AN ATLAS OF GALAXY SPECTRAL ENERGY DISTRIBUTIONS FROM THE ULTRAVIOLET TO THE MID-INFRARED

Templates and models of galaxy spectral energy distributions (SEDs) are often essential for deriving the physical properties of galaxies from observables. K-corrections, where rest-frame colors and luminosities are derived from observed galaxy photometry, rely on galaxy SED templates and models (e.g., Oke & Sandage 1968; Pence 1976; Coleman et al. 1980). Photometric redshifts rely on accurate models of the relationship between observed galaxy photometry and redshift. While this relationship can be modeled empirically for some galaxies using polynomial fits and neural networks (e.g., Connolly et al. 1995; Firth et al. 2003), models and templates are required for photometric redshifts of faint and high redshift galaxies.

The most extensively used galaxy SED template libraries are those of Coleman et al. (1980) and Kinney et al. (1996). These SEDs have proved exceptionally useful over several decades, but have understandable limitations. The wavelength coverage of Coleman et al. (1980) and Kinney et al. (1996) templates is limited to the ultraviolet and optical, so they are often extended into the infrared with stellar population synthesis models (e.g., Bruzual & Charlot 2003). The Coleman et al. (1980) and Kinney et al. (1996) spectra are often of galaxy nuclei, which may not be representative of integrated galaxy spectra. These templates also predate the availability of precisely calibrated wide-field imaging, which can be used to normalize and verify SEDs. Without such imaging, there is the potential for systematic error, especially when combining spectra from different instruments using different extraction apertures.

With the advent of imaging and spectroscopy from the Infrared Space Observatory (Kessler et al. 1996), Spitzer Space Telescope (Werner et al. 2004), Akari (Astro-F; Murakami et al. 2007), and the Wide-field Infrared Space Explorer (WISE; Wright et al. 2010), it became both important and possible to extend galaxy SED templates into the mid-infrared (e.g., Chary & Elbaz 2001; Polletta et al. 2007; Rieke et al. 2009). These templates include emission from dust, silicate absorption, and features attributed to polycyclic aromatic hydrocarbons (PAHs), which dominate the mid-infrared spectra of star-forming galaxies. However, large aperture spectrophotometry over a broad wavelength range remained unavailable (or unutilized), and consequently many of these templates rely upon model galaxy spectra rather than observed galaxy spectra.

Galaxy SEDs can be modeled using models of stellar populations, nebular emission lines, dust obscuration, and dust emission (e.g., Tinsley 1968; Silva et al. 1998; Charlot & Longhetti 2001; Bruzual & Charlot 2003; Maraston 2005; da Cunha et al. 2008; Vega et al. 2008; Pacifici et al. 2012). The models have improved considerably over the decades, with advances in relevant theory and stellar libraries, combined with better constraints from observed galaxy spectra and photometry. When provided with sufficiently precise photometry with broad wavelength coverage, galaxy SED models are able to do a remarkably good job of reproducing the spectra of some galaxies (see Section 3).

Unfortunately, SED models with a large number of free parameters cannot be expected to reproduce the SED of a galaxy with a complex star formation history when only limited photometry is available (e.g., Pforr et al. 2012). Similarly, SED models are not always the best means of predicting the observed colors of galaxies as a function of redshift, as models can generate unrealistic spectra. These issues also apply (albeit to a lesser extent) to modeling SEDs with combinations of empirical component spectra (e.g., Blanton & Roweis 2007; Assef et al. 2008). This being the case, there is still a role for galaxy SED templates derived from real galaxies.

To address this need, we present an atlas of 129 galaxy SEDs with wavelength coverage spanning from the ultraviolet to the mid-infrared. Relative to the prior literature, our atlas spans a broader range of absolute magnitudes (−14.7  >  M_g  >   − 23.2), colors (0.1  <  u  −  g  <  1.9) and galaxy types, including ellipticals, spirals, merging galaxies, blue compact dwarfs, and luminous infrared galaxies. Our templates make use of optical drift-scan spectroscopy from ground-based telescopes and mid-infrared spectroscopy from Spitzer and Akari, and these spectra were scaled and verified using matched-aperture photometry in 26 passbands. In Figure 1 we present one of the SEDs along with color composite images of the Galaxy Evolution Explorer (GALEX), Sloan Digital Sky Survey (SDSS), and Spitzer imaging used to constrain and verify the SED.

Zoom In Zoom Out Reset image size

The structure of our paper is as follows. In Section 2 we describe the galaxy sample, spectra, and photometry (all photometry presented in this paper uses the AB magnitude system). In Section 3 we describe how we compiled the SEDs, including filling gaps in spectral coverage with models and power laws. We compare our spectra against previous template libraries and the observed galaxy colors in Section 4, and summarize our major conclusions in Section 5. In Appendix A we introduce an illustrative subsample of 18 galaxies, and in Appendix B we present SEDs and optical images for the entire sample, all of which are available online.¹⁵

Our sample consists of z  <  0.05 galaxies with archival drift-scan optical spectroscopy, all of which have well calibrated multi-wavelength imaging spanning from the ultraviolet to the mid-infrared. Almost all galaxies in our sample have spectra from the Spitzer Space Telescope's Infrared Spectrograph (IRS; Houck et al. 2004), with the exceptions being elliptical galaxies with (approximately) Rayleigh–Jeans spectra in the mid-infrared. The bulk of the optical spectra are taken from Moustakas & Kennicutt (2006) and Moustakas et al. (2010), who used the Bok 2.3 m and CTIO 1.5 m telescopes, with some additional spectra of elliptical galaxies taken from Kennicutt (1992) and Gavazzi et al. (2004). Akari Infrared Camera (IRC; Onaka et al. 2007) spectroscopy is used when it captures a significant fraction of the total galaxy flux and has high signal to noise. Accurate matched-aperture photometry from the ultraviolet to the mid-infrared was measured with archival imaging from the GALEX (Morrissey et al. 2007), Swift UV/optical monitor telescope (UVOT; Roming et al. 2005), SDSS III (Aihara et al. 2011), Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006), Spitzer Space Telescope (Fazio et al. 2004; Rieke et al. 2004), and WISE (Wright et al. 2010).

Sample selection was set by the availability of suitable spectra and images, which fortunately results in a very diverse (but not necessarily representative) sample of 129 nearby galaxies. Our requirement for multi-wavelength images and spectroscopy results in our sample (largely) being a subset of other surveys of the nearby universe. For example, our sample size is smaller than that of Moustakas & Kennicutt (2006), who presented optical drift-scan spectroscopy of 417 galaxies, and our sample size is smaller than the 202 LIRGs and ULIRGs studied by the Great Observatories All-sky LIRG Survey (GOALS; Armus et al. 2009). Readers may find that the utility of our SED atlas is increased by combining it with detailed studies of nearby galaxies (many of which use data that we have now incorporated into this paper), including measurements of star formation rates (e.g., Moustakas et al. 2006; Kennicutt et al. 2009), fitting or modeling the mid-infrared spectra (e.g., Smith et al. 2007b; Inami et al. 2013; Stierwalt et al. 2013), and estimates of the contribution of active galactic nuclei (AGNs) to mid-infrared emission (Petric et al. 2011). In Table 1 we summarize the basic properties of our sample, including names, coordinates, morphologies, magnitudes, and redshifts.

2.1. Photometry

Throughout this paper we use AB apparent magnitudes that are defined by

$$\begin{eqnarray} &&m = -2.5 \log \left[ \left(\int R(\nu) \frac{f_\nu (\nu)}{h\nu } d\nu \right) \times \left(\int R(\nu) \frac{g_\nu (\nu)}{h\nu } d\nu \right)^{-1} \right]\nonumber\\ \end{eqnarray} \tag{ 1 }$$

(e.g., Hogg et al. 2002), where f_ν is the SED of the source, g_ν is the SED of a source that is 3631 Jy at all wavelengths (i.e., m = 0), R_ν is the filter response function (defined as electrons per incident photon), and hν is the energy of a photon with frequency ν. Please note that these magnitudes are based on photon counts rather than fluxes, and do not simply correspond to monochromatic flux densities at the effective wavelengths of the relevant filters.

