The Spitzer Survey of Stellar Structure in Galaxies ()

Of the 2331 galaxies in the sample, 597 had some existing data at 3.6 and 4.5 μm from the Spitzer cryogenic mission; 125 of these are part of the LVL survey, and 56 are from the SINGS survey, both of which used the same observing strategy (§ 4) that we employ for . Almost all of the archival maps have at least 240 s of integration time per pixel and are sufficiently deep. The only exceptions are NGC 5457 (96 s); NGC 0470 and NGC 0474 (150 s); and NGC 5218, NGC 5216, and NGC 5576 (192 s).

In terms of area, 82 of the archival galaxies were mapped between 1.0 and 1.5 × D₂₅, and 43 others were mapped to less than 1.0 × D₂₅. While outer disk science for these 125 galaxies would benefit from an extended map, they represent only a small fraction of the total sample, and the repetition of observations would have yielded only an incremental scientific return. Therefore, we decided not to repeat observations of any of the archival galaxies. We also decided not to map the SMC and the LMC, which meet our selection criteria but are very large and are being observed as part of other GO programs. M33, the other commonly studied Local Group galaxy, was observed during the cryogenic mission and is included in the archival part of our sample. All other galaxies that were found via the HyperLEDA database search in 2007 September are observed (see the earlier note about the modest additions to the HyperLEDA database in the last three years). Although we do not repeat observations of any of the archival galaxies, we are processing all galaxies in the sample, except the SMC and LMC, through the same reduction and analysis pipelines described in § 5.

The purpose of Pipeline 1 (P1) is to create science-ready mosaics from the two visits of each target. To do this, P1 performs two major steps. First, it matches the background level in the individual images to account for drifts in the zero point of the amplifiers. Second, it creates the final mosaic using the Drizzle package of IRAF.²⁶

To match the background levels in the individual images we find the regions of overlap between all pairs of overlapping images. Typically, we require that the overlap region contain 20,000 pixels. For all of the new post–cryogenic mission observations (and archival observations that used the same observing strategies), such an overlap is always available. However, when reducing other archival data with less overlap between the frames, we reduce the number of pixels in the overlap region to as low as 5000 pixels—this still provides an adequate number of pixels for estimating the background.

Within these regions we determine the brightness level of the 20th percentile pixel. We use the 20th percentile brightness level to better avoid detector artifacts such as muxbleed, which can affect even the median. By using the 20th percentile brightness level we have a better chance of using the true sky level. We then perform a least-squares solution using the brightness differences between all the pairs of overlapping images, minimizing the residual background difference. Since we use the difference, we need to use the background level in the first image to set the zero point of the solution. We then make corrected images by adding the solved-for corrections and use these images in the formation of the final mosaic. We create the final mosaic following the standard prescription of the Space Telescope Science Data Analysis System (STSDAS) dither package. This method removes the cosmic rays by first forming a mosaic from a median combination of the images and then comparing the individual images with the median mosaic and flagging any cosmic rays. The final mosaic is formed by drizzling the individual images. The resulting mosaic has a pixel scale of 0.75'', and we correct for the change in pixel size to keep the units of MJy sr^-1.

The relative astrometry of the mosaicked images is excellent, because the Spitzer pipeline already incorporates the 2MASS positions to update its astrometry for each of the basic calibrated data frames. The pipeline therefore does not require the type of cross-correlation that was needed previously for the SINGS pipeline to align overlapping fields. The relative astrometric accuracy is less than 0.1''.

The point-spread function (PSF) for IRAC is asymmetric and depends on the spacecraft orientation angle. Pipeline 1 images are produced by combining two different visits to the galaxy, and we therefore estimate the PSF by combining stars in the foreground and background for six typical galaxies to create a "super"-PSF, which has a typical FWHM of 1.7'' and 1.6'' at 3.6 μm and 4.5 μm, respectively. For the most distant galaxies in our survey, at a distance of ∼40 Mpc, this resolution corresponds to a linear scale of ∼300 pc. At the median distance of the galaxies in this survey of 21.6 Mpc, this corresponds to a linear scale of ∼170 pc.

After the science-ready images are produced by the pipeline, we generate masks for point sources using SExtractor (Bertin & Arnouts 1996) for both channels. Each mask is checked by eye and iterated to identify any point source otherwise missed by SExtractor. We also unmask any region of the galaxy that may be have been incorrectly identified by SExtractor. The images with their corresponding masks are then run through the third part of the pipeline (Pipeline 3, P3 hereafter).

