IRAC images at 3.6 and 4.5 m, combined with groundbased near-IR images, will provide constraints on the stellar mass distributions in the galaxies, relevant to a wide range of problems including bar structures and spiral density waves.

Galactic Nuclei and Circumnuclear Environments: Our full spectroscopic coverage of the wide range of nuclear types in the SINGS galaxy sample will provide a comprehensive inventory of the physical properties of a representative set of normal galaxies.

Large-Scale Shocks in the ISM: Large-scale shocks produced by bars, spiral density waves, and interactions are manifested in the mid-IR continuum and the shock sensitive [SiII], [FeII], and H2 features. Our imaging and low-resolution IRS SED strips should isolate examples of these shocks, which can be followed up with pointed spectroscopy by GOs.

The Radio Continuum vs Infrared Correlation: Archival VLA and WSRT radio continuum data for many SINGS galaxies will allow one to study the detailed relation between infrared and radio intensity distributions within the disks of galaxies. This will unravel the physical basis of the not-yet-understood global correlation between the far-infrared and radio continuum luminosities of galaxies revealed by IRAS (Condon 1992, Helou & Bicay 1993).

3. TECHNICAL IMPLEMENTATION PLAN

Our basic observing strategy is depicted in Figure 5, which shows an ISOCAM image of a SINGS target, NGC 6946, on three scales with the footprints of the MIPS, IRAC, and IRS observations overlaid. In this section we describe the target selection and the individual sets of observations. Complete tables of targets are attached as supplemental materials.

fields.gif (306642 bytes)

ISOCAM image of NGC 6946 (a SINGS target), shown at three different scales to illustrate our observing strategy. Left: Blue lines show the 3x4 grid of IRAC imaging, and the dashed red lines show the region to be mapped by MIPS imaging (15$^"prime$"x30$^"prime$"). Middle: Regions to be scanned at low spectral resolution with MIPS SED mode (red), IRS low-resolution 14-40 m (large yellow box) and IRS low-resolution 5-14 m (small yellow box). Right: Expanded image of an extranuclear region, with areas covered by IRS low-resolution 5-14 m (yellow), high-resolution 10-19.5 m (green), and high-resolution 19.3-37 m (violet) superimposed. The orientations shown are illustrative, and will differ depending on scheduling (the observations are not dependent on a particular orientation).

3.1  Sample Definition and Selection

To address the astrophysical goals described in the previous sections, we have chosen a sample of 75 nearby (d $<$ 30 Mpc) galaxies, covering the full range in a 3-dimensional parameter space of physical properties: morphological type (E to Im), luminosity (IR-quiescent to luminous IR galaxy), and FIR/optical ratio (over 3 orders of magnitude). The sample also covers a broad range of other parameters: nuclear activity (quiescent, starburst, LINER, Seyfert), bar structure, star formation rate (SFR), gas content, and dust temperature (see table below). The parameter space has been chosen to maximize our insight into the star formation processes of galaxies and their physical links to the ISM properties.

A wide range of galaxy environments is also included, with several examples of isolated galaxies and members of rich groups, clusters, and interacting systems. Because of its proximity and good visibility to SIRTF, we have specifically targeted the M81 group (Tully 1987), excluding 4 reserved GTO targets (NGC 2366, NGC 3077, UGC 4486, VII Zw 403) and the centers of M81 and M82 (also GTO targets). The galaxies cover the full range of inclinations, to assess the effect of dust attenuation on the overall SEDs. Our sample complements and builds on the GTO observations of nearby galaxies. Where GTO observations are relatively complete (e.g., metal-poor and IR-luminous starbursts, low surface brightness galaxies) we have avoided duplication in favor of integrating the available GTO data into our analysis. In this manner, the full SIRTF legacy on nearby galaxies can be assembled at high efficiency.

