Swift/UVOT+MaNGA (SwiM) Value-added Catalog
M. Molina, N. Ajgaonkar, R. Yan, R. Ciardullo, C. Gronwall, M. Eracleous, X. Ji, M. R. Blanton
DDraft version September 23, 2020
Typeset using L A TEX twocolumn style in AASTeX62
Swift /UVOT+MaNGA (SwiM) Value-added Catalog
Mallory Molina,
1, 2
Nikhil Ajgaonkar, Renbin Yan, Robin Ciardullo, Caryl Gronwall, Michael Eracleous, Xihan Ji, and Michael R. Blanton Department of Astronomy and Astrophysics and Institute for Gravitation and the Cosmos, The Pennsylvania State University, 525Davey Lab, University Park, PA 16803, USA eXtreme Gravity Institute, Department of Physics, Montana State University, Bozeman, MT 59715, USA Department of Physics and Astronomy, University of Kentucky, 505 Rose St., Lexington, KY 40506-0057, USA Center for Cosmology and Particle Physics, Department of Physics, New York University
ABSTRACTWe introduce the
Swift /UVOT+MaNGA (SwiM) value added catalog, which comprises 150 galaxiesthat have both SDSS/MaNGA integral field spectroscopy and archival
Swift /UVOT near-UV (NUV)images. The similar angular resolution between the three
Swift /UVOT NUV images and the MaNGAmaps allows for a high-resolution comparison of optical and NUV indicators of star formation, crucialfor constraining quenching and attenuation in the local universe. The UVOT NUV images, SDSSimages, and MaNGA emission line and spectral index maps have all been spatially matched and re-projected to match the point spread function and pixel sampling of the
Swift /UVOT uvw2 images, andare presented in the same coordinate system for each galaxy. The spectral index maps use the definitionfirst adopted by Burstein et al. (1984), which makes it more convenient for users to compute spectralindices when binning the maps. Spatial covariance is properly taken into account in propagating theuncertainties. We also provide a catalog that includes PSF-matched aperture photometry in the SDSSoptical and Swift NUV bands. In an earlier, companion paper (Molina et al. 2020) we used a subsetof these galaxies to explore the attenuation laws of kiloparsec-sized star forming regions. The catalog,maps for each galaxy, and the associated data models, are publicly released on the SDSS website a) . Keywords: galaxies: general – astronomy data analysis – catalogs – photometry: Sloan – photometry:ultraviolet – spectroscopy INTRODUCTIONThe growth and quenching of star formation withina galaxy are central elements of galaxy evolution. Inorder to fully understand this process, the appropriatemethodological tools and physical models must be inplace, including an accurate attenuation law and starformation quenching models. The former is dictated bythe intrinsic properties of the regions studied, such asthe star formation rate (SFR), stellar mass-specific SFR(log[SFR/M ∗ ]) and the chosen sightline, all of whichchange across the face of the galaxy (e.g., Charlot &Fall 2000; Calzetti et al. 2000; Wild et al. 2011; Xiaoet al. 2012; Battisti et al. 2016; Salim et al. 2018). Sim-ilarly, the physical mechanisms that drive the quench- Corresponding author: Nikhil [email protected] a) https://data.sdss.org/sas/dr16/manga/swim/v3.1/ ing of star formation could be constrained by its spatialprogression within a galaxy. For example, both mor-phological quenching and feedback from active galacticnuclei (AGN) produce inside-out quenching (with differ-ent timescales; Martig et al. 2009; Springel et al. 2005),while mechanisms such as ram-pressure stripping (Stein-hauser et al. 2016) would produce outside-in quenching.Therefore, spatially resolved measurements of the re-cent star formation history (SFH) across the faces ofgalaxies are necessary to study quenching in galaxies.The near ultra-violet (NUV) band and the nebular H α recombination line are robust SFR indicators that probedifferent timescales ( ∼
100 Myr for NUV, and ∼
10 Myrfor H α , see Kennicutt 1998; Kennicutt & Evans 2012;Calzetti 2013, for details). Combined with other stellarcontinuum features, they can provide constraints on thespatial progression of quenching. They are also crucialto understanding the relevant attenuation law in starforming regions, including the strength and contribution a r X i v : . [ a s t r o - ph . GA ] S e p of the 2175 ˚A bump (e.g., Calzetti et al. 2000; Battistiet al. 2016, 2017; Molina et al. 2020).The high spatial resolution and narrow NUV filtersrequired for the study of the quenching of star forma-tion are not attainable with the single broad NUV fil-ter and 5 (cid:48)(cid:48) angular resolution of the Galaxy EvolutionExplorer (GALEX; Martin et al. 2005). Therefore,we constructed a catalog of galaxies with Sloan Digi-tal Sky Survey IV (SDSS-IV) Mapping Nearby Galaxiesat Apache Point Observatory (MaNGA; Bundy et al.2015; Yan et al. 2016; Blanton et al. 2017) optical inte-gral field unit (IFU) spectroscopy and archival
Swift
Ul-traviolet Optical Telescope (UVOT; Roming et al. 2005)NUV images. The similar angular resolution ( ∼ . (cid:48)(cid:48)
5) of
Swift /UVOT and SDSS-IV/MaNGA provides a view ofnearby galaxies in both the NUV and optical bands ata spatial resolution of order 1 kpc.The combination of the spatially-matched NUV im-ages and optical IFU maps creates a powerful datasetthat can address a number of astrophysical questions.We have used a subset of the galaxies in this catalogin Molina et al. (2020) to explore the relevant attenua-tion law for kiloparsec-sized star forming regions. Futurework will include using the spectral information to testmodels of galaxies where star formation is being or hasrecently been extinguished.We discuss the sample selection for the Swift +MaNGA(SwiM) catalog and its basic properties in Section 2, anddetail the
Swift /UVOT and MaNGA data reduction inSection 3. The integrated photometric measurementsare described in 4. The process of spatial matching be-tween Swift and SDSS images and MaNGA IFU spec-troscopy is presented in Section 5. We discuss the AGNfraction of the sample in Section 6 and provide notesabout individual objects in Section 6.2. We summarizein Section 7. We assume a ΛCDM cosmology when quot-ing masses, distances, and luminosities, with Ω m = 0 . . = 70 km s − Mpc − . THE SwiM CATALOG2.1.
SDSS-IV/MaNGA and Swift/UVOT
Our sample of galaxies is a subset of the MaNGA sur-vey (Bundy et al. 2015; Yan et al. 2016), which is a spa-tially resolved optical IFU spectroscopic survey includedin the fourth generation of SDSS (SDSS-IV; Blanton Molina et al. (2020) used a previous version of the data re-duction presented here, which does not include errors due to co-variance or matching the
Swift /UVOT images to the resolutionof the uvw2 filter. Additionally, the D n (4000) error bars in thatversion were overestimated by including a factor of 1.98 for cali-bration uncertainties instead of 1.4. However, none of these issuesaffected the results presented in that work. et al. 2017). The main sample of the MaNGA surveywill have ∼ ,
000 galaxies that (1) create a uniformdistribution in stellar mass for M ∗ > M (cid:12) , as ap-proximated via the SDSS i -band absolute magnitude,(2) provide uniform spatial coverage in units of half-lightradius ( R e ), and (3) maximize the spatial resolution andsignal-to-noise for each galaxy (Wake et al. 2017). Thiswork uses the MaNGA Product Launch 7 (MPL-7) ver-sion of the MaNGA sample, which includes all objectsobserved by June 2017, totaling 4706 galaxies.The MaNGA data were taken with hexagonal IFUfiber bundles that contain between 19 and 127 2 (cid:48)(cid:48) fibers,extending over a diameter of between 12 (cid:48)(cid:48) and 32 (cid:48)(cid:48) (Drory et al. 2015). The fiber bundles are insertedinto pre-drilled holes on plates mounted on the Sloan2.5-m telescope (Gunn et al. 2006), and fed into thedual-channel Baryon Oscillation Spectroscopic Survey(BOSS) spectrographs (Smee et al. 2013). The BOSSspectrograph is designed with red and blue arms in or-der to provide continuous wavelength coverage from theNUV to near-IR. Therefore the MaNGA spectra spanthe wavelength range 3 , ,
350 ˚A with a resolvingpower of R ∼ . (cid:48)(cid:48)
5, while the resulting data cubeshave a spatial sampling of 0 . (cid:48)(cid:48)
5. The total exposuretime for each object is set by the sum of the squaredsignal-to-noise ratio: the sum of (
S/N ) must be at least20 pixel − fiber − in the g -band at galactic-extinction-corrected g = 22 AB mag, and 36 pixel − fiber − in the i -band at galactic-extinction-corrected i = 21 AB mag.These criteria result in an average integration time of2.5 hours (Yan et al. 2016).In order to probe both NUV and optical properties,we also make use of Swift /UVOT photometry. UVOT isa 30-cm telescope with a field of view (FoV) of 17 (cid:48) × (cid:48) ,an effective plate scale of 1 (cid:48)(cid:48) pixel − , and three NUVfilters: uvw2, uvm2 and uvw1 (Roming et al. 2005).While the UVOT PSF varies with wavelength (see Ta-ble 1), all three filters give a resolution around 2 . (cid:48)(cid:48)
5, i.e.,similar to the angular resolution of the MaNGA opticalspectra. The detector in the UVOT is a microchannelplate intensified CCD that operates in a photon count-ing mode, which can cause bright sources to suffer fromcoincidence loss (Poole et al. 2008; Breeveld et al. 2010).However, as discussed in Section 3.3, the galaxies in oursample are too faint for this to be a significant problem.2.2.
The Cross-matched Catalog
Table 1.
Swift /UVOT NUV Observation Properties
Central Spectral PSF Median Minimum FaintestWavelength a,b
FWHM a FWHM a Exposure Exposure Magnitude d Filter (˚A) (˚A) (arcsec) (s) (s) ( m AB )uvw2 1928 657 2 .