The first step in P3 is to determine the local sky level around each galaxy. We want to be sufficiently far from the galaxy to ensure that there is no contamination from the galaxy, but not so far that the background is not truly local. We do this by computing the median sky value in two concentric elliptical annuli centered at the position of the galaxy. Typically, we start the inner annulus at a distance of 1.5 × D₂₅. This annulus is divided in azimuth into 45 regions (or sky boxes). Each of these regions is then grown outward until it encompasses a total of 4000 nonmasked pixels,²⁷ as shown in Figure 4. The second annulus begins at the end of the first annulus and the same process is repeated. We compare the measured sky values in the two annuli for any radial gradients that might be indicative of light contamination from the galaxy. If we find that there is contamination, then we move the inner radius outward and repeat the process until we ensure that we are sampling the local sky.

Zoom In Zoom Out Reset image size

In our analysis so far, the inner radius of the inner annulus has varied between 1.2 and 2 × D₂₅. In all cases for the warm mission data and a majority of the archival data, it has been sufficient to set the inner radius of the annulus to either 1.5 or 2 × D₂₅. However, in a few cases where the archival data are shallow and the galaxy is relatively blue, we moved the inner radius to 1.2 × D₂₅ to be immediately outside the galaxy to get as accurate a measurement possible for the local sky level. Automating this procedure is nontrivial, because variations at these faint levels (μ_3.6μm > 27(AB) mag arcsec^-2) can be from a lopsided galaxy, debris or tidal structures, or variation in the background. Therefore, we examine every galaxy by eye to ensure that contamination from the galaxy light to the measurement of the sky background is as low as possible and move the radii outward as needed.

The sky background computation method for three of the galaxies in the sample (NGC 0337, NGC 4450, and NGC 4579) is shown in Figure 4. In the central column of this figure, we show an example case for NGC 4450, where the median sky in some of the boxes is affected by the bleeding of flux from a (masked) but saturated nearby bright field star. In such cases we ignore the affected pixels or move the annuli further out to get the best estimate for the background. Finally, we use the standard deviation obtained from the 90 sky boxes with the average standard deviation within each box to estimate the contribution of the low-frequency flat-fielding errors and possible additions to the total error budget in the background determination.

The typical sky brightness level is fainter than 26(AB) mag arcsec^-2 at 3.6 and 4.5 μm. The Poisson noise from the galaxy dominates the total error budget in the central regions, while errors in the sky background are the main source of uncertainty in the outer parts. Large-scale background errors can be particularly significant in galaxies with large apparent sizes. However, in the fainter outer parts, the noise is significantly reduced from azimuthally averaging over a large number of pixels in the outer galaxy—we are therefore able to measure the galaxy profiles well below the typical sky brightness. The typical S/N for a galaxy at the canonical μ_3.6μ = 26.5(AB) mag arcsec^-2 is greater than 3.

The images resulting from Pipeline 1 are calibrated in units of MJy sr^-1. The conversion from MJy sr^-1 to AB magnitudes is such that the zero point to convert fluxes in jansky to magnitudes does not vary with wavelength (Oke 1974):

>From this definition one can derive the corresponding expression for surface brightness:

5.3.1. P3: Measuring the Galaxy Host Properties

After the sky level has been properly measured, we use the IRAF routine ellipse to determine the radial profiles of intensity, surface brightness, ellipticity, and position angle for all galaxies in both bands. The center is determined using imcenter and kept fixed during the fit, whereas the ellipticity and position angle are left as free parameters. We generate profiles with two different radial resolutions, by incrementing the semimajor axis in 2'' and 6'' at each step, respectively. The latter profiles have a coarser resolution and a better S/N—these are used to measure the RC3-like parameters. The low- and high-frequency sky background errors described previously, together with the Poisson noise of the source along each isophote, are considered together in the final error budget of the surface photometry at each radius (see Gil de Paz et al. 2005; Muñoz-Mateos et al. 2009). The errors in the position angle and ellipticity at each radius are computed by the ellipse task from the rms of the pixel values along each fitted isophote. The corresponding errors for the values of these geometrical parameters at 25.5 and 26.5(AB) mag arcsec^-2 are interpolated between the corresponding adjacent values. The error in the semimajor axis is computed from the change in the radius affected by moving the surface brightness profile by the measured errors and an additional 10% error added in quadrature to the measured errors. The 10% error is an estimate of the aperture correction uncertainty as estimated by the Spitzer Science Center instrument team. We compute the error in the semimajor axis for each galaxy. Sample results for NGC 0337 and NGC 4579 are shown in Figure 4 and noted in Table 2. For NGC 4450, the scattered light from a background star strongly affects the computation of the surface brightness error and therefore the error in the semimajor axis is unrealistic. In such cases, the error in the semimajor axis can be assumed to be the median semimajor axis error in the sample. The maximum error in the determination of the semimajor axis is 6'', which is the step size of the coarse-resolution fitting procedure.