The spectroscopic sample is comprised of the nuclear regions of 70 of the 75 SINGS galaxies1 and 75 extra-nuclear regions. A similar physically-based approach was used to define the extra-nuclear target sample. Forty optically-selected HII/OB regions were chosen using groundbased spectroscopic surveys (Zaritsky et al. 1994 and references therein). Since it is local physical conditions that influence the properties of these regions, we need only observe objects in a small subset of nearby galaxies (9) to cover the full physical parameter space. The regions cover large ranges in metal abundance (0.1 - 3 $"Z_odot$"), ionizing luminosity ( QH =1049-1052 ph/sec), visual extinction (0-4 mag), radiation field intensity (ionization parameter log U = -2 to -4), ionizing stellar temperature (Teff = 35 - 55 kK, Kennicutt et al. 2000), and local H2/HI ratio as inferred from CO (< 0.1 to >10). We have restricted targets to those with extinction-corrected f(Himg4.gif (868
 bytes))>=2x10-16 W/m2 (or f(Bralpha.gif (868 bytes))>=6x10-18 W/m2), to ensure adequate S/N in the main fine-structure lines in a reasonable exposure time. To cover the full range of gas densities and extinctions (impossible in an optically selected sample), we will also target 35 IR-selected regions. These will be selected from our first year of IRAC/MIPS imaging, as second-look observations. Flanking aperture observations for the disk regions (IRS high-res) will provide effective background determination and together will provide composite spectra of more quiescent regions in the disk, as will also be true for the 55 "-long apertures for the IRS 5-14 m measurements.

Hubble Type     E  --  Irr
MR -12.5  --  -23.5
LV 5x106  --  2x1011 L$_odot$   (0.0003-10 L$^"*$")
L(IR) <107  --  3x1011 L$"_odot$"
L(IR)/LR <0.02  --  42
F(60)/F(100) 0.16  --  1.2
O/H (0.4 Ro) 0.05  --  3 (O/H)$_odot$
Mgas/Mstars < 0.001  --  5
SFR 0  --  15 M$"_odot$"/yr
SFR/LV < 10-8 --  10-4 M$"_odot$"/yr/L$"_odot$"

3.2  IRAC Imaging

We will image in all four IRAC bandpasses to at least the full optical radii (R25) of the galaxies. Since the majority of our galaxies are not highly inclined, we will nominally mosaic in array coordinates. We have selected mosaic patterns (3x2, 4x3, 5x4, and 6x5 fields) to cover the entire optical disks independent of spacecraft roll angle. We will use a five-position dither of 30 sec exposures as our nominal observation, as a good minimum for the removal of bad pixels and cosmic rays; using 12 sec exposures reduces the clock time by 29%, with overheads becoming a much larger percentage of the time. Because spiral arms and inner disks run from tens to hundreds of MJy/sr at 7m (Dale et al 1999), even a few seconds of integration will saturate some pixels in IRAC. We will use the high dynamic range mode to avoid this problem. Using SPOT we derive total clock times and find the total IRAC exposure time required is 44.6 hours. Galaxy disks have basically exponential profiles in the mid-IR, and the average the 7 m size matches R25 for a brightness level of 0.04 MJy/sr (Dale et al. 2000a). This corresponds to a S/N = 4.5 for a 5-position dither at 8 m. Our observing strategy is thus well matched to the goals of full imaging of quiescent galaxy disks.

3.3  MIPS Imaging

Each galaxy in the sample will be imaged in all 3 MIPS bands, using the default scale at 70m. The sensitivity goal of these observations is $"ge$"3$sigma$ detections within the optical radii (R25) of the galaxies. To estimate the surface brightness levels required, we have used the radial profiles of 11 IRAS HIRES-processed galaxies that are both nearby and nearly face-on (thus the effect of IRAS's low spatial resolution is minimized). The average surface brightness at R25 is 0.21 MJy/sr at 25 m and 0.51 MJy/sr at 60 m, in very good agreement with the mid-IR estimates above scaled by typical galaxy SEDs (Dale et al. 2000b). We have adopted these values as the 3$"sigma$" sensitivity goals at 24 m and 70 m. Surface brightnesses at 160 m are comparable to or greater than those at 60 m (e.g., Engargiola 1991), so we have adopted a sensitivity goal of 0.51 MJy/sr at 160 m. The 3$"sigma$" goals above are equivalent to $1sigma$ point-source sensitivities of 67Jy, 1.9mJy, and 6.1mJy, respectively. Using the medium-background plots for MIPS photometry, the nominal sensitivity is achieved in 1 photometry cycle at 70 and 160 m, and is within 10% of being achieved at 24 m, with 10 sec exposures. However we will perform multiple photometry cycles (2-6, depending on wavelength) to ensure sufficient redundancy, meet the sensitivity goals, enable super-resolution capability (160 m), and produce consistently high-quality images.