The 2MASS and Multiband Imaging Photometry for SIRTF (MIPS) filter curves (available from their respective archives) are defined as electrons per f_λ (then renormalized to peak at 1.00), so we modified these curves by dividing the filter curves by λ. The MIPS zero-points provided with the archival imaging are defined using a flux density at a specific wavelength and a spectrum where the flux density varies with wavelength (i.e., a Rayleigh–Jeans spectrum). For these data, we applied corrections to the zero-points so the resulting photometry is defined by Equation (1). Photometry originally defined using the Vega magnitude system (e.g., 2MASS) has been shifted into the AB magnitude system.¹⁶ The SDSS calibration is not quite on the AB magnitude system, so we added corrections of −0.04 and +0.02 mag to the u and z photometry, respectively, to shift it back to the AB system.¹⁷

Foreground dust extinction was modeled using the Planck dust extinction maps (Planck Collaboration et al. 2011, 2013). We use the Fitzpatrick (1999) model of dust extinction as a function of wavelength, with a modified parameterization for c₄ (Peek 2013),

$$\begin{equation} c_4=4.64-11.64 \times E(B-V). \end{equation} \tag{ 2 }$$

This accounts for the enhanced ultraviolet extinction observed at high Galactic latitudes (Peek & Schiminovich 2013) relative to the expectation of the Fitzpatrick (1999) extinction curve.

Accurate matched-aperture photometry was essential for normalizing and verifying our spectra, and for constraining models used to fill gaps in our spectral coverage. Where possible, we made use of archival imaging that is readily available online. Ultraviolet imaging was taken from GALEX Release 6 (GR6) and the Swift UVOT (Roming et al. 2005), with the latter often requiring stacking of individual exposures using Swarp (Bertin et al. 2002). While some galaxies have ultraviolet imaging from the XMM-Newton optical monitor (Mason et al. 2001), these images often suffer from scattered light and were not used in the analyses presented here. The only sample galaxy without GALEX or Swift imaging is UGCA 410 (a blue compact dwarf with weak PAH emission) where we used 20''  ×  10'' elliptical aperture fluxes from the International Ultraviolet Explorer (Kinney et al. 1993) to measure an approximate total magnitude.

All galaxies in the sample have optical imaging from SDSS III (Aihara et al. 2011), near-infrared imaging from 2MASS (Skrutskie et al. 2006), and mid-infrared imaging from WISE. When galaxies overlapped the SDSS III or 2MASS image boundaries, we used mosaic images from the NASA-Sloan Atlas (Blanton et al. 2011), the 2MASS Large Galaxy Atlas (Jarrett et al. 2003), or generated new Swarp mosaics (Bertin et al. 2002). For the WISE bands, we used release version 4.1 atlas images from the WISE All Sky Data Release. When available, we utilized Spitzer Infrared Array Camera (IRAC; Fazio et al. 2004), IRS, and MIPS (Rieke et al. 2004) mosaics available from the Spitzer Infrared Nearby Galaxies Survey (SINGS; Kennicutt et al. 2003) or the Spitzer Heritage Archive.

For a small number of galaxies we recombined individual exposures to improve signal to noise and cosmic-ray rejection. In some instances, where multiple images were available for the same galaxy in the same band (e.g., GALEX NUV images from two surveys), we measured photometry using both images to check consistency. The background of the images was estimated using the median of the pixel values that are outside of the aperture. The background in the 2MASS and Spitzer images can vary as a function of position, so we visually inspected all of the images, and for those images with significant background variations we subtracted a model background determined with IRAF's imsurfit task.

We defined our photometric aperture to (where possible) approximate the rectangular apertures used by Kennicutt (1992), Gavazzi et al. (2004), Moustakas & Kennicutt (2006), and Moustakas et al. (2010) for the optical drift-scan spectrophotometry. In some instances the photometric aperture was reduced in size to mitigate contamination from neighboring sources or improve signal to noise at some wavelengths (particularly the ultraviolet and mid-infrared). The photometric apertures used for each of the galaxies are defined in Table 2. Bright stars are manually flagged, masked, and not included in the aperture photometry presented in this paper. As the masked regions are very small relative to the large apertures used to measure galaxy photometry, we did not attempt to model the galaxy light within the masked regions. For some galaxies the aperture only covers a fraction of the galaxy, and the fraction of light captured in the g band can be estimated by comparing (approximate) total and aperture magnitudes presented in Tables 1 and 3, respectively. However, for more than 70% of our sample the g-band aperture magnitudes are within 0.3 mag or brighter than Sérsic model fit magnitudes provided by the NASA-Sloan Atlas. In this paper we largely use photometry corrected for foreground dust extinction, but in Table 3 the photometry is presented without dust extinction corrections (as the corrections are model dependent). The modeled Milky Way dust extinction for each sample galaxy and band is presented in Table 4.