92 2375 187 22.3uvm2 c .
45 2093 166 22.3uvw1 2600 693 2 .
37 1658 120 21.1 a All filter properties are from Breeveld et al. (2010). b The central wavelength assumes a flat spectrum in f ν . c The uvm2 exposure time statistics only include galaxies with uvm2 im-ages. d The faintest magnitude in our data set for each filter.
We constructed our sample by cross-referencing theMPL-7, i.e., the SDSS Data Release 15 (DR15) (Aguadoet al. 2019), with the
Swift /UVOT NUV archive as ofApril 2018. The UVOT archive is a combination ofstars, active and normal galaxies, and gamma ray burstsources. The objects included in this sample are not al-ways targeted, but instead fall within the FoV of UVOT.We required all objects to have usable, science-readydata cubes from MaNGA and have uvw1 and uvw2 ob-servations in the
Swift /UVOT archive. We do not re-quire uvm2 data for inclusion in the sample. Using thesecriteria, we obtained a sample of 150 galaxies, 87% ofwhich have uvm2 observations. We report the UVOTproperties, and exposure time statistics, and the faintestmagnitude detected in each filter in Table 1.The basic properties of the SwiM catalog galaxies, in-cluding the SDSS classification from DR15 (see Boltonet al. 2012, for details), derived quantities, and observa-tion information from MANGA and UVOT, are storedin a table available in its entirety in the electronic edi-tion of the Astrophysical Journal. We show the formatfor this table in Table 3 in Appendix A.1, and the datamodel for the spatially-matched maps in Tables 4–9 inAppendix A.2. 57% of the sample have either “Galaxy”or no available SDSS classification, while 31% are classi-fied as star-forming, 8% as AGN and 4% as starbursts.The median redshift is 0.033, which corresponds to aluminosity distance of 134.3 Mpc, and a spatial scale of0.6 kpc / (cid:48)(cid:48) . The stellar masses of the galaxies are pro-vided by the NASA Sloan Atlas catalog (NSA; Blan-ton et al. 2011) v1 0 1 based on the aperture-correctedelliptical-Petrosian photometry (for details see Wakeet al. 2017). The stellar masses for this SwiM sample arein the range 8 . ≤ log( M ∗ /M (cid:12) ) ≤ .
11, with a me-dian of log( M ∗ /M (cid:12) ) = 10 .
02. The SFRs are calculatedby the MaNGA Data Analysis Pipeline (Westfall et al.2019) from the H α emission line flux within one effectiveradius, 1 R e , via the scaling relation given in Kennicutt& Evans (2012), after an internal reddening correction assuming the O’Donnell (1994) extinction law. The re-sulting SFRs span 0 (cid:46) SFR(H α ) (cid:46) M (cid:12) yr − , with amedian value of 0.18 M (cid:12) yr − .2.3. Comparison of SwiM catalog to MaNGA
The MaNGA main sample is the combination of threedifferent subsamples: Primary, Secondary, and Color-Enhanced samples (Wake et al. 2017). The catalog pre-sented here contains 63 objects from the Primary sam-ple, 57 from the Secondary sample, and 31 from theColor-Enhanced sample. Therefore, while these individ-ually defined samples from MaNGA can be combined indifferent configurations, we compare the SwiM catalogto the combination of all three, i.e., the “MaNGA mainsample” as defined in MPL-7.We compare the distributions of redshift, g − r color,size, and axial ratio of SwiM catalog galaxies to those ofthe MaNGA main sample in Figure 1. The size is probedby the elliptical Petrosian half-light semi-major axis de-fined in the r -band, and all plotted quantities are takenfrom the NASA-Sloan Atlas. The majority of our galax-ies fall within the range z (cid:46) . R Pet , < (cid:48)(cid:48) , and b/a (cid:38) .
6. The distributions of redshift, R Pet , and b/a in the SwiM catalog are similar to that of the MaNGAmain sample. We also show the distributions of stellarmass, the SFR and the “star-forming main sequence,”i.e., the SFR (as given by H α within 1 R e ) vs. stellarmass for our catalog as compared to the full MaNGAsample in Figure 2. The majority of our galaxies fallwithin the range of log( M ∗ /M (cid:12) ) (cid:38) .
5. The reducedMaNGA emission-line flux maps are already correctedfor foreground Milky Way extinction, but not for inter-nal attenuation in the target galaxy. While this can becomplicated in the UV, most attenuation curves agreeat redder wavelengths including H α . We therefore as-sume the O’Donnell (1994) law, along with an intrin-sic H α /H β ratio of 2.86 (Osterbrock & Ferland 2006,chapter 11), and correct both the MaNGA main sampleand the SwiM catalog for internal attenuation. The cor-rected SFRs are shown in both the top right and bottompanels of Figure 2. The SwiM catalog generally recov-ers the star forming main sequence seen in the MaNGAsample, but is overly populated at the low-mass, low-SFR end, and sparsely populated relative to MaNGA atthe high-mass, high-SFR end. This trend is also quali-tatively seen in the SFR 1–D histogram on the top rightof Figure 2.2.4. Correction Factors to Volume-Limited Weights
The MaNGA sample was selected to have a flat num-ber density distribution with respect to the stellar mass(as approximated by the SDSS i -band magnitude) and .
00 0 .
05 0 .
10 0 . . . . . . . . N o r m a li z e d N u m b e r o f G a l a x i e s MaNGA Main SampleSwiM Sample 0 . . . . . . g − r . . . . . N o r m a li z e d N u m b e r o f G a l a x i e s R Pet , (arcsec)0 . . . . . N o r m a li z e d N u m b e r o f G a l a x i e s . . . . . b/a (Axis Ratio)0 . . . . . . . N o r m a li z e d N u m b e r o f G a l a x i e s Figure 1.
Distribution of the redshift, g − r color, Petrosian half-light radius (derived from r -band photometry), and the r -bandaxis ratio ( b/a ) of the galaxies in the SwiM catalog as compared to the MaNGA main sample. All histograms are normalizedby the total number of galaxies in each data set. In each panel, the blue dashed outline represents the distribution of the SwiMcatalog, while the gray shaded region represents the MaNGA main sample. The SwiM catalog has a similar distribution to theMaNGA main sample for all four properties. We perform a quantitative comparison of the two data sets in Section 2.4. Alldata were taken from the NASA-Sloan Atlas. thus cannot be described as magnitude- or volume-limited. Wake et al. (2017) have provided weights foreach MaNGA target that will statistically correct thesample to that of a volume-limited data set. If our cata-log is consistent with a random sampling of the MaNGAmain sample, then the weights calculated by Wake et al.(2017) should be applicable. We show the g − r vs. stel-lar mass distribution of the entire MaNGA sample asblack contours, with the SwiM catalog overlaid as redpoints in Figure 3. While we sample a large portionof the parameter space, we do not have an even sam-pling of the MaNGA catalog in the color-mass space. We test for a quantitative similarity between our cata-log and a random sample of the same size pulled fromthe MaNGA distribution using the 2–D K-S test (Pea-cock 1983). Specifically, we create 1000 samples of 150galaxies randomly drawn from the MaNGA main sam-ple and compare the resulting 2–D K-S test statistic be-tween that sample and the MaNGA main sample. Weshow the distribution of the test statistic, along withthe measured test statistic for the SwiM catalog in Fig-ure 4; the test statistic is defined such that a largervalue denotes a lower probability that the two distri-butions are quantitatively similar in the observed 2–D M ∗ /M (cid:12) )0 . . . . . . . N o r m a li z e d N u m b e r o f G a l a x i e s − − − − Re[H α ]/( M (cid:12) yr − )]0 . . . . . . . N o r m a li z e d N u m b e r o f G a l a x i e s MaNGA Main SampleSwiM Sample M ∗ /M (cid:12) ] − − − − l og [ S F R R e [ H α ] / ( M (cid:12) y r − ) ] Figure 2.
Top Left:
Same as Figure 1, but for stellar mass. The SwiM catalog has an over density of low-mass objects comparedto the MaNGA main sample.
Top Right:
Same as Figure 1, but for the SFR(H α ) within 1 R e . The L (H α ) measurements havebeen corrected for foreground extinction and internal attenuation in both catalogs, assuming the O’Donnell (1994) law andR V = 3 .
1. The SwiM catalog has an over density of low-SFR objects, and an under density of high-SFR objects compared tothe MaNGA main sample.
Bottom:
SFR(H α ) within 1 R e vs. stellar mass for the SwiM catalog as compared to the full MaNGAsample. The L (H α ) measurements have been corrected for internal attenuation as described above. The catalog recovers thegeneral distribution, except for the high-mass, high-SFR end of the star forming main sequence. parameter space. We find that the SwiM catalog lies atthe 96.1 percentile of the distribution presented in Fig-ure 4, and therefore there is only a ∼
4% chance thata random sample pulled from the MaNGA main samplewould have properties similar to the SwiM catalog. Wetherefore conclude that there are some selection effectsthat make the SwiM catalog different from a randomly-drawn sample and the Wake et al. (2017) weights mustbe scaled in order to statistically correct our catalog tothat of a volume-limited sample. In order to quantify those scaling corrections, webinned both the SwiM catalog and the MaNGA mainsample using a linearly spaced, 10 ×
10 binning schemein the g − r vs. stellar mass space as shown in Figure 5.When plotted this way, the difference between the twodata sets become obvious: the MaNGA main sample hasa strong peak in the high-mass red end of the diagramwith a second weaker peak at the low-mass, blue endof the diagram. Meanwhile the SwiM catalog samplesthe red sequence in a different way (i.e., the band of M ∗ /M (cid:12) ]0 . . . . . . g − r Figure 3.