We will use compact-source photometry modes for sources smaller than about 2', and large-source modes for sources up to about 4'. Any source larger than 4' will be imaged using the MIPS scan map mode. To improve sampling and mitigate the effects of bad pixels at 160 m, each scan leg will be offset from the previous one by one-half the array width, using the 148" cross-scan step in both the forward and reverse directions. Saturation concerns (similar to the IRAC ones) will be addressed during Phase 1. The decision on whether to use shorter exposures, use fine-scale 70 m channel, or fall back on the SED data to correct saturated pixels in the broad-band maps will be based on inspection of the IRAS and ISO data for each galaxy. The total time required for the MIPS imaging is 124 hours.

3.4  MIPS SED Observations  (Second-Generation AOT)

We will perform a small raster of SED mode observations (52-99 m) in a radial strip in each galaxy, designed to overlap with the IRS low-resolution strip scans. In most galaxies ($D_{25} e 10$$^"prime$"), this raster will consist of 3 adjacent slits to cover in full a 1'x4' region. In galaxies larger than 10$^"prime$", a second, similar set of SED observations will be added to produce a ~1'x8' map. We prefer that these strips be oriented as closely as possible to the major axis, but deviations can be tolerated. However we wish to overlap the MIPS and IRS SED strips as much as possible (a 15  maximum rotation results in benign scheduling constraints). We used the radial profiles of nearby galaxies discussed above to estimate the typical surface brightnesses at 2' to 4' radius and compute the integration times required. The low end of the range is around 1.2 MJy/sr, which implies a $"1sigma$" sensitivity of 2.7 mJy on a point source to achieve $"5sigma$" sensitivity at that surface brightness. This is achieved in 5 cycles of 10 sec integrations. From the prescription for the MIPS SED mode (SIRTF web site documentation), we derive a total time for MIPS SED scans of 77 hours.

3.5  14-40 m IRS Radial Strip Spectroscopy  (Second-Generation AOT)

We will obtain 1'-wide spectral maps for all galaxies from 14.2 to 40 m using the IRS low resolution mode, maximally overlapping the SED strip. These strips will extend radially to about 0.55$"R_{25}$", where the average surface brightness is $f_
u(15 mu m) sim$1 MJy/sr (Dale et al. 2000a). The IRS spectral mapping mode is much more efficient than the standard "Step and Stare" mode in covering spatial areas significantly larger than the slit. To minimize the effects of cosmic rays and bad pixels, we will spatially double-sample by stepping across each galaxy's major axis diameter, in a direction parallel to the long-low slits, by half the length of the long-low subslits (see Fig 5). Since the IRS short-low subslits are about one arcmin long and are almost perpendicular to the long-low slits, we can complement this map with short-low slit scans along the long-low scan region, to obtain complete wavelength coverage from 5.3 to 40 $"mu$"m in the nuclear regions (below).

Again using the ISO imaging-based model of Dale et al. (2000b), we find that scan exposure times of 30 seconds give S/N $"ge$" 5 for a brightness $f_
u(15 mu m) sim$1 MJy/sr, for the cirrus-like SEDs in the outer disks. >From the prescription for IRS spectral mapping mode (SIRTF web site documentation), we derive a total clock time of 78.3 hours to observe these major axis strips between 14 and 40 $"mu$"min all 75 galaxies. These times include all overheads, as well as a low accuracy initial peak-up exposure. This will provide pointing good to 2 $^{"primeprime}$", useful for positioning our strips and necessary to ensure adequate alignment of the short-low and long-low scans, observations which may not occur consecutively.