Table 2. Photometric and Spectroscopic Apertures

Name	Photometric Aperture		Optical	Akari Aperture		Spitzer IRS SL Aperture		Spitzer IRS LL Aperture
	Size	P.A.	Spectrum	Size	P.A.	Size	P.A.	Size	P.A.
	(arcsec)	(deg)		(arcsec)	(deg)	(arcsec)	(deg)	(arcsec)	(deg)
Arp 256 N	40 × 60	90	M06	...	...	33.3 × 18.5	154	40.6 × 40.6	71
Arp 256 S	40 × 40	90	M06	29.2 × 60.0	246	25.9 × 18.5	153	61.0 × 25.4	69
NGC 0337	95 × 55	70	M10	...	...	53.7 × 31.5	149	121.9 × 71.1	65
CGCG 436−030	35 × 40	90	M06	7.3 × 60.0	248	36 slit	−28	106 slit	67
NGC 0474	60 × 150	90	M06	...	...	...	...	...	...
NGC 0520	140 × 100	90	M06	29.2 × 60.0	248	75.8 × 57.4	−27	106.7 × 96.5	−111
NGC 0584	138 × 55	70	M10	...	...	27.8 × 18.5	148	40.6 × 35.6	64
NGC 0628	346 × 55	70	M10	...	...	48.1 × 27.7	152	330.2 × 50.8	68
NGC 0660	130 × 120	90	M06	...	...	36 slit	151	106 slit	248
III Zw 035	20 × 35	90	M06	29.2 × 60.0	69	20.0 × 35.0	−26	50.8 × 40.6	−110
NGC 0695	60 × 45	90	M06	25.9 × 60.0	249	36 slit	−24	106 slit	71
NGC 0750	80 × 30	90	M06	...	...	...	...	...	...
NGC 0855	86 × 55	70	M10	...	...	24.0 × 22.2	156	116.9 × 45.7	73
NGC 1068	155 × 180	90	M06	26.1 × 5.0	251	57.3 × 24.1	−31	106.7 × 25.4	−115
Arp 118	95 × 60	90	M06	...	...	40.7 × 24.1	−31	76.2 × 25.4	−115
NGC 1144	50 × 60	90	M06	...	...	40.7 × 24.1	−31	76.2 × 25.4	−115
NGC 1275	75 × 40	90	M06	29.2 × 60.0	254	40.7 × 18.5	−22	76.2 × 25.4	−106
NGC 1614	80 × 60	90	M06	7.3 × 60.0	259	36 slit	−28	106 slit	68
NGC 2388	60 × 30	90	M06	7.3 × 60.0	−82	38.9 × 31.4	−174	81.3 × 30.5	102
NGC 2403	657 × 56	127	M10	...	...	129.5 × 59.2	31	549.3 × 50.8	127
NGC 2537	100 × 60	90	M06	...	...	36 slit	180	106 slit	277
NGC 2623	90 × 40	60	M06	7.3 × 60.0	104	40.7 × 40.7	11	61.0 × 55.9	−73
IRAS 08572+3915	30 × 20	90	M06	29.2 × 60.0	−72	36 slit	10	106 slit	107
UGC 04881	60 × 40	90	M06	29.2 × 60.0	110	36 slit	197	106 slit	294
NGC 2798	84 × 55	103	M10	...	...	53.7 × 31.4	23	106.7 × 55.9	120
UGCA 166	20 × 20	90	M06	...	...	36 slit	176	106 slit	272
UGC 05101	50 × 30	90	M06	29.2 × 60.0	116	36 slit	45	106 slit	142
NGC 3049	72 × 55	115	M10	...	...	44.4 × 24.1	15	101.6 × 50.8	112
NGC 3079	100 × 330	90	M06	...	...	53.6 × 24.0	−160	116.8 × 25.4	116
UGCA 208	30 × 20	90	M06	...	...	36 slit	40	106 slit	137
NGC 3190	144 × 55	115	M10	...	...	51.8 × 31.5	19	203.2 × 50.8	115
NGC 3198	180 × 55	120	M10	...	...	48.1 × 29.6	36	203.2 × 55.9	133
NGC 3265	42 × 55	120	M10	...	...	51.8 × 27.8	22	71.1 × 50.8	119
Mrk 33	33 × 55	110	M10	17.5 × 3.0	−60	38.9 × 27.7	42	38.9 × 27.7	42
NGC 3310	90 × 65	90	M10	...	...	36 slit	38	106 slit	135
NGC 3351	245 × 55	115	M10	...	...	53.7 × 35.2	18	238.7 × 55.9	114
NGC 3379	180 × 120	90	K92	...	...	...	...	...	...
UGCA 219	35 × 30	90	M06	...	...	36 slit	41	106 slit	137
NGC 3521	263 × 56	110	M10	...	...	57.4 × 35.2	18	254.0 × 55.9	114
NGC 3627	200 × 55	115	M10	...	...	53.7 × 29.6	23	223.5 × 55.9	119
IC 0691	40 × 40	90	M06	...	...	25.9 × 7.4	51	71.1 × 20.3	−33
NGC 3690	90 × 60	90	M06	29.2 × 60.0	−52	40.7 × 31.4	57	111.8 × 40.6	−27
NGC 3773	38 × 55	115	M10	...	...	24.0 × 24.0	−159	35.6 × 35.6	117
Mrk 1450	20 × 15	90	M06	...	...	36 slit	197	106 slit	294
UGC 06665	30 × 60	90	M06	...	...	36 slit	197	106 slit	294
NGC 3870	50 × 40	90	M06	...	...	29.6 × 7.4	51	61.0 × 20.3	−33
UM 461	25 × 20	90	M06	...	...	36 slit	197	106 slit	294
UGC 06850	36 × 40	90	M06	...	...	36 slit	197	106 slit	294
NGC 3938	177 × 56	133	M10	...	...	49.9 × 29.6	37	49.9 × 29.6	37
NGC 4088	140 × 300	135	M06	...	...	48.1 × 49.9	−149	127.0 × 66.0	128
NGC 4125	190 × 56	90	M10	...	...	24.0 × 24.1	173	24.0 × 24.1	173
NGC 4138	60 × 120	90	M06	...	...	50.0 × 51.8	−151	96.5 × 55.9	125
NGC 4168	105 × 84	122	G04	...	...	...	...	...	...
NGC 4194	125 × 30	165	M06	34.8 × 60.0	−53	36 slit	178	106 slit	274
Haro 06	30 × 20	90	M06	...	...	9.2 × 7.4	22	35.6 × 20.3	−62
NGC 4254	177 × 55	120	M10	...	...	49.9 × 35.2	19	208.3 × 50.8	115
NGC 4321	245 × 56	121	M10	...	...	53.6 × 31.5	25	254.0 × 55.9	121
NGC 4365	209 × 148	45	G04	...	...	...	...	...	...
NGC 4387	110 × 49	140	G04	...	...	...	...	...	...
NGC 4385	100 × 60	90	M06	...	...	36 slit	196	106 slit	293
NGC 4450	173 × 55	120	M10	...	...	48.1 × 29.6	31	48.1 × 29.6	31
NGC 4458	108 × 91	11	G04	...	...	...	...	...	...
NGC 4473	242 × 120	90	G04	...	...	...	...	...	...
NGC 4486	300 × 300	90	G04	...	...	...	...	...	...
NGC 4536	250 × 55	115	M10	...	...	48.1 × 31.5	22	254.0 × 55.9	118
NGC 4550	237 × 52	0	G04	...	...	...	...	...	...
NGC 4551	120 × 96	34	G04	...	...	...	...	...	...
NGC 4552	169 × 55	120	M10	...	...	44.4 × 29.6	19	121.9 × 40.6	116
NGC 4559	354 × 55	135	M10	...	...	49.9 × 31.4	41	264.2 × 50.8	137
NGC 4569	172 × 55	120	M10	...	...	50.0 × 31.5	26	127.0 × 55.9	122
NGC 4579	194 × 55	120	M10	...	...	55.5 × 33.3	20	254.0 × 55.9	116
NGC 4594	287 × 55	113	M10	...	...	51.8 × 35.2	18	218.4 × 55.9	114
NGC 4625	72 × 56	143	M10	...	...	37.0 × 29.6	46	37.0 × 29.6	46
NGC 4621	460 × 247	165	G04	...	...	...	...	...	...
NGC 4631	512 × 56	101	M10	...	...	49.9 × 35.1	4	218.4 × 55.9	101
NGC 4660	113 × 74	88	G04	...	...	...	...	...	...
NGC 4670	80 × 40	90	M06	...	...	22.2 × 18.5	−165	55.9 × 20.3	−42
NGC 4676 A	40 × 155	90	M06	...	...	38.9 × 29.6	−173	35.6 × 40.6	103
NGC 4725	292 × 55	120	M10	...	...	27.7 × 18.5	31	162.6 × 50.8	128
NGC 4826	330 × 55	120	M10	...	...	29.6 × 31.5	30	29.6 × 31.5	30
NGC 4860	38 × 46	38	G04	...	...	...	...	...	...
NGC 4889	60 × 60	90	K92	...	...	...	...	...	...
IC 4051	55 × 40	75	G04	...	...	...	...	...	...
NGC 4926	72 × 59	63	G04	...	...	...	...	...	...
NGC 5033	354 × 56	150	M10	...	...	51.8 × 35.1	53	259.1 × 55.9	150
IC 0860	30 × 40	90	M06	7.3 × 60.0	115	36 slit	180	106 slit	277
UGC 08335 NW	50 × 45	90	M06	29.2 × 60.0	−39	24.0 × 24.1	−136	30.5 × 30.5	156
UGC 08335 SE	35 × 50	90	M06	7.3 × 60.0	141	24.1 × 24.1	−136	35.6 × 35.6	156
NGC 5055	416 × 56	120	M10	...	...	50.0 × 31.5	23	376.0 × 50.8	119
IC 0883	90 × 20	130	M06	7.3 × 60.0	118	36 slit	189	106 slit	285
NGC 5104	40 × 60	90	M06	7.3 × 60.0	−67	36 slit	191	106 slit	287
NGC 5194	370 × 56	158	M10	...	...	51.8 × 35.2	62	391.2 × 55.9	158
NGC 5195	190 × 56	118	M10	...	...	53.7 × 35.2	25	198.2 × 55.9	121
NGC 5256	50 × 30	90	M06	7.3 × 60.0	126	40.7 × 22.2	−160	61.0 × 25.4	116
NGC 5257	70 × 90	90	M06	...	...	37.0 × 37.0	19	45.7 × 40.6	−65
NGC 5258	80 × 90	90	M06	...	...	46.3 × 29.6	17	40.6 × 40.6	−66
UGC 08696	15 × 85	90	M06	7.3 × 60.0	−44	36 slit	91	106 slit	188
Mrk 1490	40 × 20	90	M06	7.3 × 60.0	128	36 slit	206	106 slit	302
NGC 5653	80 × 30	90	M06	...	...	48.1 × 35.1	30	76.2 × 35.6	−54
Mrk 0475	30 × 20	90	M06	...	...	36 slit	63	106 slit	294
NGC 5713	90 × 55	113	M10	...	...	46.2 × 35.2	19	137.2 × 50.8	115
UGC 09618 S	30 × 47	90	M06	...	...	30.0 × 47.0	177	30.0 × 47.0	94
UGC 09618	30 × 110	90	M06	7.3 × 60.0	111	75.8 × 31.4	177	45.7 × 81.3	94
UGC 09618 N	30 × 50	90	M06	7.3 × 60.0	111	30.0 × 50.0	177	30.0 × 50.0	94
NGC 5866	154 × 55	120	M10	...	...	35.2 × 33.3	30	172.8 × 50.8	126
CGCG 049−057	35 × 30	90	M06	7.3 × 60.0	107	29.6 × 16.7	24	45.7 × 30.5	−60
NGC 5953	60 × 75	90	M06	21.9 × 5.0	−73	29.6 × 20.4	−167	61.0 × 25.4	109
IC 4553	80 × 36	90	M06	...	...	36 slit	9	106 slit	105
UGCA 410	25 × 15	90	M06	...	...	36 slit	166	106 slit	262
NGC 5992	40 × 40	90	M06	21.9 × 60.0	115	36 slit	177	106 slit	273
NGC 6052	50 × 60	90	M06	29.2 × 5.0	104	35.2 × 22.2	33	55.9 × 25.4	−50
NGC 6090	45 × 20	90	M06	7.3 × 60.0	125	36 slit	25	106 slit	121
NGC 6240	50 × 80	90	M06	29.2 × 60.0	−79	66.6 × 57.4	15	106.7 × 76.2	−69
IRAS 17208−0014	25 × 25	90	M06	29.2 × 60.0	−86	36 slit	171	106 slit	268
II Zw 096	50 × 40	90	M06	29.2 × 60.0	250	38.8 × 38.9	−12	81.3 × 45.7	−98
NGC 7331	200 × 55	90	M10	...	...	48.1 × 16.6	149	182.9 × 55.9	65
UGC 12150	55 × 40	90	M06	7.3 × 60.0	61	36 slit	−21	106 slit	181
CGCG 453−062	50 × 30	90	M06	7.3 × 60.0	244	36 slit	−26	106 slit	70
IC 5298	40 × 30	90	M06	7.3 × 60.0	243	36 slit	−20	106 slit	75
NGC 7585	130 × 60	90	M06	...	...	...	...	...	...
NGC 7591	80 × 75	90	M06	29.2 × 60.0	66	38.9 × 24.0	−28	91.4 × 30.5	−111
NGC 7592	60 × 58	90	M06	29.2 × 60.0	67	35.1 × 35.2	−29	61.0 × 45.7	−113
NGC 7673	70 × 40	90	M06	...	...	40.7 × 20.4	−27	86.4 × 25.4	−111
NGC 7674	60 × 60	90	M06	7.3 × 60.0	246	31.5 × 20.3	−27	61.0 × 25.4	−111
NGC 7679	80 × 30	90	M06	29.2 × 60.0	62	36 slit	−28	106 slit	67
Mrk 0930	25 × 15	90	M06	...	...	36 slit	−35	106 slit	60
NGC 7714	60 × 60	90	M06	17.5 × 5.0	246	36 slit	−30	106 slit	65
NGC 7771	130 × 50	90	M06	29.2 × 60.0	65	51.8 × 27.8	−23	116.8 × 35.6	−107
Mrk 0331	40 × 40	90	M06	7.3 × 60.0	64	36 slit	−21	106 slit	75