The g − r color vs. stellar mass distribution of theentire MaNGA main sample, shown as black contours, com-pared to the SwiM sample, represented by red filled circles.Both quantities are measured within the elliptical Petrosianradius apertures and are from the NASA-Sloan Atlas (Blan-ton et al. 2011). The SwiM catalog sample captures thegeneral distribution of the total MaNGA sample over ourmass range of 8 . ≤ log(M ∗ / M (cid:12) ) ≤ . higher density bins with 0 . (cid:46) g − r (cid:46) . ∼ N SwiM /N MaNGA ) is well-described by a binomial distri-bution. We use the un-normalized, binned 2–D distribu-tion (the ratio of the two top panels in Figure 5) to pro-vide scaling factors in addition to the Wake et al. (2017)‘esweights’ or the volume weights for the MaNGA mainsample. We present these ratios as the scale factors foreach galaxy, along with the uncertainty as derived fromthe binomial distribution. To calculate the new weights,one should simply multiply the inverse of this scalingfactor by the ‘esweight’, both of which are provided inthe Swim all catalog file. We do note that there are a .
05 0 .
10 0 . F r e q u e n c y SwiMCatalog
Figure 4.
Histogram of the 2–D K-S test statistic (Peacock1983) for 1000 simulated random samples drawn from theMaNGA main sample. The test statistic for the SwiM cat-alog is denoted by the red vertical line. The SwiM cataloglies at the 96.1 percentile of the distribution, and thereforethere is 3 . significant number of bins, particularly on the edges ofthe 2–D distribution, that the SwiM catalog does notcover. In fact, 8% of galaxies in the MaNGA mainsample fall in bins that have no SwiM catalog galaxies.Thus, while these scaling factors can be used, there issome uncertainty in the correction that is highly depen-dent on the galaxy population of interest. We thereforecaution users to only use scaling factors where the galax-ies of interest lie within the bins populated by the SwiMcatalog. SWIFT /UVOT AND MANGA DATAREDUCTION3.1.
Swift/UVOT Pipeline
All of the UVOT data in our catalog are archival andare drawn from observations obtained from the HighEnergy Astrophysics Science Archive Research Center(HEASARC). The UVOT data are processed using the
Swift
UVOT Pipeline , an automated and updated ver-sion of the subroutine uvot deep.py from the UVOTMosaic program, written by Lea Hagen . github.com/malmolina/Swift-UVOT-Pipeline github.com/lea-hagen/uvot-mosaic M ∗ /M (cid:12) ]0 . . . . . . g − r MaNGA N u m b e r o f G a l a x i e s M ∗ /M (cid:12) ]0 . . . . . . g − r SwiM N u m b e r o f G a l a x i e s M ∗ /M (cid:12) ]0 . . . . . . g − r . . . . . . . ( n S w i M / n M a N G A ) Figure 5. g − r vs. stellar mass. The two sample distributions on thetop show high density as yellow, and low density as shades of purple, denoted by the color bar. The MaNGA main samplehas a strong peak in the high-mass portion of the red sequence and a secondary peak in the low-mass, blue portion of thediagram. Meanwhile the SwiM catalog samples the red sequence in a different way and includes a smaller fraction of low-massblue galaxies. The bottom shows the ratio of the number densities of the SwiM catalog ( n SwiM ) and the MaNGA main sample( n MaNGA ). Both number densities are calculated by normalizing to the total number of objects in each sample (150 and 4498,respectively). If the number densities are equal, the bin color is white, while over-densities in the SwiM catalog are representedby shades of red and under-densities by shades of blue, as denoted by the color bar. The qualitative number density differencesbetween the two catalogs seen in the top two panels are quantified here.
The uvot deep.py subroutine reads in data alreadydownloaded from HEASARC, and follows the basic dataprocessing procedures for UVOT images as described inthe UVOT Software Guide as described below. Theprogram ensures that both the counts and exposuremaps are aspect corrected, reducing the uncertainty inthe defined world coordinate system to 0 . (cid:48)(cid:48)
5. Occasion-ally uvot deep.py will produce errors that cause theexposure map to have 0 or NaN values for small regionsof pixels. This error only occurs in 5 galaxies out ofthe 150 galaxy sample. Furthermore, the error is onlypresent within the galaxy itself for two objects. We pro-vide masks to correct for this issue, which is describedin Appendix A.1.All UVOT images are mosaics of single frames withvery short exposures that are stacked together to pro-duce a deep image. UVOT does allow for differentframe exposure times according to the science goal ofthe observation. However, the UVOT software willnot combine frames with different frame times, asthis would greatly complicate the analysis. Currently uvot deep.py requires the standard full frame exposuretime of 11.0322 ms for inclusion in the final image. Ad-ditionally, all individual frames must be 2 × (cid:48)(cid:48) pixel − . If an individualframe meets both of these requirements and is aspectcorrected, then it is added to the final image. As bothcriteria are standard for non-event mode data, we re-tain the majority of frames in this process. The finalcounts and exposure maps are corrected for large scalestructure (see Breeveld et al. 2010, for details) and haveall bad pixels masked.The Swift
UVOT Pipeline automates uvot deep.py ,and will reduce multiple Swift images in a single exe-cution. In one run, the pipeline will parse data alreadydownloaded from HEASARC and run a modified ver-sion of uvot deep.py . This modified version will alignthe large scale structure correction map (if needed), skipany image that does not meet the requirements of theoriginal uvot deep.py , and store that information in alog file for reference. This process is then repeated foreach image of interest.The distribution of exposure times for the
Swift /UVOTNUV observations are shown in Figure 6. The majorityof objects have exposure times of less than 5000 s inall three filters, with the median exposure time listedin Table 1. The limiting visual magnitude for a 5 σ de-tection for a 5 ks observation is m V = 22 .
45 for uvw2, m V = 22 for uvm2, and m V = 22 . heasarc.gsfc.nasa.gov/docs/swift/analysis N u m b e r o f G a l a x i e s uvw2uvm2uvw1 Figure 6.
The distribution of exposure times for eachSwift/UVOT filter in kiloseconds. The exposure time dis-tribution for uvw2 is shown in as the grey filled histogram,while that of uvm2 and uvw1 are shown in dashed blue andsolid red lines, respectively. The majority of exposures in allthree filters are less than 5 ks. The minimum visual magni-tudes for a 5 σ detection are listed in Section 3.1. ally, the limiting visual magnitude for each filter, giventheir respective median exposures, are m V = 22 .
04 foruvw2, m V = 21 .
53 for uvm2, and m V = 21 . Further UVOT Data Processing
After the images are processed through the
Swift
UVOT Pipeline, they are corrected for both the deadtime and degradation of the detector. The UVOT detec-tor is a microchannel plate intensified CCD, operatingin a photon counting mode. As a result, approximately2% of the full frame time is dedicated to transferringcharge out of the detector. This is corrected by increas-ing the count rate (Poole et al. 2008). Meanwhile, thedecline in count rate due to the degradation of the de-tector is well characterized and provided in the UVOTcalibration documents ; this results in a 2.5% correctionfor the most recent observations.Cosmic ray corrections are not necessary for UVOTimages, due to its operation mode. For each ∼
11 msframe, all individual events are identified, and the cen-troid of the event location is saved. When the final im-age is created, each event is recorded as a count, with itslocation on the image given by the calculated centroiddescribed above. In this regime, a cosmic ray that hitsthe detector will register at most a few counts in a sin- heasarc.gsfc.nasa.gov/docs/heasarc/caldb/swift/docs/uvot gle location, while a stationary astrophysical source willregister thousands of counts. Therefore cosmic rays areincorporated into the background counts as they affectvery few frames.3.3. Coincidence Loss in Swift/UVOT Images
Coincidence loss occurs when two or more photons ar-rive at a similar location within the same ∼
11 ms frame,causing a pile-up that affects the measured count rate.As UVOT operates in a photon-counting mode, this be-comes a significant issue for bright sources. Poole et al.(2008) characterized this effect for a single-pixel detec-tor, while Breeveld et al. (2010) describe coincidence losswhen a point source is in front of a diffuse background(e.g., knots of star formation on top of a galaxy). Thusthe Breeveld et al. (2010) model appears to be the mostappropriate for our data set.However, in order to avoid PSF variation in individualfilters, Breeveld et al. (2010) recommended a minimumaperture size of 3 (cid:48)(cid:48) for all the three NUV filters, corre-sponding to a physical size of ∼ . II regions which have typicalsizes of no more than a few hundred parsecs (Kennicutt1984; Garay & Lizano 1999; Kim & Koo 2001; Hunt &Hirashita 2009). Therefore, the point source plus diffusebackground model described by Breeveld et al. (2010) isnot appropriate.Instead we approach this problem in the spirit ofPoole et al. (2008). The formulated coincidence losscorrections are only valid for point sources, but theeffect is insignificant when the count rate is below10 counts s − pixel − . Across all of the UVOT obser-vations in our sample the maximum count rate is 1 . − pixel − . This translates to a correction of < . Swift/UVOT Sky Subtraction
In order to quantify the local background in eachUVOT image, we construct an annulus using two aper-tures, the inner elliptical aperture with a semi-majoraxis of twice the elliptical Petrosian semi-major axis ( R p ,from NASA-Sloan Atlas; Blanton et al. 2011), and theouter circular aperture with a radius of 4 R p . We mea-sure the background counts with this annulus.We calculate the sky background using a three stepprocess to better describe the local background emission.First, we mask all astrophysical contaminants withinthe annulus. Second, we mask all pixels within the an-nulus that do not have the same exposure time as the galaxy. After this processing, we use the biweight esti-mator (Beers et al. 1990) on the remaining pixels withinthe annulus to calculate the final background counts forthe galaxy. We describe each step in detail below.Given the size of the sky annuli, neighboring brightstars or galaxies may fall within it. To mitigate thiseffect, we run Source Extractor (Bertin & Arnouts1996) and set the contrast parameter (DEBLEND MIN-CONT) to 0, which identifies even the faintest localpeaks inside the annular region and generates a back-ground mask free of these objects.As Swift /UVOT images are made up of of individualframes that are stacked, there may be an uneven expo-sure map around the galaxy of interest. Thus, we onlyconsider sky pixels with exposure times equal to that ofthe center of the galaxy. We therefore mask out pix-els that do not meet this criterion. This step ensuresthat all pixels used for the background calculation havethe same observational properties as the galaxy, givinga more faithful measurement.After applying both of these masks to the sky annu-lus, we measure the background counts on the remainingpixels within the annulus using the biweight estimator(Beers et al. 1990). Even with these masks, we retain asufficient number of sky pixels for each galaxy to havea robust background measurement. Our masked annulistill have a median sky coverage of ∼ ∼
800 square arc-seconds. 3.5.