3.6  Targeted Low-Res Spectra (5-14 m)  (Second-Generation AOT)

Low-resolution scan maps at 5-14 m, covering a 0.3'x0.9' region, will be obtained for the centers of each galaxy and the 75 extranuclear regions. Since the galaxy centers will also be covered by the 14-40 m and 52-99 m SED strips, we will have nearly complete spectral coverage of these regions. Moreover, the 55 " sub-slit length will enable useful observations of extended sources, and the serendipitous regions simultaneously covered by the other sub-slit will provide data on the local background and quiescent disk regions. We will use a scan exposure of 14 seconds for the nuclei, and 60 seconds for extranuclear regions. Based on the range of observed mid-infrared surface brightnesses for the ISO Key Project galaxies (~10-300 MJy/sr), we expect signal-to-noise ratios of 5 or higher in the core of each target. Exposure times may be revised during Stage 1, based on detailed examination of IRAS and available ISO maps for each target. We derive a total clock time of 47 hours for these observations, including all overheads.

3.7  Targeted High-Resolution Spectra (10-37 m)

For the targeted high-resolution spectra, we will use the step-and-stare mode AORs, for both the short-high and long-high modules. Most of our targets are spatially extended, with significant emission structure over scales comparable to the sizes of the IRS short-high aperture. Our AORs provide: 1) adequate spatial coverage to obtain accurate line ratios over the same physical region in the sources; 2) S/N $"ge$" 5 in the principal diagnostic lines (below); 3) flexible scheduling and roll angle. Both AORs are 5x3 grids, with five, half-slit-width steps in the dispersion direction (perpendicular to the length of the rectangular slit) and three half-slit-length steps in the cross-dispersion direction (parallel to the length of the rectangular slit), as shown in Fig 5. The short-high AOR covers an area of 15.7"x 23.6", while the long-high AOR covers an area of 33.5" x 44.8". The larger spatial coverage in long-high is necessitated by the larger PSF at the longer wavelengths. Our basic unit of coverage for the nuclei and HII regions is well-matched to the area covered by the short-high slit. We expect to achieve 1 $"sigma$" noise levels for unresolved emission lines of approximately $1-2 	imes 10^{-18}$ W/m$^"2$" in the short-high and long-high data, except for the region beyond about 32 $"mu$"m which may be a factor of 2-4 noisier. Thus, emission lines at the $10-20 	imes
10^{-18}$ W/m$^"2$" level, similar to what we expect for [NeII] and [SIII], should be detected at S/N $"ge$" 5-10 over the entire mapped area, and higher S/N in the centers of the scanned regions, where we benefit from aperture oversampling.

Because we are mapping, an IRS low accuracy peak-up on the galaxy nucleus provides sufficiently accurate pointing control. The peak-ups will also provide us with deep, 15 m images of the same areas and they will be a valuable addition to our dataset. We estimate that we will achieve a 1 $"sigma$" point-source sensitivity in these images of 50-100 Jy at 15 m.

Our observing plan is summarized in the table below.

Instrument/Mode Bands Targets Time
  (microns)   (hours)
IRAC 3.6, 4.5, 5.8, 8.0 75 45
MIPS 24, 70, 160 75 124
Total Imaging     169
MIPS SED 52 - 99 75 77
IRS SED 14 - 40 75 78
Total SED     155
IRS Low-Short 5.3 - 14.2 150 47
IRS High-Res 10 - 37 150 141
Total Spectra     188
Grand Total     512

3.8  Summary

Robustness Against Variations in Observatory Performance: We have attempted to design the observations to be relatively flexible in terms of available spacecraft roll angles, and have distributed targets over the sky to avoid sensitivity on spacecraft launch date. If significant degradations in instrument performance occur after launch, we would modify the observing strategy in one or more of the following ways:

IRS High-Res: reduce number of spectroscopic targets
IRS Low-Res or MIPS SED: reduce area of SED radial strips
IRAC or MIPS: map with reduced S/N or reduce galaxy sample (in extreme case).

Validation of Observing Strategy: Three of our highest-priority galaxies (M51, M81, NGC 6946) have long visibility windows and at least one of the three is visible at any time of the year. Depending on launch date we will request that one or more of these galaxies be observed in all of the first-generation AOT modes. The pipeline-processed data will be analyzed immediately to confirm the robustness of the observing plan, and these data will form part of our project's early release observations.