Download table as:  ASCIITypeset images: 1 2

Table 3. AB Magnitudes for the Sample with Neither Foreground Nor Intrinsic Dust Extinction Corrections

Name	FUV	UVW2	UVM2	NUV	UVW1	U	u	g	V
	r	i	z	J	H	KS	W1	[3.6]	[4.5]
	W2	[5.8]	[8.0]	W3	PUI Blue	W4	W4'	PUI Red	[24]
Arp 256 N	16.26 ± 0.10	⋅⋅⋅	⋅⋅⋅	15.91 ± 0.10	⋅⋅⋅	⋅⋅⋅	15.31 ± 0.05	14.43 ± 0.05	⋅⋅⋅
	14.03 ± 0.05	13.77 ± 0.05	13.62 ± 0.05	13.33 ± 0.05	13.28 ± 0.05	13.34 ± 0.05	13.83 ± 0.10	13.85 ± 0.10	14.22 ± 0.10
	14.34 ± 0.10	13.12 ± 0.10	11.98 ± 0.10	12.06 ± 0.10	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	11.23 ± 0.10
Arp 256 S	16.37 ± 0.10	⋅⋅⋅	⋅⋅⋅	16.00 ± 0.10	⋅⋅⋅	⋅⋅⋅	15.26 ± 0.05	14.48 ± 0.05	⋅⋅⋅
	14.05 ± 0.05	13.80 ± 0.05	13.58 ± 0.05	13.25 ± 0.05	13.03 ± 0.05	13.15 ± 0.05	13.28 ± 0.10	13.16 ± 0.10	13.40 ± 0.10
	13.49 ± 0.10	11.92 ± 0.10	⋅⋅⋅	10.53 ± 0.10	⋅⋅⋅	8.95 ± 0.10	9.12 ± 0.10	⋅⋅⋅	8.90 ± 0.10
NGC 0337	15.32 ± 0.10	14.77 ± 0.10	14.95 ± 0.10	14.63 ± 0.10	14.31 ± 0.10	⋅⋅⋅	13.37 ± 0.05	12.36 ± 0.05	⋅⋅⋅
	11.88 ± 0.05	11.65 ± 0.05	11.51 ± 0.05	11.27 ± 0.05	11.12 ± 0.05	11.34 ± 0.05	11.89 ± 0.10	11.85 ± 0.10	12.24 ± 0.10
	12.36 ± 0.10	11.24 ± 0.10	10.31 ± 0.10	10.46 ± 0.10	10.31 ± 0.10	9.63 ± 0.10	9.79 ± 0.10	⋅⋅⋅	9.55 ± 0.10
CGCG 436−030	17.48 ± 0.10	⋅⋅⋅	⋅⋅⋅	16.98 ± 0.10	⋅⋅⋅	⋅⋅⋅	15.85 ± 0.05	14.79 ± 0.05	⋅⋅⋅
	14.34 ± 0.05	14.03 ± 0.05	13.87 ± 0.05	13.46 ± 0.05	13.15 ± 0.05	13.17 ± 0.07	13.39 ± 0.10	13.21 ± 0.10	12.94 ± 0.10
	12.96 ± 0.10	11.74 ± 0.10	10.62 ± 0.10	10.54 ± 0.10	⋅⋅⋅	8.79 ± 0.10	8.99 ± 0.10	⋅⋅⋅	8.70 ± 0.10
NGC 0474	19.16 ± 0.12	⋅⋅⋅	⋅⋅⋅	17.14 ± 0.10	⋅⋅⋅	⋅⋅⋅	14.05 ± 0.05	12.41 ± 0.05	⋅⋅⋅
	11.66 ± 0.05	11.25 ± 0.05	10.99 ± 0.05	10.67 ± 0.05	10.59 ± 0.05	10.74 ± 0.05	11.44 ± 0.10	11.52 ± 0.10	12.02 ± 0.10
	12.12 ± 0.10	12.30 ± 0.10	12.91 ± 0.10	13.32 ± 0.10	⋅⋅⋅	14.08 ± 0.10	14.15 ± 0.10	⋅⋅⋅	⋅⋅⋅

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Table 4. Modeled Milky Way Foreground Dust Extinction