MaNGA
The MaNGA spectra come fully reduced via theMaNGA Data Reduction Pipeline (DRP; Law et al.2016), which completes both the basic extraction andcalibration steps needed to produce datacubes. Thesedatacubes are then processed using the MaNGA DataAnalysis Pipeline (DAP; Westfall et al. 2019), whichproduces the best-fitting model spectra for all pixelsthat were successfully fit. The DAP also creates the2–D maps of the measured emission line strengths andspectral indices, and measured quantities such as theH α emission within 1 effective radius. All MaNGAmaps are corrected for foreground extinction using the E ( B − V ) value from Schlegel et al. (1998), and assum-ing the O’Donnell (1994) Milky Way dust extinctioncurve with R V = 3 . n (4000) measurements to allow for binnedmeasurements, as described in Section 5.4. INTEGRATED PHOTOMETRICMEASUREMENTSWe present the observed AB apparent magnitudes inthe
GALEX
FUV,
Swift /UVOT NUV and SDSS op-tical filters for all the galaxies in the SwiM catalog inthe SwiM all catalog file. The
GALEX and SDSS mea-surements come from the NASA-Sloan Atlas (Blantonet al. 2011), while those from
Swift /UVOT are mea-sured using our dataset. We measure the integratedphotometry for these bands using the same apertureused by NSA v1 0 1, which is the r -band elliptical Pet-rosian aperture. These integrated magnitudes need tobe corrected for the light lost outside the aperture due to the instrument PSF. This could be a larger effect in GALEX and
Swift /UVOT, as their PSFs are wider thanthe SDSS images. The NASA-Sloan Atlas already cor-rects the
GALEX and SDSS photometry for this issue.Thus, we must apply a similar correction to the UVOTNUV measurements. The corrections are completed ina three-step process: first the galaxy’s r -band 2–D lightprofile, as projected on the sky, is modeled to create asimulated galaxy. The model is then convolved with thePSF of the UVOT NUV filter of interest (uvw2, uvm2,or uvw1) to simulate an observation in that filter. Fi-nally the original r -band integrated measurement fromSDSS is compared with that of the simulated galaxythat was “observed” by UVOT in order to calculate thefraction of light lost. This procedure is completed foreach galaxy and the corrections are applied to the el-liptical r -band Petrosian integrated galaxy UVOT NUVmagnitudes.The photometric measurements are not corrected foreither foreground extinction or internal attenuation, andare not K-corrected. SPATIAL MATCHING OF SDSS DATAPRODUCTS TO
SWIFT /UVOTIn order to enable a joint analysis using SDSS imag-ing, Swift/UVOT imaging, and a MaNGA spectral dat-acube, we have to transform all the images and maps tothe same spatial resolution and spatial sampling. TheSwift/UVOT uvw2 filter has the coarsest PSF (2 . (cid:48)(cid:48) (cid:48)(cid:48) pixels. Moreover, the uvw2photometry tends to have a lower S/N than uvw1 pho-tometry. Thus, we choose to resample all data to matchthe sampling in the uvw2 band. We describe this processbelow for all of the quantities of interest.Given the relatively low S/N of the UV data, furtherbinning is likely necessary to make use of this dataset.We thus strive to present the final data in a format thatwould facilitate binning of the end user’s choice.5.1. SDSS and Swift/UVOT Images
The uvw2 images are kept in the original format. Weconvolve each of the uvw1, uvm2, and SDSS u, g, r, i, z images with an appropriate kernel to match the PSF in1the uvw2, and then reproject them to the uvw2 sam-pling. We set the convolution kernel to a 2D Gaussianwith σ = (cid:115) (FHWM uvw2 − (cid:15) ) − FWHM , (1)where FWHM x represents the FWHM of the PSF ofthe corresponding filter, and (cid:15) is a correction term wewill discuss below. In general, PSFs are not Gaussians;they are closer to Moffat functions, which have addi-tional power-law wings. However, given that most ofour sources are faint, only the core and not the addi-tional wings are strongly detected. We therefore ap-proximate the PSFs as Gaussians. The FWHM for uvw1and uvm2 are given in Table 1. For SDSS images, weuse a FWHM of 1 . (cid:48)(cid:48)
4, typical of the seeing condition forthe SDSS data. Small variations in the seeing will notsignificantly change the kernel size.We then reproject the convolved images for uvw1,uvm2, and SDSS u, g, r, i, z -bands to the pixel positionsin the uvw2 image using the flux-conserving sphericalpolygon intersection algorithm. This is achieved by us-ing the reproject.reproject exact function in astropy (As-tropy Collaboration et al. 2018).This reprojection process brings an additional broad-ening to the effective PSF. With simulations, we foundthat this additional PSF broadening varies dependingon the amount of shift in the pixel grid. For uvw1 anduvm2 with 1 (cid:48)(cid:48) pixels, the broadening can vary from 0 to0.1 (cid:48)(cid:48) in the σ of the PSF for a Gaussian with a FWHM of2.92 (cid:48)(cid:48) . A polynomial fit as a function of fractional pixelshifts could predict this broadening effect to better than0.001 (cid:48)(cid:48) . Thus, for each galaxy, for uvw1 and uvm2 fil-ters, we apply the corresponding correction factor ( (cid:15) )in Equation 1 depending on the amount of fractionalpixel shift between the pixel grids. For MaNGA andSDSS, due to their smaller pixels, the PSF-broadeningeffect is smaller and shows much less variation with frac-tional pixel shifts. For SDSS, the broadening variesfrom 0.03 (cid:48)(cid:48) to 0.04 (cid:48)(cid:48) with a median around 0.0376 (cid:48)(cid:48) for σ .For MaNGA, the broadening varies from 0.03 (cid:48)(cid:48) to 0.05 (cid:48)(cid:48) with a median around 0.0419 (cid:48)(cid:48) for σ . In both of thesecases, we use the median correction for all galaxies. Theamount of remaining error is at most 0.012 (cid:48)(cid:48) in σ and0.028 (cid:48)(cid:48) in FWHM. This is less than 1% of the final PSFwidth and is negligible, as the measurement error of thePSF is usually larger than this.The exposure maps and masks are also processed inthe same way through the convolution and reprojection.Masked pixels are ignored in the computation. The finalprocessed mask is rounded to 0 or 1 using a threshold of 0.4 . If more than 40% of a pixel area comes from badpixels, then the final pixel is considered bad (mask=1).5.2. Spatial Covariance for Swift/UVOT and SDSSimages
The convolution and reprojection introduce covari-ance between neighboring pixels. This not only meansthe final uncertainty is larger than that computed basedon error propagation without including covariance, butit also means that the final images contain covariance. Ifone were to bin the final images further, one would needto take this covariance into account when estimating thephotometric errors.We compute the final covariance matrix in the follow-ing way. For SDSS and Swift uvw1 and uvm2 images,we start by constructing a covariance matrix with onlydiagonal elements containing the variance of all the pix-els, basically assuming all pixels are independent of eachother. We refer to this matrix as G . Then we constructthe matrices corresponding to the convolution and re-projection processes, which we refer to as W and Z ,respectively. For an image with an initial size N × N and a final reprojected size M × M . The W matrix hasthe shape of N × N , and the Z matrix has the shape of M (rows) × N (columns) . The final covariance matrixcan then be computed as C = ( Z × W ) × G × ( Z × W ) T (2)The final covariance matrix has a size of M × M .The diagonal elements of C contain the variance for thefinal images. We then take the square root to obtain the1- σ uncertainty map.The final map still contains covariance. This can becharacterized by the correlation length. We recast thecovariance matrix, C , to the correlation matrix, ρ , bycomputing ρ ij = C ij / (cid:112) C ii C jj for all i and j from 1 to M . The correlation matrix has all diagonal elementsequal to 1. Other elements give the correlation strengthbetween each pair of pixels. Only pairs of pixels withsmall spatial distance have non-zero values. Followingthe example given by Westfall et al. (2019) and usinggalaxy, MANGID 1-44745, as a typical example, we findthe correlation can be well fit by a Gaussian functionof the pair-wise distance. The scale parameter of theGaussian is 0.925 for uvm2, 0.907 for uvw1, and 1.538pixels for SDSS filters. These are shown in Figure 7.In lieu of providing the full covariance matrix, herewe provide a functional form as an approximation forthe effect of the covariance. We construct a mock mapwith unity errors in all pixels and bin N pixels togetherand propagate the errors in two ways, with and with-out taking covariance into account. We plot their ratio,2 − − − − ρ u v m σ uvm2 =0.925 − − − − ρ u v w σ uvw1 =0.907 − − − ρ S D SS σ SDSS =1.538 D (pixels) − − − − ρ M a N G A σ MaNGA =1.48
Figure 7.