4. REFERENCES


Alonso-Herrero, A., Engelbracht, C.W,. Rieke, M.J., Rieke, G., Quillen, A.C. 2000, ApJ, in press
Alton, P.B. et al. 1998, A&A, 335, 807
Beichman, C.A. 1987, ARAA, 25, 521
Bell, E., & Kennicutt, R.C. 2000, ApJ, submitted
Boselli, A. et al. 1997, A&A, 324, L13
Boselli, A. et al. 1998, A&A, 372, 38
Burstein, D. 1994, Nature, 372, 38
Calzetti, D., Kinney, A.L., & Storchi-Bergmann, T. 1994, ApJ, 429, 582
Condon, J.J. 1992, ARAA, 30, 575
Dale, D.A. et al. 1999, AJ, 118, 2055
Dale, D.A. et al. 2000a, AJ, 120, 583
Dale, D.A. et al. 2000b, ApJ, submitted
Draine, B.T., & Bertoldi, F. 2000, in "H2 from Space", eds. F. Combes & G. Pineau des Forets, Cambridge University Press, in press
Draine, B.T., Roberge, W. G., & Dalgarno, A. 1983, ApJ, 264, 485
Engargiola, G. 1991, ApJS, 76, 875
Giovanelli, R. et al. 1994, AJ, 107, 2036
Goodfrooij, P. & de Jong, T. 1995, A&A, 298, 784
Gordon, K. D., Misselt, K. A., Witt, A. N., & Clayton, G.C. 2000, ApJ, submitted
Heckman, T.M. 1999, Ap&SS, 266, 3
Helou, G. 1986, ApJ, 311, L33
Helou, G., & Bicay, M. 1993, APJ, 415, 93
Helou, G., Lu, N.Y., Werner, M.W., Malhotra, S., & Silbermann, N.A. 2000, ApJ, in press
Ho, L.C., Filippenko, A.V., & Sargent, W.L.W. 1997, ApJ, 487, 568
Hollenbach, D.J., & McKee, C. F. 1989, ApJ, 342, 306
Hollenbach, D.J., & Tielens, A.G.G.M. 1999, Rev Mod Phys, 71, 173
Kaufman, M., Wolfire, M., Hollenbach, D., & Luhman, M. 1999, ApJ, 527, 795
Kennicutt, R.C. 1983, ApJ, 272, 54
Kennicutt, R.C. 1992, ApJS, 79, 255
Kennicutt, R.C. 1998a, ARAA, 36, 189
Kennicutt, R.C. 1998b, ApJ, 498, 541
Kennicutt, R.C., Bresolin, F., French, H., & Martin, P. 2000, ApJ, 537, 589
Kennicutt, R.C., Keel, W.,C., & Blaha, C. A. 1989, AJ, 97, 102
Kwan, J., & Xie, S. 1992, ApJ, 398, 105
Leitherer, C. et al. 1999, ApJS, 123, 3
Li, D., Goldsmith, P., & Xie, T. 1999, ApJ, 522, 897
Lilly, S. et al. 1999, ApJ, 518, 641
Lonsdale, C. J., & Helou, G. 1987, ApJ, 314, 513
Luhman, M.,L., Jaffe, D., Kellar, L., & Pak, S. 1994, ApJ, 436, L185
Malhotra, S., Kaufman, M., Hollenbach, D., & Helou, G. et al. 2000, ApJ, submitted
Merluzzi, P. 1998, A&A, 338, 807
Mobasher, B., Cram, L., Georgakakis, A., & Hopkins, A. 1999, MNRAS, 308, 45
Nelson, A.E., Zaritsky, D., & Cutri, R.M. 1998, AJ, 115, 2273
Rieke, G.H., Loken, K., Rieke, M.J., & Tamblyn, P. 1993, ApJ, 412, 99
Roche, P., Aitken, D., Smith ,C., & Ward, M. 1991, MNRAS, 248, 606
Roussel, H. et al. 1999, in "The Universe as Seen by ISO", eds. P. Cox & M. Kessler, ESA-SP 427.
Rowan-Robinson, M. & Crawford, J. 1989, MNRAS, 238, 523
Sturm, E. et al. 2000, A&A, 358, 481
Thornley, M.D. et al. 2000, ApJ, 539, 641
Tully, R.B. 1987, ApJ, 321, 280
Valentijn, E.A. 1994, MNRA, 266, 614
Xu, C, & Helou, G. 1996a, ApJ, 456, 152
Zaritsky, D., Kennicutt, R.C., & Huchra, J. 1994, ApJ, 420, 87