Name	A(FUV)	A(UVW2)	A(UVM2)	A(NUV)	A(UVW1)	A(U)	A(u)	A(g)	A(V)
	A(r)	A(i)	A(z)	A(J)	A(H)	A(KS)	A(W1)	A([3.6])	A([4.5])
	A(W2)	A([5.8])	A([8.0])	A(W3)	A(PUI Blue)	A(W4)	A(W4')	A(PUI Red)	A([24])
Arp 256 N	0.27	0.27	0.29	0.27	0.20	0.16	0.15	0.12	0.10
	0.08	0.06	0.05	0.025	0.017	0.012	0.006	0.006	0.005
	0.005	0.004	0.003	0.002	0.001	0.001	0.001	0.001	0.001
Arp 256 S	0.27	0.27	0.29	0.27	0.20	0.16	0.15	0.12	0.10
	0.08	0.06	0.05	0.025	0.017	0.012	0.006	0.006	0.005
	0.005	0.004	0.003	0.002	0.001	0.001	0.001	0.001	0.001
NGC 0337	0.84	0.87	0.90	0.86	0.65	0.49	0.48	0.37	0.30
	0.25	0.19	0.15	0.080	0.054	0.037	0.021	0.019	0.015
	0.015	0.012	0.009	0.006	0.004	0.003	0.003	0.003	0.003
CGCG 436−030	0.49	0.50	0.52	0.50	0.37	0.28	0.28	0.21	0.17
	0.15	0.11	0.08	0.046	0.031	0.021	0.012	0.011	0.009
	0.009	0.007	0.005	0.003	0.003	0.002	0.002	0.002	0.002
NGC 0474	0.31	0.31	0.33	0.31	0.23	0.18	0.17	0.13	0.11
	0.09	0.07	0.05	0.029	0.020	0.013	0.007	0.007	0.006
	0.005	0.004	0.003	0.002	0.002	0.001	0.001	0.001	0.001

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Many galaxies in our sample are bright so photon counting noise is likely to be small relative to systematic errors. For the brightest galaxies in our sample, we assumed the uncertainties had a floor of 0.05 mag in the SDSS and 2MASS bands, and we assumed the uncertainties in the other bands had a floor of 0.10 mag. For the fainter galaxies in our sample (i.e., m_FUV > 18, $m_{K_S}>13$), we estimated the uncertainties and corrected for background errors by measuring fluxes at 24 locations (offset from the galaxy position). The distribution of the fluxes in these apertures was used to determine the uncertainties while the median of the fluxes was used to correct for background errors. For a small number of star-forming galaxies, the 2MASS photometry has very large uncertainties so we used the near-infrared photometry of Engelbracht et al. (2008) and Vaduvescu et al. (2005) (when available) for constraining and verifying the spectra.

2.1.1. Zero-point and Point-spread Function Corrections

Large aperture photometry will overestimate fluxes if the zero-points have been determined with small apertures that do not capture all the light from relevant standard stars. The Spitzer IRAC imaging was calibrated using 24'' diameter aperture photometry (without corrections for flux beyond the aperture) and IRAC large aperture photometry suffers from scattered light within the aperture, which can increase measured galaxy fluxes by 10% or more (Fazio et al. 2004). To mitigate these issues, we used the IRAC point-spread function (PSF) to determine the fraction of light beyond the aperture and we used the corrections described in the IRAC Instrument Handbook¹⁸ to model scattered light as a function of aperture area. Swift UVOT photometry was calibrated using 10'' diameter aperture photometry (without corrections for flux beyond the aperture) and we correct for this using approximations to the curves of growth presented in Breeveld et al. (2010). Spitzer IRS peak-up blue and red channel imaging is calibrated using 12 and 13 pixel radius apertures, respectively, and we made small corrections for this using the peak-up imager PSFs. The WISE all sky release zero-points were determined by fitting PSF models to the inner region of the PSF, and consequently there are small zero-point errors for large aperture photometry. To correct for this, we added 0.03, 0.04, 0.03, and −0.03 to the measured WISE W1, W2, W3, and W4 magnitudes, respectively (Jarrett et al. 2012).

For galaxies with small angular sizes, we are effectively measuring total magnitudes and the smallest apertures approach the size of the GALEX, IRS, MIPS, and WISE PSFs. For the ultraviolet and mid-infrared photometry we applied an approximate correction for flux beyond the aperture using radially averaged models of the PSF. For very small apertures the WISE W4 PSF results in PSF corrections of 50% or more, so for isolated galaxies measured with small spectroscopic extraction apertures (less than 30'' on either side) we increased the W4 aperture size to at least 30'' on a side.

2.1.2. Photon Coincidence Loss Corrections

The Swift UVOT is a micro-channel plate intensified CCD and suffers from significant photon coincidence losses (Roming et al. 2005). Coincidence loss corrections have been determined for point sources and a surface brightness limit has been established, below which coincidence losses have a minimal impact on extended source photometry (Breeveld et al. 2010; Hoversten et al. 2011). However, coincidence loss corrections for bright extended sources are not available in the prior literatures. To measure the photon coincidence losses for extended sources, we compared Swift UVOT U-band and SDSS u-band surface brightness measurements for bright galaxies. Figure 2 compares Swift UVOT and SDSS measurements of the surface brightness for NGC 2403 and NGC 4486, and the impact of photon coincidence losses can be clearly seen. The losses approach 50% for a surface brightness of U ∼ 20 mag arcsec⁻², corresponding to a photon count rate of 0.5 photons s⁻¹ arcsec⁻². The photon coincidence losses occur at relatively faint magnitudes due to the large (4''  ×  4'') physical pixel scale of the UVOT detector and the use of five pixels when centroiding photon splashes.

Zoom In Zoom Out Reset image size

We applied an approximate correction to the photon counts given by

$$\begin{eqnarray} C_{{\rm cor}} &=& \frac{ {-}\ln (1- 80 \times C_{{\rm raw}}\times ft -2[80 \times C_{{\rm raw}}\times ft]^2) }{ 80\times ft(1-df)},\nonumber\\ \end{eqnarray} \tag{ 3 }$$

where C_raw is the counts per second per square arcsecond, ft is the frame time (0.011088 s) and df is the deadtime fraction (0.0155844). As this is an approximation, we measured Swift photometry using images with and without the photon coincidence loss corrections applied, and generally only include measurements when the correction to the photometry is less than 0.3 mag. (To increase the number of blue compact dwarfs in our sample, we include Swift photometry for Mrk 1450 with corrections up to 0.5 mag.) A consequence of the coincidence loss corrections and the 0.3 mag criterion is much of the Swift optical photometry (including all the B band) is not included in the final sample.

2.1.3. WISE W4 Filter Curve Correction

The pre-launch WISE W4 (22 μm) filter response curve does not match the on-sky performance of the WISE W4 measurements (Wright et al. 2010). The WISE photometric measurements are calibrated using a network of A stars and K–M giants (Jarrett et al. 2011), with typical spectral indices (α) between 1 and 2, where the spectral index is defined by f_ν∝ν^α. Hence, by definition, the photometry is well behaved for objects with Rayleigh–Jeans SEDs, including early-type galaxies. Star-forming galaxies and AGNs with SEDs that rise toward longer wavelengths have WISE W4 magnitudes that are at least 10% brighter than what is expected from Spitzer IRS spectrophotometry and Spitzer 24 μm photometry (Wright et al. 2010; Jarrett et al. 2013). As we show in Figure 3, we have been able to measure the anomaly with greater fidelity with our sample, and measure it as a function of ∼22 μm spectral index. We find the residual is well approximated by

$$\begin{equation} \Delta m_{W4} = 0.035 \times (\alpha _{22} - 2), \end{equation} \tag{ 4 }$$

with an rms of 0.05 mag. We present our original and corrected WISE W4 photometry in Table 3, with the latter being denoted by W4'.