The spatial correlation strength in the final mapsas a function of distance between pixels, for galaxy 1-44745.The data points and error bars show the mean and stan-dard deviation for each distance bins. The curves show theGaussian fits over the plotted range (with amplitude fixed to1 and centered at 0). The four panels, from top to bottom,show the results for uvm2, uvw1, SDSS images, and MaNGAmaps, respectively. The correlation can be well described byGaussian functions with the scale indicated in the legend. f covar = σ covar /σ no covar , as a function of the numberof pixels binned together. Because the correlation is be-tween neighboring pixels, the effect of the covariance de-pends on the shape of the bin. We compute two extremecases to bracket different situations. The maximum co-variance case is for a bin that is nearly a square, similarto the case of Voronoi binning. The minimum covari-ance case is for a long rectangular bin (with a maximumlength of 27 pixels), similar to the case of annular bin-ning. Figure 8 shows how f covar scales with the numberof pixels in the bin, under these two cases. We fit themusing the same functional form as suggested by Huse-mann et al. (2013). The fit results are listed below. Forlarge N bin , the scaling factor asymptotes to a constant value.UVM2: f covar = (cid:40) .
61 log( N bin ) if N bin < .
16 otherwise (3)UVW1: f covar = (cid:40) .
59 log( N bin ) if N bin < .
12 otherwise (4)SDSS: f covar = (cid:40) .
207 log( N bin ) if N bin < .
41 otherwise (5)There is a caveat for the uncertainty maps of the SDSSimages. Many inverse variance images in the NASASloan Atlas contain features due to satellite tracks. Butthese features do not appear in the flux images. If theyare not masked in the inverse variance images, theywould become the dominant feature in the final uncer-tainty maps. We applied additional masking to removethese features in our processing.5.3.
MaNGA Emission Line Maps
For emission line fluxes and equivalent widths (EWs),we start from the Gaussian-fitted 2-D emission line fluxand EW maps generated by the MaNGA DAP (Westfallet al. 2019). Because the EW is a ratio between the lineflux and the continuum, all the convolution and repro-jection steps should be carried out on the line flux andcontinuum images first, before deriving the EW at theuvw2 resolution and sampling positions.By taking the ratio between the line flux and EWmaps from DAP, we first derive the continuum mapat the original MaNGA resolution and sampling posi-tions. We convolve the flux and continuum maps withan appropriate Gaussian kernel to match the PSF ofthe
Swift /UVOT uvw2 filter. The maps were thenreprojected to the
Swift /UVOT uvw2 pixel positionsusing the reproject.reproject exact function in astropy.Masked pixels are ignored in the computation and thefinal masks are produced by convolving and resamplingthe mask, which is then rounded to 0 or 1 to produce thefinal mask. If users of the catalog wish to bin the datafurther, the best approach is to divide the flux mapsby the EW maps and then bin flux and continuum sep-arately before computing the binned EW. We providethese measurements for all 22 emission lines providedby the DR15 version of MaNGA DAP (Westfall et al.2019).3 . . . f c o v a r b uvm2 = . . . f c o v a r b uvw1 = . . . f c o v a r b SDSS = N bin f c o v a r b MaNGA = Figure 8.
Scaling factor between errors propagated withand without including covariance, as a function of the num-ber of pixels binned together. The black dots show thesimulated difference between the two sets of errors for thetwo extreme cases of binning, as described in Section 5.2.The red curves show the fit using the functional form of f covar = 1+ b log( N bin ). The four panels, from top to bottom,are for uvm2, uvw1, SDSS, and MaNGA maps, respectively. MaNGA Spectral-Index Maps
Spectral-index maps, including the Lick indices andD n (4000) maps, are very useful for constraining the stel-lar populations of galaxies. Producing these propertiesat the uvw2 resolution requires more care as they allrepresent ratios of flux densities.For example, D n (4000) is the ratio of the average fluxdensity per unit frequency ( f ν ) between a red band(4000-4100˚A) and a blue band (3850-3950˚A). Simplypresenting the final convolved D n (4000) maps is not suf-ficient to allow further binning. Therefore, we have re-measured the blue and red band flux densities in thedata using the DRP LOGCUBE files. We then con-volved and resampled them to the Swift uvw2 PSF and pixel coordinates following the same process as done tothe emission line flux maps. Instead of presenting theirratio in the final file, we provide the two flux densitymaps for each galaxy. The variance maps are also pro-cessed in the same way, and the final 1-sigma uncer-tainty maps are presented for each flux density map.The mask for D n (4000) is derived from the DAP maskfor D n (4000), and is processed in the same way as thatfor emission line maps.In order to allow flexibility in further binning of theresulting map, we chose a different definition of the Lickindices from the standard definitions adopted by theMaNGA DAP (see Section 10 of Westfall et al. 2019).The standard definitions given by Trager et al. (1998),define the continuum as a sloped line between the twoside bands. It then integrates the fractional deficit influx over the central band. Because the ratio betweenflux and continuum is inside the integral and the denom-inator is not a constant, this definition is inconvenientfor spatial binning. Under this definition, proper spa-tial binning would require one to go back to the spectraldatacube, bin the spectra, and then remeasure the in-dices.To allow more convenient spatial binning, we adoptan older definition of Lick indices, which is first used byBurstein et al. (1984) and described in detail by Faberet al. (1985). Instead of a sloped continuum, this defini-tion adopts a constant as the continuum in calculatingthe integral of the fractional flux deficit, as in I a = (cid:90) (cid:18) − f λ f C0 (cid:19) dλ in ˚A − . (cid:18) λ (cid:90) f λ f C0 dλ (cid:19) in magnitudes(6)Here, f λ is the total flux density per unit wavelength inthe index band, f C0 is the continuum flux density perunit wavelength and ∆ λ is the width of the index band.The value of ∆ λ for all Lick indices is given in Table 6of Appendix A. Here f C0 is not a function of λ and canbe taken out of the integral. Therefore, Equation 6 canbe simplified to I a = ∆ λ − F I f C0 in ˚A − . (cid:18) λ F I f C0 (cid:19) in magnitudes (7)where F I is the integrated flux in the index band ( F I = (cid:82) f λ dλ ). With the continuum flux density taken out ofthe integral, one could get binned Lick indices without4having to go back to the spectra, as long as both thecontinuum and the integrated flux in the passband isprovided for each spaxel, and the constant continuum isdefined to be strictly additive when spectra are addedtogether. To get binned Lick indices, one would simplybin both the map of the continuum and the map of theflux before dividing them.To define the continuum that is strictly additive, wefirst define the average flux density in the red and theblue bands as the following, f R = 1( λ − λ ) (cid:90) λ λ f λ dλ (8) f B = 1( λ − λ ) (cid:90) λ λ f λ dλ (9)where λ , λ , λ and λ are the end points of thered and blue bands. We define the linear continuum fluxas, f C0 = ( f R − f B ) λ IM − λ BM λ RM − λ BM + f B (10)where λ RM and λ BM are the mid-points of the red andthe blue bands and λ IM is the mid-point of the indexband. This continuum level would be strictly additivewhen multiple spaxels are combined.This definition of the Lick indices is also quite usefulwhen it is applied to composite stellar population mod-els. One simply has to measure the continuum and theintegrated flux in the index band for each simple stel-lar population with a certain age and metallicity. Whenconstructing the composite models, the Lick indices canbe computed by adding the flux and the continuum sep-arately before computing the index for the compositemodel.Under this definition, we need only to provide themaps for F I , f C0 , the associated uncertainty, and masksat the Swift UVOT resolution and sampling positions.For each spaxel, we take the spectrum from the MaNGADRP LOGCUBE file, subtract from it the best-fitemission-line spectrum, then transform it to the rest-frame given the redshift and the stellar velocity providedby DAP. We measure for each spaxel the index bandintegral and continuum for each Lick index, using thepassbands of the MaNGA DAP (Westfall et al. 2019).We then convolve the resulting maps to the same PSFas the Swift uvw2, and reproject it to the Swift uvw2pixel positions.5.5. Spectral Resolution for the Lick indices
The values of the Lick indices depend on the spec-tra resolution of the spectra. the stellar velocity dis-persion of the target, and “beam smearing” resulting from any systematic variation in stellar velocities withinthe aperture used for the measurement. The traditionalLick-index system is defined for a constant instrumentalresolution of 8.4 ˚A FWHM, and a fixed stellar velocitydispersion. This instrumental resolution is too coarse forthe higher resolution spectra from SDSS and MaNGA.We also argue that it is undesirable to smooth the datato match a fixed velocity dispersion, or to make an ap-proximate and model-dependent correction using a fit-ting formula based on an object’s velocity dispersion. Tocomplicate the matter further, the instrumental resolu-tion of the BOSS spectrograph varies with wavelength,whether it is specified in wavelength units or velocityunits. A better approach is, therefore, to smooth themodel spectra to match the combined effective disper-sion in the data, which includes both the instrumentaldispersion and the stellar velocity dispersion. Therefore,we provide, as part of our data products, the maps ofthe combined dispersion for each Lick index.The combined dispersion is constructed by adding inquadrature the stellar velocity dispersion with the in-strumental dispersion for each spaxel and each index.The instrumental dispersion is taken at the center ofthe index band. For the convolution and reprojectionprocess, we apply the square of the combined dispersionand weight the computation by the integrated flux inthe index band. We also propagate the uncertaintiesand provide the associated masks.5.6.
Spatial Covariance in the MaNGA maps
The uncertainty maps for MaNGA emission line andspectral index properties are also produced by takinginto account the covariance. In contrast to the Swift andSDSS images, the MaNGA maps come with significantcovariance between spaxels. This means the G matrixin Eqn. 2 contains non-zero off-diagonal elements. Toconstruct G for MaNGA, we start by creating a correla-tion matrix ( ρ ) using a correlation scale of 1.92 spaxels,as provided by Westfall et al. (2019) (see Fig. 8 in thatpaper). We then multiply this matrix by the reformat-ted variance maps of each MaNGA property to build thecovariance matrix, G , by G ij = ρ ij G ii G jj . The rest of thesteps are similar to those described in Section 5.2.The final correlation in the resulting maps has a scalefactor of 1.48 pixels, as shown in the last panel of Fig-ure 7. This is for an example galaxy with MANGAID,1-44745. For different galaxies, with different PSFs inthe MaNGA data cube, the results could differ slightly.If one wants to bin the map further, to include covari-ance in the error propagation, one should use the f covar given below to scale the error propagated without co-variance. The fits for this covariance factor are shown5in the last panel of Figure 8 f covar = (cid:40) .