In Figures 5, 6, and 7 we show three sample sets of profiles for three galaxies in the sample: NGC 0337 (SBd), NGC 4450 (SAab), and NGC 4579 (SABb). The basic RC3-like galaxy properties (semimajor axis, axial ratio, and position angle) are tabulated at the levels of 25.5 and 26.5(AB) mag arcsec^-2 in both the 3.6 μm and the 4.5 μm images and are quoted in Table 2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Along with the surface photometry we also measure the curve of growth (e.g., Muñoz-Mateos et al. 2009; Gil de Paz et al. 2007) with the integrated flux up to each radius. By fitting the accumulated magnitude as a function of the magnitude gradient at each point we obtain the asymptotic magnitude as the y-intercept of that fit (see Gil de Paz et al. 2007). Once the total magnitude is known, it is straightforward to locate the radii containing a given percentage of the total galaxy luminosity, from which concentration indices can be determined. In Table 3 we quote the asymptotic magnitudes of the three sample galaxies considered here, together with the C₃₁ (de Vaucouleurs et al. 1977) and C₄₂ (Kent 1985) concentration indices.

In Pipeline 4, we decompose the two-dimensional stellar distribution in each galaxy into different subcomponents using the version 3 of GALFIT (Peng et al. 2002, 2010). GALFIT is a parametric fitting algorithm that offers a large flexibility on the fitted models; the optimal solution for the model parameters is found with the Levenberg-Marquardt algorithm, performing nonlinear least-squares χ² minimization of the difference between observed and model images. For our pipeline, we could have also used bulge/disk decomposition analysis (BUDDA; Gadotti 2008) and bulge/disk/bar (BDBAR; Laurikainen et al. 2005). We tested all three algorithms using a set of galaxy images and verified that GALFIT produces the same results as BUDDA and BDBAR if the three codes are run with the same set of input images, models, and initial parameters. We chose GALFIT because it is currently the most sophisticated algorithm and also well known by a large portion of the astronomical community. Other studies in which software was developed for similar purposes include Simard (1998), Pignatelli et al. (2006), and Méndez-Abreu et al. (2008).

To simplify the use of GALFIT for a large number of cases, a set of new IDL-based tools has been created (GALFIDL). The tools include automatic creation of the input files for decompositions, containing reasonable first guesses for the initial parameters based on P3 products, and routines for running and visualizing the decompositions. GALFIDL will be available to the community concurrently with the publication of the data and the GALFIT analyses.²⁸

As a standard part of the pipeline analysis, decompositions include bulges and disks (and, if appropriate, bars, nuclear components, and, in some cases, multiple exponential disks). Taking into account that decompositions, particularly at high redshift, are often made in a simpler manner, we also compute the two-component bulge/disk decompositions and one-component Sérsic function fits. A decomposition for NGC 1097 is shown in Figures 8 and 9: the five-component model includes an exponential disk, a Sérsic function for the bulge, and a Ferrers function for the bar. The galaxy has an extremely prominent nuclear ring, which is fit using a Gaussian function, and a nuclear point source using the corresponding PSF. Without deep, almost dust-free, observations (such as those in ), the measurement of bulge structural parameters (such as like shape, size, and profile), might be compromised in some cases if the quantification of underlying structures (such as the disks, bars, and rings) is inaccurate. GALFIT 3.0 allows for even more sophisticated decompositions, which will be made for selected subsamples of galaxies. For example, it is possible to add more components, or one may want to fit the spiral arms, an example of which is shown in Figure 10.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A major area of study for is a quantitative analysis of stellar structures in the faint outer regions of galaxies, where the stellar surface density we trace is extremely low and comparable with, or lower than, the atomic gas reservoir. Deep optical studies have already shown that in a majority of spiral galaxies, the outer disks exhibit a secondary exponential component. Stellar disks appear to extend beyond 1.5 × D₂₅ (e.g., Pohlen & Trujillo 2006; Erwin et al. 2008) with truncated or antitruncated profiles. GALEX (Martin et al. 2005) has found extended UV disks even farther out, where it was assumed only H i gas existed (e.g., Gil de Paz et al. 2005; Thilker et al. 2005, 2007; Muñoz-Mateos et al. 2007; Zaritsky & Christlein 2007). Not only is significantly deeper than the previous optical data, but it also traces the stellar profiles into the H i disk of the galaxy—in over 800 galaxies, we image an area greater than 3D₂₅.