Zoom In Zoom Out Reset image size

2.1.4. The Observed Colors of Nearby Galaxies

The aperture photometry of the sample galaxies is presented in full in Table 3 and the optical color–magnitude diagram for the sample is plotted in Figure 4. The symbols in Figure 4 are a function of RC3 morphological T-Type (de Vaucouleurs et al. 1991), and sample galaxies that are not in the RC3 have almost exclusively irregular and peculiar morphologies (as shown in Table 1). For comparison, in Figure 4 we also plot the colors and magnitudes of g < 14 galaxies drawn from the NASA-Sloan Atlas (Blanton et al. 2011). Our sample spans the observed range of galaxy colors (0.1 < u − g < 1.9), a broad range of absolute magnitude (−14.7 > M_g > −23.2) and a broad range of galaxy types. While our sample spans a broad range of galaxy properties, some galaxy types are not included in the sample (e.g., ultra-compact dwarf galaxies) while other galaxy types (such as LIRGs) are over-represented. We caution that while galaxy morphological types are correlated with galaxy spectra, there are outliers including elliptical galaxies with star formation (e.g., NGC 855) and disk galaxies with low star formation rates (e.g., NGC 4450).

Zoom In Zoom Out Reset image size

In Figure 5 we plot the optical colors of sample galaxies and NASA-Sloan Atlas galaxies. The majority of galaxies fall along a relatively narrow locus running diagonally across the diagram. Our galaxy locus is slightly offset from the locus of NASA-Sloan Atlas galaxies, due to our use of aperture photometry and the use of Sérsic model fit photometry by NASA-Sloan Atlas. The largest differences between our photometry and the NASA-Sloan Atlas photometry are seen for galaxies where the NASA-Sloan Atlas Sérsic model fits show significant residuals (e.g., NGC 660, NGC 4088) and our aperture does not span the entire galaxy (e.g., NGC 4725, NGC 7331). The optical colors of elliptical galaxies are slightly offset from the locus of star-forming galaxies (including dust reddened galaxies), and this offset is known to become more apparent when combinations of optical and near-infrared colors are used (e.g., Cardamone et al. 2010). The bluest galaxies have strong emission lines that contribute significantly to the observed ugr photometry, and these galaxies fan across the lower left of the diagram. Many of the galaxies falling to the right of the galaxy locus are heavily obscured (e.g., NGC 660) while those falling to the left have strong Balmer breaks (e.g., NGC 5992).

Zoom In Zoom Out Reset image size

The diversity of the sample is illustrated by Figures 6 and 7, where we plot the ultraviolet and infrared colors of sample galaxies. Elliptical galaxies have a broad range of ultraviolet continuum slopes, and this has been known for several decades (e.g., Code & Welch 1979; Donas et al. 1987, 2007). The distribution of ultraviolet colors tightens as one moves toward the left of Figure 6, with blue dwarf galaxies having a relatively narrow range of ultraviolet colors.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Figure 7 we plot the infrared colors of the sample galaxies. Galaxies with a Rayleigh–Jeans spectrum at ∼4 μm fall along the bottom of the figure, with galaxies with weak and strong PAH emission at ∼6 μm falling on the left and right of the plot respectively. Galaxies that lie above the locus in Figure 7 include AGNs, galaxies with ice absorption at ≃ 3.1 μm and galaxies with hot dust emission. We caution that our large sample does not capture the full diversity of local galaxy populations, as some populations are not included (e.g., ultra-compact dwarfs, low surface brightness galaxies, quasars) and it does not capture the full diversity of other populations (e.g., LIRGs; Howell et al. 2010). However, our galaxy sample does capture much of the diversity of local galaxy populations (as discussed in Section 4) and illustrates why a relatively small number of galaxy templates can struggle to model the full diversity of local galaxies.

2.2. Optical Spectra

2.3. Spitzer Infrared Spectra

2.4. Akari Infrared Spectra

Galaxy SEDs were produced by normalizing the observed spectra with our photometry and filling the gaps with spectral coverage with power-law fits or (more frequently) model spectra produced using the Multi-wavelength Analysis of Galaxy Physical Properties code (MAGPHYS; da Cunha et al. 2008). MAGPHYS includes stellar population synthesis models (Bruzual & Charlot 2003) combined with a self-consistent model of dust emission, dust absorption, and PAH emission.

The multiplicative factors used to renormalize the optical spectra are plotted in Figure 8 and listed in Table 6. Assuming our SDSS g-band photometry is valid, the measured optical spectra typically underestimate the fluxes of the true optical spectra by approximately 10%, with larger offsets seen for brighter (and more extended) galaxies. As these underestimates also apply to the nebular emission lines, star formation rate indicators calibrated with the spectra (e.g., Kennicutt et al. 2009) may systematically underestimate the true star formation rates. A possible cause for this systematic underestimation is sky background subtraction using a slit of finite length (∼33 for the Bok 2.3 m telescope), which may be contaminated by galaxy light for the very largest galaxies in the sample (e.g., NGC 4569). A recalibration of star formation rate indicators (e.g., mid-infrared luminosities between 10 and 24 μm) using our SEDs is beyond the scope of this paper and will be the subject of a future work.

Zoom In Zoom Out Reset image size

MAGPHYS does not include nebular emission lines, so we produced synthetic optical ugr photometry with the nebular emission lines subtracted, using the emission line fluxes from Moustakas & Kennicutt (2006) and Moustakas et al. (2010). In cases where [Ne iii] 3869 Å was not measured, we used the relation of Pérez-Montero et al. (2007) to approximate the [Ne iii] 3869 Å flux using the measured [O iii] 5007 Å flux. We show the resulting optical ugr color–colour diagram in the right panel of Figure 5. It is immediately clear from this diagram that the galaxy locus is approximately a straight line once the contribution of emission lines has been removed. Very blue star-forming galaxies that initially have comparable optical colors can move to different regions of the color–color diagram, in part due to differing contributions from [O iii] 5007 Å and Hα. Consequently, it can be very difficult to determine some properties of emission line galaxies (e.g., stellar masses) when only broadband photometry is available (e.g., Schaerer & de Barros 2009; Atek et al. 2011).

As we illustrate in Figure 9, the MAGPHYS and observed spectra can approximate each other remarkably well. However, the MAGPHYS and observed spectra do differ from each other, and simply switching from one to the other can produce false spectral breaks. To produce smooth and continuous spectra, we rescaled the MAGPHYS SEDs near the optical, Akari and Spitzer spectra so the models joined the observed spectra smoothly. For example, for each galaxy near 6800 Å we determined a scaling factor by comparing the MAGPHYS and optical spectrum fluxes over a wavelength range corresponding to the last 500 Å of the optical spectrum. We then multiplied the MAGPHYS fluxes by this scaling factor at ∼6800 Å, and used progressively smaller scaling factors until the MAGPHYS fluxes were unchanged at 10700 Å. For galaxies with both Akari and Spitzer spectra, we used linear interpolation to fill the small gap between the Akari and Spitzer spectra. For UGCA 410, MAGPHYS did not approximate the spectra near 6 μm so we interpolated between 4.5 μm and the Spitzer spectra using a double power law.

Zoom In Zoom Out Reset image size

For a small number of early-type galaxies we do not have Spitzer IRS spectra and we use power laws to model the spectra of these galaxies at long wavelengths. The ∼20 μm spectral index is left as a free parameter, as some galaxies do not precisely match a Rayleigh–Jeans spectrum. There are also some galaxies (e.g., AGNs) where the photometry is better matched (over a limited wavelength range) by a power law than a MAGPHYS model (e.g., UGC 5101). For these galaxies we replaced the MAGPHYS model with a power law over the relevant wavelength range. Some galaxies have significant AGN emission in the mid-infrared which cannot be approximated by MAGPHYS (NGC 1068, NGC 1275, and NGC 6240), so the mid-infrared photometry was excluded when fitting MAGPHYS models to the photometry of these galaxies. Some galaxies have significant AGN emission in the near-infrared that cannot be approximated by MAGPHYS models nor power laws, and (unfortunately) these had to be rejected from the sample (e.g., Mrk 231).