156 log( N bin ) if N bin < .
31 otherwise (11)5.7.
Organization of the Maps
Most of the per-galaxy data we provide are in the formof maps. Given that all the images and maps are con-volved to the same PSF and reprojected to the samesampling positions, they also share the same World Co-ordinate System. We group these images and maps intoseveral groups: broadband images, emission line fluxes,Lick indices, and D n (4000). Each group contains imagesin multiple broadband filters, multiple emission lines, ormultiple indices. We stack all the images and maps ineach group together in a 3D array with different channels(layers) corresponding to different filters/lines/features.The uncertainty and masks are also presented in corre-sponding 3D arrays. All these arrays are presented asdifferent header data units (HDU) in a FITS file withthe extension name indicating the group and whetherthe file contains measurements, uncertainties, or a mask.The detailed data model is given in Appendix A.1.5.8. Uncertainties of MaNGA-based measurements
We provide formal errors associated with the data inthis value-added catalog (VAC). However, these for-mal errors could be underestimated or overestimated.It is much more reliable to use repeated observationsto evaluate the uncertainty. Using repeated observa-tions, the MaNGA team (Belfiore et al. 2019) evaluatedthe uncertainty associated with the emission line fluxmeasurements, and found the actual uncertainty is onlyslightly larger than the formal error, by 25% for H α , andby similar levels for other strong emission lines. There-fore, to get a realistic error estimates, one simply has tomultiply the H α flux and EW errors by 1.25. For moreinformation on this process, please see Belfiore et al.(2019).Similarly, for D n (4000), Westfall et al. (2019) showedthat a realistic error estimate based on repeated obser-vation is about 1.4 times that of the formal error. Fluxcalibration systematics could be one of the contribut-ing factors. However, here, we do not scale our errorestimates for D n (4000) because we are presenting theerrors associated with the red band and the blue bandseparately. We recommend that the users propagate theformal errors to the final error for D n (4000) and thenmultiply it by 1.4.For Lick indices, one could also derive these scalingfactors. Westfall et al. (2019) found that the error scal-ing factor is 1.2 for H β and H δ A absorption EW, 1.6 for the Fe5335 index, 1.4 for the Mgb index, and 1.5 for theNaD index. ACTIVE GALAXIES IN THE SwiM CATALOG6.1.
Identification of AGN
Because our only requirement for inclusion in the sam-ple is the availability of
Swift /UVOT data, there areAGNs present in the catalog. We present our AGN iden-tification method in this section, while providing noteson individual objects in Section 6.2. We searched forthese AGNs using a three step process. We first usedthe spatially-resolved BPT diagrams from MaNGA toidentify all objects with at least 10 MaNGA 0.5 (cid:48)(cid:48) -pixelswithin 0 . R e that fall within the Seyfert, LINER, orAGN regions of the [S II ]/H α , [N II ]/H α or [O I ]/H α BPTdiagrams. These are combined with all objects that havean SDSS classification of “AGN”, “QSO” or “Broadline”to make the subset of 47 AGN candidates that are usedin steps two and three.Second, we identify all objects with detectable X-rayemission by utilizing archival data from the
Swift
X-Ray Telescope (XRT; Burrows et al. 2000), as all UVOTimages have a corresponding XRT observation. The
Swift /XRT data come from the UK
Swift
Science DataCentre. The X-ray properties of all detected objectsare obtained via the automated spectral fitting web tool(Evans et al. 2009), while the upper limits either comefrom the automated light curve web tool (Evans et al.2007, 2009) or the XRT point source catalog (1SXPSC;Evans et al. 2014). We have 100% coverage of our sampleand 17 galaxies have detectable X-ray emission. Whilehard X-ray emission can be indicative of AGN, both low-and high-mass X-ray binaries (XRBs) can also producehard X-rays. The PSF of
Swift /XRT is 18 (cid:48)(cid:48) at 1.5 keV,which encompasses the entire galaxy for almost all ofthe objects in our sample. Thus, the XRBs present inthe galaxy are contributing to the observed X-ray emis-sion. The contribution from low-mass XRBs is propor-tional to the stellar mass, while the contribution fromhigh-mass XRBs is proportional to the SFR (Fabbiano2006; Lehmer et al. 2010). Therefore the observed X-ray emission must be stronger than the contributionfrom XRBs in order to be ascribed to an AGN. Wecalculate the XRB contribution using the stellar massand SFRs for the galaxy presented in SwiM all cata-log file, and the L galHX calculation given in equation (3)of Lehmer et al. (2010). We report the photon indexfor the assumed power law, the unabsorbed 0.3–10 keVluminosities of the galaxies with detectable X-ray emis-sion, and the XRB contribution from the galaxy in Ta-ble 2. While there is a well-defined relationship betweenthe H α and the X-ray luminosities in AGN, e.g., Panessa6et al. (2006), the MaNGA maps do not include the broadH α component. In order to complete this test, detailedmeasurements of the broad components of AGN spectramust be made separately, which is beyond the scope ofthis paper. Due to short exposure time in the X-rayfor the undetected objects, their upper limits are in therange L (0 . ∼ –10 erg s − and are not verymeaningful. Thus, we do not report them in this paper.The calculated XRB contribution presented here islimited by several effects: (1) the contamination of theobserved H α emission from the potential AGN, and (2)the SFR is calculated within 1 R e . The first effect willincrease the expected H α contribution, and thus arti-ficially increase the fraction of hard X-ray emission ex-plained by XRBs. While the second effect does not allowus to calculate the total X-ray emission from the galaxy,we are only interested in the nuclear emission, which isenclosed in the chosen aperture.The 50% light radius used for the SFR calculation isbased on the SDSS r -band, which peaks around 6200˚Aand encompasses the H α emission line. We thereforeassume that the r -band 50% light radius can be approx-imately applicable to H α . However, even if we dou-ble the SFR, using the fact that H α luminosity is di-rectly proportional to the SFR (i.e., Kennicutt & Evans2012), the XRB contribution to the hard X-ray luminos-ity changes by at most 50%; the observed 0.3–10 keVluminosities are often more than a factor of 2–3 largerthan the quoted L galHX . Therefore, despite the large PSFof Swift /XRT, the resulting measurements can still beused to identify an AGN. We conclude that AGNs arecontributing to the observed hard X-ray emission in 12out of the 17 objects.In the final step, we compared the emission from thenuclear resolution element (circular aperture with a di-ameter of 2 . (cid:48)(cid:48)
92 to the star forming models from Kewleyet al. (2006) for the [S II ]/H α , [N II ]/H α and [O I ]/H α BPT diagrams, as shown in Figure 9. We also plotthe composite object line from Kauffmann et al. (2003)and the LINER and Seyfert boundary lines from Kew-ley et al. (2006). We required all integrated emissionline measurements to have S/N > Notes on Individual Objects – This galaxy is NGC 1149, and hasan upper limit on the X-ray luminosity of L (0 . < × erg s − , as reported inthe 1SXPSC. The measured emission line ra-tios are consistent with non-star-forming ioniza-tion mechanisms in all three BPT diagrams, butthe [S II ]/H α emission line ratio fall within theLINER-region section of the diagram, which arenot necessarily indicative of an AGN. The centralpixel from MaNGA shows extremely weak emis-sion lines and no AGN-like continuum. We can-not confidently conclude that NGC 1149 harborsan AGN. – This galaxy has a hard to soft X-rayratio close to zero and all three BPT diagramsare consistent with star formation. We thereforeconclude that this object does not harbor an AGN. – This galaxy is UGC 4056, and hasan upper limit on the x-ray luminosity of L (0 . < × erg s − , as reported inthe 1SXPSC. The measured emission line ra-tios are consistent with non-star-forming ioniza-tion mechanisms in all three BPT diagrams, butthe [S II ]/H α emission line ratios fall within theLINER-like region of the BPT diagram, which isnot necessarily indicative of an AGN. Given theweak X-ray emission and ambiguous emission lineratios, we cannot confirm that UGC 4056 harborsan AGN. – This galaxy is Mrk 290, a well-knownAGN with strong X-ray emission. The nuclear re-gion is not fit by the MaNGA DAP so it is notincluded in Figure 9. – This galaxy is identified as “Star-forming” by the SDSS collaboration, but exhibitsstrong X-ray emission and has Seyfert-like emis-sion line ratios in the central MaNGA pixel. How-ever, this signal is washed out by the 2 . (cid:48)(cid:48)
92 apertureused to construct the [S II ]/H α and [O I ]/H α BPTdiagrams. Given the conflicting evidence but verystrong X-ray emission, we conclude that this ob-ject may harbor a low-luminosity AGN. – This object has an upper limit on itsX-ray luminosity of L (0 . < erg s − ,and the spectra are not fit in the nuclear regionof the IFU by the MaNGA DAP. This object islisted as a black hole candidate based on HubbleSpace Telescope imaging by Greene & Ho (2007),but we cannot confirm that this object harbors anAGN.7
Table 2.