Another observation that has been much debated is the break between the inner and outer exponential profiles (e.g., Bakos et al. 2008; Azzollini et al. 2008). With we are measuring the break radius and comparing it with the disk size and mass. The catalog of these breaks will provide strong quantitative constraints for galaxy evolutionary models. In particular, this catalog will be key to quantifying the possible role of radial stellar migration in shaping the outskirts of disks (Roškar et al. 2008; Sánchez-Blázquez et al. 2009).

In elliptical galaxies, deep optical observations have revealed the presence of faint tidal tail and shell-like structures (e.g., Bennert et al. 2008; Canalizo et al. 2007). contains 188 ellipticals and lenticulars—with this set and the depth, we are quantifying the frequency and mass of shells and other debris features in the outskirts of ellipticals and lenticulars.

Bulges are an inhomogeneous class of objects, and different types have been proposed in the literature (Kormendy & Kennicutt 2004; Athanassoula 2005). Kormendy & Kennicutt (2004) distinguish bulges from pseudobulges, while Athanassoula (2005) distinguishes three categories: classical bulges; boxy/peanut bulges, which are a part of a bar; and disklike bulges, formed out of disk material. Several studies have analyzed their observational properties and compared them with N-body models (e.g. Kuijken & Merrifield 1995; Carollo et al. 1997; Bureau & Freeman 1999; Erwin et al. 2003; Bureau & Athanassoula 2005; Fisher 2006; Drory & Fisher 2007; Fisher & Drory 2008; Gadotti 2009). Even in lenticular galaxies there seems to be evidence for disklike bulges, based on structural decompositions (e.g., Laurikainen et al. 2005, 2007; Gadotti 2008; Graham & Worley 2008) and kinematics (e.g., Peletier et al. 2007; Falcón-Barroso et al. 2006). But until now few studies have analyzed the structural properties of bulges in a large number of galaxies; none have done it in a sample as large and deep as . Moreover, encompasses a range of large-scale structures, allowing us to examine the environmental influences on the formation of bulges. This may have important implications, since massive bulges are expected to form from mergers.

From a preliminary analysis of the data, Buta et al. (2010) are finding that many late-type ellipticals in the RC3 have disks and thus should instead be classified as S0s, consistent with other deep near-infrared studies (Laurikainen et al. 2010). With the data we will examine how common the disk structure is in all types of ellipticals.

Finally, the extremely deep data from allows us to best determine the structural properties of the bulge component—in shallower data, estimates of shape, size, and profile are uncertain when measurements of underlying components such as disks, bars, and rings are imprecise. As described in § 5.4, the project is carrying out detailed decomposition of all the galaxies and expects to compile the most detailed database of bulges of all kinds. The combined advantages of the data (deep and insensitive to dust attenuation) and the uniform reduction pipeline (careful and detailed structural analysis) will allow for a suitable examination of galaxies with composite bulges: i.e., galaxies hosting more than one bulge category (e.g., Erwin et al. 2003; Kormendy & Barentine 2009; Nowak et al. 2010). The spatial resolution of our survey is adequate for resolving typical bulges (our resolution is ∼100 pc at the median distance of the survey volume), but is inadequate for precise measurements of the bulge structure from the inability to resolve compact nuclear structures that are often present at the centers of galaxies. These unresolved point sources affect the values of Sérsic parameters, which, on average, are overestimated in ground-based optical images, compared with Hubble Space Telescope studies (e.g., Carollo et al. 1997; Balcells et al. 2003). We also expect that the reduced dust extinction may reveal triaxial structures in bulges and thus allow us to probe beyond the standard symmetric Sérsic fits.