MAGPHYS struggled to fit the ultraviolet photometry of some LIRGs and AGN host galaxies, which is not unexpected given MAGPHYS does not include AGN emission and does not model very complex dust geometries. We applied two corrections to the spectra of CGCG 436−030, IC 860, IC 883, IC 5298, II Zw 96, III Zw 035, IRAS 08572+3915, IRAS 17208−0014, Mrk 331, NGC 2623, NGC 3690, NGC 5256, NGC 6240, NGC 7674, UGC 5101, UGC 8335 S, UGC 8696, and UGC 9168 N so they were forced to agree with the GALEX FUV and NUV photometry. The two corrections vary linearly with wavelength, with one correction forced to equal zero beyond 2315 Å and the other forced to equal zero beyond 3561 Å. The corrections can be as large as ∼60% near the GALEX FUV band, but smaller values are more typical. While the approximations are crude, the resulting spectra should be better than the prior literature (as we discuss in Section 4) and will produce improved broadband colors of LIRGs and AGN host galaxies.

We used the resulting SEDs to determine synthetic photometry, and we compared this with the observed photometry used to normalize and verify the spectra. Any residuals larger than 0.2 mag are flagged and visually inspected. In some cases outliers were removed by improved masking of foreground stars or improved background subtraction. For some galaxies the signal to noise of the optical and Spitzer spectra was low and these were rejected from the sample. For a handful of galaxies discrepancies were evident in multiple bands and the source of the error could not be identified nor remedied, so the relevant galaxy was rejected from the sample.

The residuals of the observed photometry and photometry synthesized from the spectra are summarized in Table 5. As some bands were used to normalize the spectra (i.e., g band), we caution that some bands have small residuals by construction. The median and standard deviation of the offsets between the photometry and spectra are less than 10% for almost all of the bands.

After applying the correction for the W4 filter curve error, the most significant residual is found in 5.8 μm IRAC Channel-3 photometry. In Figure 10 we plot the residuals as a function of galaxy color, with different symbols denoting different redshift ranges. While galaxies with Rayleigh–Jeans spectra show relatively modest errors, the photometry of star-forming galaxies is systematically brighter than the spectroscopy. As the presence of the residual depends on SED shape and is not seen in the other IRAC bands, it unlikely to result from errors in the scattered light corrections we have previously applied to the photometry. The 6.2 μm PAH feature is near the red end of the 5.8 μm filter curve, and should start moving out of the band toward the upper end of the redshift range. The systematic offset is removed if we shift the 5.8 μm filter curve redward by 1.5%, although doing this may not be justified. The blue end of the IRS spectroscopy also overlaps the 5.8 μm filter curve, and the spectra can have poor signal to noise shortward of the 6.2 μm PAH feature. One could apply corrections to the IRS spectroscopy of star-forming galaxies, but doing so produces spectra that poorly match the Akari spectra at shorter wavelengths. We do not apply any corrections to the photometry or spectra 5.8 μm, but readers should be cautious of the spectra and photometry near this wavelength.

Zoom In Zoom Out Reset image size

In Figure 11 we provide plots for four of the galaxies drawn from the illustrative subsample described in Appendix A. Spectra and images of all the galaxies in the sample are provided in Appendix B. Figure 11 shows good agreement between the photometry and spectroscopy, although we caution that some of this agreement is by construction (Section 4 compares the spectra with an independent set of photometry). Figure 11 also illustrates the diversity of the sample, including a passive early-type galaxy (NGC 4125), a merging galaxy (NGC 5953), and a low metallicity dwarf galaxy (UGCA 219).

Zoom In Zoom Out Reset image size

Below we directly compare our SEDs (of particular galaxies) with those from the prior literature, and compare how well different template libraries reproduce the observed colors of galaxies. Systematic differences between our SEDs and those of previous template libraries are inevitable due to differences in methodology and data, with much of the data presented here not being available a decade ago. A key difference is the size of the aperture used to measure photometry and extract spectra, with some of the prior literature being restricted to galaxy nuclei (e.g., Kinney et al. 1996), which may not be representative of entire galaxies. We have deliberately produced spectra for individual galaxies, whereas the prior literature sometimes merged spectra from different galaxies to produce spectra spanning broad wavelength ranges. That said, these limitations and sources of error were well understood and discussed by the relevant authors, who constructed the best SEDs possible with the data then available.

The biggest differences between our templates and the prior literature should be for LIRGs and ULIRGs, as these are difficult to model with stellar population synthesis models and are relatively faint in the ultraviolet. In Figure 12 we plot our spectrum of IC 4553 (Arp 220), along with other IC 4553 templates from the recent literature (Chary & Elbaz 2001; Donley et al. 2007; Polletta et al. 2007; Rieke et al. 2009). While IC 4553 templates are frequently used to model the spectra of ULIRGs, the properties of this galaxy are unusual (e.g., Soifer et al. 1984) and ULIRG properties do display considerable diversity (e.g., Brandl et al. 2006; Howell et al. 2010; U et al. 2012).

Zoom In Zoom Out Reset image size

It is immediately obvious from Figure 12 that the ultraviolet SEDs vary enormously, with the prior literature overestimating the ∼2000 Å flux of IC 4553 by over 100%. As with this work, the IC 4553 templates from the prior literature use combinations of spectra and model fits to photometry, using the data available at the time of publication. While it was reasonable for the prior literature to extend their IC 4553 SEDs into the ultraviolet (e.g., to approximate the observed optical colors of high redshift galaxies), previous IC 4553 SEDs had few (if any) constraints in ultraviolet (as noted by Chary & Elbaz 2001), and consequently they systematically overestimated the ultraviolet flux of IC 4553. However, we caution that even with GALEX and Swift, we are constraining the ultraviolet SED of IC 4553 with just three photometric data points.

As can be seen in Figure 12, the shapes of the various IC 4553 (Arp 220) templates vary in the optical and mid-infrared too, with our spectrum featuring a relatively red u − r color, strong PAH emission, and 3.1 μm ice absorption feature. As all of the recent IC 4553 templates make use of (or have been updated with) Spitzer IRS spectra, the broad agreement beyond 5 μm is not surprising. As IC 4553 template spectra are often used to interpret the observations of high redshift galaxies, it is plausible that errors in template spectra have led to erroneous conclusions. As we illustrate in Figure 13, the issues seen in IC 4553 templates are also seen in other LIRG SEDs. However, as we discuss below, there is better agreement between different template libraries for galaxy types with little dust emission and no AGN content (e.g., ellipticals).

Zoom In Zoom Out Reset image size

To compare our ensemble of templates with the prior literature, we generated synthetic colors and compared them to observed galaxy colors. We compare the performance of our templates with those from Coleman et al. (1980), Kinney et al. (1996) and Polletta et al. (2007), which have been used by a large variety of K-correction and photometric redshift codes over the past two decades (e.g., Fernández-Soto et al. 1999; Benítez 2000; Bolzonella et al. 2000; Ilbert et al. 2009). For this comparison we used optical and mid-infrared colors of z ∼ 0.3 galaxies in the Boötes field (Jannuzi & Dey 1999; Ashby et al. 2009). By changing redshift we avoid comparing our SEDs with photometry that has the same rest-frame wavelengths as the photometry used to constrain and verify the SEDs. The Boötes field optical imaging, taken from the National Optical Astronomy Observatory Deep Wide-Field Survey, also uses a different set of optical filters than the SDSS. The spectroscopic redshifts used for this comparison are taken from I < 20.4 AGN and Galaxies Evolution Survey (Kochanek et al. 2012). The Boötes photometric catalogs are described in Brown et al. (2008) and we use a magnitude dependent aperture size to capture the vast majority of the galaxy light.

Figure 14 compares our templates, the Coleman et al. (1980) templates and the Kinney et al. (1996) templates with the observed optical colors of z = 0.3 and z = 0.4 galaxies in Boötes. Both the Coleman et al. (1980) and Kinney et al. (1996) templates are systematically offset from the observed galaxy locus, and in some cases the offset corresponds to several tenths of a magnitude. Our templates broadly follow the observed galaxy loci in Figure 14. Some templates do fall off the galaxy locus, including blue emission line galaxies (which are generally fainter than I = 20.4 at these redshifts) and LIRGs (that are relatively rare at z ∼ 0.3). Some regions of the color-space that are populated by Kinney et al. (1996) templates are not populated by our templates. Given our large sample size and accurate photometry, we suspect this is due to errors in the Kinney et al. (1996) templates rather than incompleteness in our template library.