X-Ray Detections in the SwiM catalog N Ha L obsa L galHXb ConsistentObject I.D. Γ a (cm − ) (10 erg s − ) (10 erg s − ) with AGN?1-37336 2 +4 − < × +10000 − c . +0 . − . < × ±
800 ... Yes d . +0 . − . < × +50 − − . +0 . − . < × +300 − ± +5 − × +3 × − . ± . < × +9 − e . × ± . ± . < × +90 − f . ± . ± × +500 − ± < × < × g . +0 . − . < × ±
40 0.88 Yes1-419607 7 +30 − < × < × g . +0 . − . < × ±
20 0.01 No c . +0 . − . . +0 . − . × ± . +0 . − . < × +800 − . ± . < × ± . ± . +4 − × ±
200 0.78 Yes1-620993 1 . ± . × +80 − a The “photon index,” Γ is the power-law index of the X-ray photon number spectrum, i.e. N ( E ) ∝ E − Γ ( E is the photon energy and N ( E ) is the number of photons per unit energy).It is used along with the column density, N H , are used to calculate the unabsorbed 0.3–10 keV luminosity, as described in Section 6. b The hard X-ray luminosity contribution from XRBs, based on the stellar masses and SFRsfrom the SwiM all catalog file, and equation (3) of Lehmer et al. (2010). c We do not conclude this object harbors an AGN due to the lack of spectroscopic evidenceand the fact that the bulk of the X-ray emission in in the soft (0.3–2 keV) band. Section 6.2for details. d The object 1-90242 is Mrk 290, which is a known AGN (i.e., Bentz & Katz 2015). We cannotreport the calculated L galHX from MaNGA data as the DAP cannot handle the strong AGNcontribution and thus masks most of the galaxy. e Object 1-155975 has detectable X-ray emission but the spectrum cannot be automaticallyfit by the 2SXPSC software. In this case, a fixed photon index of 1.7 and Galactic N H isassumed f This object resides in a large cluster, and thus the strong X-ray detection could be gasassociated with the cluster. See Section 6.2 for details. g This object has weakly detected X-ray emission, but the conversion to unabsorbed luminosityresults in an upper limit. We do not conclude this object harbors an AGN, due to thepoorly constrained X-ray emission and conflicting spectroscopic evidence of AGN activity.See Section 6.2 for details. – This galaxy is NGC 6166C, the cen-tral cD galaxy of the Richness Class 2 cluster Abell2199. The observed X-rays in the 0.3–10 keV bandcould therefore be from to the system’s intraclus-ter medium. Additionally, the nuclear [O I ] emis-sion is very weak, with a S/N <
3. In combinationwith the weak X-ray emission and the very weak[O I ] emission, we cannot confidently conclude thatNGC 6166C harbors an AGN. – This galaxy has an upper limit onthe X-ray luminosity of L (0 . < × erg s − , as reposrted in the 1SXPSC. Theobject’s measured emission line ratios are consis-tent with non-star-forming ionization mechanismsin all three BPT diagrams, though the [O I ]/H α and [S II ]/H α ratios are in the LINER region ofthe corresponding diagrams. We therefore do notconclude that this object harbors an AGN. – The X-ray luminosity of this galaxyis L (0 . < × erg s − , as reported inthe 1SXPSC. The measured emission line ratiosare again consistent with non-star-forming ioniza-tion mechanisms, but the [S II ]/H α emission lineratio fall within the LINER region of the BPT di-agram, which is not necessarily indicative of anAGN. In addition, the [O I] emission is not mea-sured in all pixels within the central 2 . (cid:48)(cid:48)
92 aperture,which prevents a robust AGN identification. Wetherefore cannot conclude that this object harborsan AGN.8 − α ) − . − . . . . l og ( [ O III ] / H β ) − α ) − − α ) Figure 9.
The [S II ]/H α , [N II ]/H α and [O I ]/H α BPT diagrams for the nuclear resolution element of the AGN candidates thathave a S/N > – This galaxy has strong X-ray emissionin the 0.3–10 keV band, but the H β emission lineflux has a S/N < – This galaxy has an upper limit onthe X-ray luminosity of L (0 . < × erg s − , as reported in the 1SXPSC. Thegalaxy shows Seyfert or AGN-like emission line ra-tios in the [O I ]/H α and [N II ]/H α diagrams, butits [S II ]/H α ratio is consistent with excitation bystar formation. Given the conflicting evidence andnon-detection in the X-ray, we cannot confirm thepresence of an AGN in this galaxy. – This galaxy has poorly-constrained X-ray emission as measured by XRT. While there is“observed” emission, the unabsorbed X-ray flux isstatistically consistent with zero. The object alsodisplays Seyfert or AGN-like [N II ]/H α ratios, butfalls within the LINER-like region of the [S II ]/H α and [O I ]/H α diagrams. The central pixel of theMaNGA data shows very weak emission lines andno AGN-like continuum. We cannot confidentlyconclude that this object harbors an AGN. –This galaxy is not identified as anAGN by its emission line, but has X-ray emissiondetected by XRT. However, the galaxy resides in the Coma cluster, so the X-ray emission could be aresult of the hot gas from the cluster and not neces-sarily indicative of an AGN. We therefore concludethe object does not harbor an AGN. – This galaxy shows no spectroscopicsignatures of an AGN, and the X-ray emission ispoorly constrained. While X-rays are detected,the unabsorbed X-ray emission is still statisticallyconsistent with zero. Therefore we do not concludethat this object harbors an AGN. –This galaxy is identified as Star-forming by the SDSS collaboration, but exhibitsstrong X-ray emission. The measured emissionline ratios are consistent with non-star-formingionization mechanisms in all three BPT diagrams,but the [O I ]/H α and [S II ]/H α emission line ratiosfall within the LINER-like region of the diagram,which are not necessarily indicative of an AGN.Given the strong X-ray detection, we concludethis object harbors an AGN. –This galaxy is IC 4227, and the up-per limit on its x-ray luminosity, L (0 . < × erg s − as reported in the 1SXPSC doesnot rule out the presence of an AGN. Addition-ally, the measured emission line ratios are consis-tent with AGN or Seyfert-like ionization mecha-nisms in all three BPT diagrams, and the centralMaNGA pixel shows strong narrow emission lines9consistent with a Seyfert 2 spectrum. We concludethat IC 4227 harbors an AGN. SUMMARYIn this paper, we present the SwiM value added cata-log, which consists of 150 galaxies in the MaNGA mainsample that have also been observed by
Swift /UVOTin both the uvw2 and uvw1 filters. In this dataset,we provide the integrated photometry in the UVOTuvw2, uvm2, and uvw1 filters measured consistentlywith the SDSS and
GALEX photometry provided bythe NSA, along with selection weights and scaling factorsfor correcting back to a volume-limited sample. We alsopresent resolution- and sampling-matched SDSS andSwift images, along with matching maps for emission-line and spectral indices based on MaNGA spectra. Er-rors have been propagated taking into account of the co-variance throughout the reduction. We also present thecorrelation between pixels in the resulting maps. All theimages and maps have a final resolution of 2 . (cid:48)(cid:48)
92 FWHMand are sampled on 1 (cid:48)(cid:48) A. SwiM VAC DATA MODELA.1.
Catalog Data Model
This appendix provides the SwiM VAC data model for the SwiM all catalog file. The catalog file holds basicproperties of the galaxies included in the SwiM catalog, as well as the integrated
GALEX , Swift /UVOT, and SDSSphotometry. The names and contents of each extension in this file is given in Table 3.
Table 3.
SwiM Catalog Data Model
Column Units Description
MANGAID . . . MaNGA ID for the object (e.g 1-210754)
PLATE . . . Plate ID for the object
IFUDSGN . . . IFU design ID for the object (e.g. 12701)
MNGTARG1 . . . MANGA TARGET1 maskbit for galaxy target catalog
MNGTARG3 . . . MANGA TARGET3 maskbit for galaxy target catalog
NAME . . . Galaxy Name
SDSS CLASS . . . SDSS DR15 object classification
EBV . . . E ( B − V ) value from Schlegel et al. (1998) dust map for this galaxy RA deg Right-ascension of the galaxy center in J2000 DEC deg Declination of the galaxy center in J2000
NSA ELPETRO PHI deg Position angle (east of north) used for elliptical apertures
NSA ELPETRO TH50 R arcsec Elliptical Petrosian 50% light radius (semi-major axis) in SDSS r-band
NSA ELPETRO THETA arcsec Elliptical Petrosian radius (semi-major axis) in SDSS r-band
NSA ELPETRO BA . . . Axis ratio used for elliptical apertures
NSA ELPETRO MASS h − solar masses Stellar mass from K-correction fit for elliptical Petrosian fluxes NSA Z . . . Heliocentric redshift from the NASA-Sloan Atlas
NSA ELPETRO FLUX nanomaggies Elliptical SDSS-style Petrosian flux the
GALEX and SDSS filters inbands [FNugriz] (using r-band aperture)
NSA ELPETRO FLUX IVAR nanomaggies − Inverse variance of
NSA ELPETRO FLUX [FNugriz]
SWIFT ELPETRO FLUX nanomaggies Elliptical SDSS-style Petrosian flux in bands [uvw2, uvm2, uvw1](aperture corrected using r-band aperture)
SWIFT ELPETRO FLUX IVAR nanomaggies − Inverse variance for
SWIFT ELPETRO FLUX [uvw2, uvm2, uvw1];if there is no uvm2 measurement that element is − SWIFT EXPOSURE sec Exposure times for Swift/UVOT bands [uvw2, uvm2, uvw1];if there is no uvm2 measurement that element is − APERCORR . . . Aperture correction factor f a /f b for Swift/UVOT bands [uvw2, uvm2, uvw1]; f a and f b are the r-band integrated fluxes (of the mock galaxy) before and afterthe Swift/UVOT PSF convolution (see Section 4) SFR 1RE . . . Dust corrected log(SFR/1 M (cid:12) yr − ) using H α flux within 1 effective radius reportedin MaNGA DAP (see Westfall et al. 2019, and Section 2.3 of this work) SCALING FACTOR . . . Scaling factors that represent the number of objects in the SwiM catalogdivided by the number in MaNGA main sample in each bin (see Section 2.4)
SCALING FACTOR ERR . . . 1 σ uncertainty for SCALING FACTOR s ESWEIGHT . . . Volume weights from Wake et al. (2017) for Primary+ and full secondary sample
Map HDU Data Models
In this appendix we present the data model for the SwiM VAC map files. The maps files contain the spatially-matched MaNGA IFU maps,
Swift /UVOT and SDSS photometry. The map data files have 17 total HDUs, with 5main groups: D n (4000) (HDU 0), spectral indices (HDU 1–8), emission line flux and equivalent width (HDU 9–14), Swift and SDSS photometry (HDU 15–16), and the raw
Swift information (HDU 17). The HDU format for each groupis described below, including notes on how to use the maps. The names and descriptions of the header data units(HDUs) are given in Table 4, while the formatting of the HDUs are described in Tables 5 through 9.All MaNGA maps and UVOT images have masks, where science-ready pixels are indicated by 0 and 1 otherwise.The MaNGA masks are based on those in DR15, but has been simplified to a 0 or 1 mask given the analysis presentedin this work. The masks for the UVOT images only affect two objects as discussed in Section 3.