Stellar bars are dynamically important structures in galaxies. A bar-induced gas flow leads to a number of evolutionary effects from the triggering of circumnuclear starbursts to the buildup of bulges (Scoville & Hersh 1979; Simkin et al. 1980; Zaritsky & Lo 1986; Athanassoula 1992; Friedli & Benz 1993; Martin & Roy 1994; Friedli & Benz 1995; Sakamoto et al. 1999; Sheth et al. 2000, 2002, 2005). The cosmological redshift evolution of the bar fraction is also an important signpost of the growth and dynamic maturity of galaxies (e.g., Abraham et al. 1999; Sheth et al. 2003; Jogee et al. 2004; Elmegreen et al. 2004; Sheth et al. 2008). Whereas the previous smaller and shallower studies gave mixed results for the change in the bar fraction with redshift, Sheth et al. (2008) show that not only does the overall bar fraction in disk galaxies decline significantly over the last seven billion years, but also that the decline depends on the galaxy type. They show that the formation and evolution of bars is strongly correlated with the host galaxy mass, bulge dominance, and presumably their dark-matter halo. The data are critical to establish the precise local frequency of bars and its variation with galaxy type. The data also provide a measurement of the light fraction in the bar component (e.g., Durbala et al. 2008; Gadotti 2008; Weinzirl et al. 2009) and accurate measurements of properties such as the bar length, shape, and ellipticity (e.g., Menéndez-Delmestre et al. 2007; Erwin 2005; Gadotti 2010). Several techniques may be used to derive bar strengths, based on calculating bar torques (Laurikainen & Salo 2002; Buta et al. 2003) and estimating the bar-spiral arm contrast (Elmegreen & Elmegreen 1985). Analysis of the relative Fourier intensity profiles of bars is compared with models of the evolution of barred galaxies (e.g., Athanassoula & Misiriotis 2002; Debattista et al. 2004) to constrain the history of their evolution and interaction with their dark-matter halo (Athanassoula 2003).

Another study of interest is the presence of secondary or nuclear bars that may be responsible for feeding the central black holes and/or stellar clusters (Shlosman et al. 1990). The observations, unaffected by the high and patchy extinction often present in galactic centers, allow us to detect all nuclear bars with sizes of r > 500 pc (our resolution, 2'' ∼ 200 pc at the median distance of ) in galaxies of all types.

Outer rings are large, optically low, surface brightness features that dominate the outer disks of some early-type barred and weakly barred galaxies (see the review by Buta & Combes 1996). The properties of the rings constrain their formation epoch, the dynamical time-scale at their respective radii, and the evolution of the bar pattern speed (e.g., Athanassoula 2003). Finding outer rings around unbarred galaxies is also of great interest, because it would suggest that some unbarred galaxies may in fact be highly evolved former barred galaxies. Whereas ground-based near-IR imaging is rarely deep enough to detect outer rings reliably to study their stellar populations, colors, and other characteristics, the data will allow us to not only detect many new outer rings, but also to make a complete census of these features in normal galaxies and to measure their stellar mass, shape, width, and ellipticity. Likewise, provides an unprecedented view of circumnuclear and inner rings, which are typically found in barred spirals. These rings should have a broad underlying older stellar population background as they evolve, and it is this type of feature we will see in the data.