Zoom In Zoom Out Reset image size

In Figure 15 we compare our templates and the Polletta et al. (2007) templates with the observed mid-infrared colors of z ∼ 0.2 galaxies in Boötes. Galaxies with Rayleigh–Jeans SEDs and strong PAH emission form two distinct loci on the left and right of the plots. The PAH locus is not populated by the Polletta et al. (2007) templates, while multiple templates from our library populate the PAH locus. Our ability to populate the color–color diagram with templates is a consequence of the input photometry and spectroscopy, combined with our large sample size. However, even with 129 templates some regions of the color–color diagram are sparsely populated, including galaxies with mid-infrared colors that do not quite match Rayleigh–Jeans SEDs (e.g., galaxies similar to NGC 4450 and NGC 4725).

Zoom In Zoom Out Reset image size

We have constructed an atlas of 129 SEDs for nearby galaxies.²² The atlas combines optical, Spitzer, and Akari spectra with MAGPHYS models, all of which have been normalized, constrained and verified with matched-aperture photometry. The redundancy of the 26 bands of matched-aperture photometry allows us to identify and mitigate systematic errors known to be present in the imaging data, including scattered light in IRAC images, coincidence losses in the Swift UVOT imaging and errors in the pre-launch WISE W4 filter curve. Comparison of the photometry and spectroscopy reveals that spectra used to calibrate star formation rate indicators systematically underestimate Hα fluxes by 10% or more. The SEDs are typically in good agreement with the photometry, with residuals being less than 10% in most cases.

Our atlas of SEDs spans a broad range of galaxy absolute magnitudes (−14.7 > M_g > −23.2), colors (0.1 < u − g < 1.9), and types, including ellipticals, spirals, merging galaxies, blue compact dwarfs, and luminous infrared galaxies. Within individual object classes we see considerable diversity in our SEDs. For example, there is the well known diversity of ultraviolet SEDs for elliptical galaxies, and some elliptical galaxies have significant star formation and dust emission in the mid-infrared. Multi-wavelength photometry reveals the diversity of galaxy properties and shows that this diversity cannot be adequately modeled with a small number of galaxy templates.

Our SEDs differ significantly from those in the prior literature, particularly for LIRGs. Relative to the prior literature, our SEDs provide a better match to the observed optical and mid-infrared colors of z < 0.5 galaxies. Our atlas will provide improved K-corrections, photometric redshifts, and local analogs of distant galaxies than existing template libraries. However, significant improvements can still be made including extending the atlas to longer wavelengths and filling gaps in the spectral coverage in the ultraviolet and near-infrared.

We thank Vianney Lebouteiller for responding to our queries about and making improvements to the Cornell Atlas of Spitzer/IRS Sources (CASSIS Lebouteiller et al. 2011). We thank Michelle Cluver for using and providing feedback on a preliminary version of the templates (Cluver et al. 2013, 2014). We also thank Richard Beare, Mark Brodwin, Vassilis Charmandaris, Charlie Conroy, Simon Driver, Peppo Gavazzi, Will Hartley, Erik Hoversten, Hanae Inami, Rob Kennicutt, and Lauranne Lanz for their assistance and feedback as the spectral templates were being developed. Michael Brown acknowledges financial support from The Australian Research Council (FT100100280), the Monash Research Accelerator Program (MRA), and the International Science Linkages Program. Part of the research for this paper was conducted during a visit to the Aspen Center for Physics in 2012 June. Masa Imanishi is supported by Grants-in-Aid for Scientific Research (No. 22012006).

Swift UVOT was designed and built in collaboration between MSSL, PSU, SwRI, Swales Aerospace, and GSFC, and was launched by NASA. GALEX is a NASA Small Explorer, launched in 2003 April. We gratefully acknowledge NASA's support for construction, operation, and science analysis for the GALEX mission, developed in cooperation with the Centre National d'Etudes Spatiales of France and the Korean Ministry of Science and Technology.

Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Office of Science. The SDSS-III Web site is http://www.sdss3.org/. SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration including the University of Arizona, the Brazilian Participation Group, Brookhaven National Laboratory, University of Cambridge, University of Florida, the French Participation Group, the German Participation Group, the Instituto de Astrofisica de Canarias, the Michigan State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley National Laboratory, Max Planck Institute for Astrophysics, New Mexico State University, New York University, Ohio State University, Pennsylvania State University, University of Portsmouth, Princeton University, the Spanish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University. The NASA-Sloan Atlas was created by Michael Blanton, with extensive help and testing from Eyal Kazin, Guangtun Zhu, Adrian Price-Whelan, John Moustakas, Demitri Muna, Renbin Yan, and Benjamin Weaver. Funding for the NASA-Sloan Atlas has been provided by the NASA Astrophysics Data Analysis Program (08-ADP08-0072) and the NSF (AST-1211644).

This research is based in part on observations taken with telescopes of the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This research is based in part on observations with Akari, a JAXA project with the participation of ESA.

This work is based in part on observations made with the Spitzer Space Telescope, obtained from the NASA/IPAC Infrared Science Archive, both of which are operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with the National Aeronautics and Space Administration. The Cornell Atlas of Spitzer/IRS Sources (CASSIS) is a product of the Infrared Science Center at Cornell University, supported by NASA and JPL. This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration.

Facilities: Akari (IRC) - Akari (ASTRO-F), Bok (Boller & Chivens spectrograph) - Steward Observatory 2.3 meter Bok Telescope, CTIO:1.5m (R-C spectrograph) - Cerro Tololo Inter-American Observatory's 1.5 meter Telescope, CTIO:2MASS - 2MASS Telescope at Cerro Tololo Inter-American Observatory, FLWO:2MASS - 2MASS Telescope at Fred Lawrence Whipple Observatory, GALEX - Galaxy Evolution Explorer satellite, Mayall (MOSAIC-1 wide-field camera) - Kitt Peak National Observatory's 4 meter Mayall Telescope, MMT (Hectospec) - MMT at Fred Lawrence Whipple Observatory, Sloan - Sloan Digital Sky Survey Telescope, Spitzer (IRAC, IRS, MIPS) - Spitzer Space Telescope satellite, Swift - Swift Gamma-Ray Burst Mission, WISE - Wide-field Infrared Survey Explorer

In Figure 16 and Table 7 we present an illustrative subsample of 18 galaxy spectra. We present spectra of individual galaxies rather than averaging spectra, as averaging can produce spectra that do not match those of real galaxies. As we are providing spectra of individual galaxies, it is also possible for others to add spectra to these templates as new data becomes available. The subsample was selected to span a broad range of NUV − u color, and for each NUV − u color bin shown in Figure 16 we chose three galaxies which have differing mid-infrared SEDs. Where possible, we selected galaxies that had Akari spectra, particularly when galaxies had strong mid-infrared emission and were thus likely to have strong PAH emission and the 3.1 μm ice absorption feature. The observed colors of the subsample galaxies are plotted in Figure 17. We caution that our subsample does not include some of the most extreme galaxies from the overall sample (e.g., IRAS 08572+3915, UGC 5101) and none of the subsample galaxies meet the Stern et al. (2005) AGN selection criterion. With some exceptions, including LIRGs, the optical colors of the subsample galaxies lie close to the locus of NASA-Sloan Atlas galaxies.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Figure 18 we present the spectra and gri images of the sample galaxies (the full figure is available online). Overlaid on the gri images are the photometric apertures and the apertures used to extract the Akari and Spitzer spectra (defined in Table 2). When Spitzer stare mode spectra were used, the length of the slit used for the extraction varies with wavelength, so we plot a quarter of the slit width in these instances. The plots of the individual galaxy spectra include the dust corrected photometry used to constrain and verify the spectroscopy (also see Table 3), including corrections for Swift UVOT coincidence losses and errors in the WISE W4 filter curve.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (129 panels)

Tarball of figure set images