A systematic study of spiral arm amplitudes as a function of radius at 3.6 and 4.5 μm, free of confusing effects from dust extinction, promises to initiate a new set of spiral arm studies. Although the discrepancies in the classical density wave model were solved with the swing amplification theory (Toomre 1981), there has been very little testing of this theory with observations. Modal studies by Bertin et al. (1989a, 1989b) predict that there should be amplitude variations along the arms from resonances between inward- and outward-moving spiral waves, and the precise amplitude variations have been modeled (Elmegreen & Elmegreen 1984; Elmegreen et al. 1989; Elmegreen & Thomasson 1993; Regan & Elmegreen 1997). Most recently, a new theory has been proposed for the formation of rings and spiral arms, which argues in favor of chaotic orbits, confined by invariant manifolds emanating from the L1 and L2 Lagrangian points of a bar (Romero-Gómez et al. 2007). Several comparisons between the results of this theory and observations have been carried out successfully (Athanassoula et al. 2009a, 2009b) and several more will be possible with the database. will cover the full range of galaxies with varying spiral arm strengths and galactic structures. In Figure 11, we show a preliminary result of the type of analysis that is possible with the data. In this figure we show the arm-interarm contrast measured from the 3.6 μm images for NGC 7793, a flocculent galaxy, in contrast with M51, a grand-design spiral galaxy (see a model for amplitude variation of M51 in Salo & Laurikainen 2000b). The contrasts are measured beyond R₂₅, further than in previous optical studies. The spiral structure traced in the 3.6 μm band is different in detail from the optical images, but still shows weak arms in NGC 7793 and increasingly stronger arms in M51. We will carry out a detailed analysis of spiral arm amplitudes and variations and compare the observational data with the models.

Zoom In Zoom Out Reset image size

For the NGC 5194 (M51) system, there may well be a connection between the spiral structure and the presence of the companion (NGC 5195). Indeed, detailed H i observations by Rots et al. (1990) show the presence of a long tail, which emanates from the west side of the galaxy, running south of the spiral arm visible in the image, and turning up toward the north, further eastward. The H i kinematics are very difficult to model with an interaction involving a single passage, but Salo & Laurikainen (2000a, 2000b) propose a model with a second passage (Fig. 12), which gives a better fit to some of the aspects of the observations. The inner spiral structure in the main spiral is then amplified by the action of the companion, and wave interference causes amplitude modulations, which are well reproduced in the observations (Fig. 13). This shows the interest of the data for detailed modeling of the dynamics of individual galaxies.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

contains 188 galaxies of T-type < 0 and therefore constitutes one of the largest and most homogeneous mid-infrared data sets for early-type galaxies. The shapes of ellipticals and lenticulars may reflect different evolutionary paths, with wet mergers leading to disky shapes and dry mergers leading to boxy profiles (Naab et al. 2006; Pasquali et al. 2007; Kang et al. 2007). In addition, Kormendy et al. (2009) show that deep imaging can reveal departures from the Sérsic profiles in elliptical galaxies, which are diagnostics of their formation. They found a dichotomy in which ellipticals that have cuspy cores at their inner radii can be separated from those that show an excess of light at the center. The data offer us a unique opportunity to revisit the frequency of the different shapes and profiles as a function of stellar mass and environment, free of dust obscuration effects.

Although dwarfs are the most abundant type of galaxy in the universe, their low surface brightnesses of typically μ_B > 22 mag arcsec^-2 have restricted detailed observations of these stellar systems. In particular, the characterization of the underlying stellar structure in these faint galaxies still remains largely unexplored. Using an absolute magnitude criterion of M_B > -17, we have identified a preliminary sample of 206 dwarf galaxies in . The sample includes 87 late-type DDO dwarfs previously unobserved with Spitzer, to complement the 48 that have been observed. The dwarfs therefore constitute a representative sample of nearby dwarf galaxies in which to study the properties of the underlying stellar component.

Dwarf galaxies in allow for a detailed analysis of the underlying stellar disk in both the 3.6 and 4.5 μm data. By deriving the radial scale lengths for these galaxies, more insight can be gained into the evolutionary link between dwarfs and their giant counterparts. At the same time, the relatively large sample of dwarfs in gives a unique opportunity to look at how these parameters may vary with environment. We can also derive stellar masses for the dwarfs that will be used to assess their stellar mass to dark-matter ratio and will likely provide crucial evidence to the long-standing debate of whether these galaxies are truly dark-matter-dominated systems (see, e.g., Strigari et al. 2008).

In addition to these areas, there are at least three other main areas of astrophysics that these data will address uniquely: the reported absence of bulges in some disk galaxies, the vertical stellar structure, and the Tully-Fisher relationship. There is also likely to be significant amounts of spin-off science for with its large and accurate inventory of mass and galactic structure in the nearby universe: e.g., the relationship between the Tully-Fisher relationship and properties of the fundamental plane (e.g., Zaritsky et al. 2008), dust heating in outer disks, the interaction/merger fraction from a study of the debris around these galaxies, and diagnosis of AGN activity from tracing hot dust very close to the AGN from the 3.6–4.5 color.