Table 4.
SwiM Maps HDU Descriptions
Index Name Channels Units Description0
Dn4000 − cm − Hz − arcsec − Maps required to calculate the D n (4000) measurementsand its uncertainty1 SPECINDX FLUX
43 erg s − cm − arcsec − Spectral index flux maps ( F I in equation 7)2 SPECINDX CONT
43 erg s − cm − ˚A − arcsec − Spectral index continuum maps ( F C0 in equation 10)3 SPECINDX FLUX SIGMA
43 erg s − cm − arcsec − σ uncertainties for SPECINDX FLUX SPECINDX CONT SIGMA
43 erg s − cm − ˚A − arcsec − σ uncertainties for SPECINDX CONT SPECINDX MASK
43 . . . Masks for
SPECINDX FLUX , SPECINDX FLUX SIGMA , SPECINDX CONT and
SPECINDX CONT SIGMA COMBINED DISP
43 km s − Flux-weighted combined dispersion maps7
COMBINED DISP SIGMA
43 km s − σ uncertainties for COMBINED DISP COMBINED DIPS MASK
43 . . . Masks for
COMBINED DISP and
COMBINED DISP SIGMA ELINE FLUX
22 10 − erg s − cm − arcsec − Gaussian-fitted emission line flux maps based on MPL-7 DAP10
ELINE FLUX SIGMA
22 10 − erg s − cm − arcsec − σ uncertainties for ELINE FLUX ELINE FLUX MASK
22 . . . Masks for
ELINE FLUX and
ELINE FLUX SIGMA ELINE EW
22 ˚A Gaussian-fitted equivalent width maps based on MPL-7 DAP13
ELINE EW SIGMA
22 ˚A 1 σ uncertainties for the ELINE EW SIGMA ELINE EW MASK
22 . . . Masks for
ELINE EW and
ELINE EW SIGMA SWIFT/SDSS
Swift /UVOT and SDSS sky-subtracted images[uvw2,uvw1,uvm2,u,g,r,i,z]16
SWIFT/SDSS SIGMA σ uncertainties for SWIFT/SDSS SWIFT UVOT
12 . . . Swift/UVOT non-sky-subtracted counts, exposure,counts error and mask maps [uvw2, uvw1, uvm2]. HDU 0: D4000 – This HDU contains the maps necessary to calculate D n (4000) measurements and their uncer-tainties. The data are a 3–D array with the third dimension having a size of 5 corresponding to the two data channels,their uncertainties and the mask. To calculate D n (4000), use the relation D n (4000) = f ν, red /f ν, blue , and its uncertaintyas σ D n (4000) = D n (4000) (cid:115)(cid:18) σ f ν, red f ν, red (cid:19) + (cid:18) σ f ν, blue f ν, blue (cid:19) . (A1)Covariance has been properly taken into account in our data processing. The final uncertainty must be multiplied by1.4 to account for calibration errors described in Westfall et al. (2019). See Section 5.8 for more information.All maps have the units erg s − cm − Hz − arcsec − , except for the mask, which uses 0 to indicate a science-readypixels and 1 otherwise. The structure of the HDU is given in Table 5. Table 5.
HDU0: D4000 Channel Description
Channel Name Description0
Fnu Red
Flux density per unit wavelength in the red window1
Fnu Blue
Flux density per unit wavelength in the blue window2
Sigma Red
Uncertainty in flux density in the red window3
Sigma Blue
Uncertainty in flux density in the blue window4
Mask D n (4000) mask HDU 1-8: SPECINDX – These HDUs contain the information needed to calculate the spectral indices includedin this VAC (the first 43 listed in Westfall et al. 2019). We put D n (4000) spectral index in HDU 0 because of itsdifferent definition and units. The spectral indices can be calculated using Equation 7.This relation is described in more detail in Section 5.4. HDUs 1 and 3 contain the flux and the uncertaintymeasurements for the index flux F I in units of erg s − cm − arcsec − , while HDUs 2 and 4 contain the same informationfor the continuum flux density f C in units of erg s − cm − ˚A − arcsec − . HDU 5 is the mask for the spectral indexmaps, where 0 denotes science-ready pixels, and 1 denotes otherwise.Each HDU contains a 3D array with the third dimension having a size of 43 (channels) corresponding to the 43indices. The channel-to-index mapping is provided in the header and in Table 6. Here we also include the ∆ λ for eachindex which is needed to compute the final indices using Equation 7. The index bandpasses are identical to that givenin Table 4 of Westfall et al. (2019).HDUs 6–8 contain the flux weighted combined stellar and instrumental dispersion maps, its uncertainty, and mask.The data in HDUs 6 and 7 are in units of km s − , while the masks in HDU 8 have the same definitions as that ofHDU 5. These HDUs also have 43 channels corresponding to the 43 indices as given in their header and in Table 6.5 Table 6.
HDU 1–8: Spectral Index Channels Descrip-tion
Channel Name ∆ λ a (˚A)0 CN1
CN2
Ca4227
G4300
Fe4383
Ca4455
Fe4531
C24668 H β Fe5015
Mg1
Mg2
Mgb
Fe5270
Fe5335
Fe5406
Fe5709
Fe5782
NaD
TiO1
TiO2 H δ A H γ A H δ F H γ F CaHK
CaII1
CaII2
CaII3
Pa17
Pa14
Pa12
MgICvD
NaICvD
MgIIR
FeHCvD
NaI bTiO aTiO
CaH1
CaH2
NaISDSS
TiO2SDSS a The ∆ λ presented here is the width of the index band. HDU 9–14: ELINE FLUX and ELINE EW – These HDUs contain the emission line flux maps and EW maps,and their associated uncertainties. The emission line fluxes come from the Gaussian-fitted measurements from theMPL-7 DAP. Again each HDU contains a 3D array with the third dimension corresponding to different emission linechannels. The channel-to-line mapping can be found in the header and in Table 7.HDUs 9 and 10 contain the measured flux and uncertainty in units of 10 − erg s − cm − arcsec − , while HDUs 12and 13 contain the EW information in units of ˚A. HDUs 11 and 14 contain the masks for flux and EW, respectively,defined so that 0 denotes science-ready pixels and 1 denotes otherwise. Table 7.
HDU 9–14: Emission Line Channels Descrip-tion
Channel Ion λ rest (˚A) a [O II ] [O II ] H θ H η [Ne III ] H ζ [Ne III ] H (cid:15) H δ H γ [He II ] H β [O III ] [O III ] [He I ] [O I ] [O I ] [N II ] H α [N II ] [S II ] [S II ] a Vacuum rest wavelengths from the National Institute ofStandards and Technology (NIST) and are used by theMaNGA DAP. HDU 15 –16: SWIFT/SDSS – HDU15 contains the sky-subtracted NUV images from
Swift and optical im-ages from SDSS, and HDU16 contains their corresponding uncertainty images. All images are provided in units ofnanomaggies. To convert these maps to the AB magnitude ( m ) system, use m = 22 . − . ( f / nanomaggie). Toconvert to µ Jy use 1 nanomaggie = 3 . µ Jy.For 5 out of the 150 galaxies, there are bad pixels that report incorrect exposure times (usually 0 or NaN) withinthe map’s FoV. In this file, the values are stored as − inf or NaN, which will cause errors in photometric measurementsif not masked. We provide masks for all Swift images in HDU 17. We strongly recommend users always use the masksfrom HDU 17 when working with Swift images. If there are no bad pixels, then the mask will not change the image.For SDSS, masked pixels have an uncertainty of 0.In these two HDUs, the data are also given in 3D arrays with the third dimension corresponding to different filters.The correspondence are given in the header and in Table 8.
Table 8.
HDU 15–16: Photometry ChannelDescription
Channel Name Central Wavelength (˚A)0 uvw2 uvw1 uvm2
SDSS u
SDSS g
SDSS r
SDSS i
SDSS z HDU 17: SWIFT UVOT – This HDU contains the
Swift /UVOT counts, uncertainty, exposure and mask mapsfor all three NUV filters. The masks have a value of 0 for science-ready pixels and 1 otherwise. Unlike HDU 15 and16, these images are not sky-subtracted. However the calculated sky counts are provided in the header under thekeywords
SKY W1 , SKY M2 and
SKY W2 . The AB magnitude system zero points of the filters and f λ conversion factorsare also provided in the header as ABZP ∗ and FLAMBDA ∗ , where the ∗ represents the desired filter. The structure ofthis HDU is given in Table 9. Table 9.
HDU 17: Swift/UVOT Channel Description
Channel Name Description0 uvw2 Counts
Fully reduced, non-sky subtracted uvw2 counts1 uvw1 Counts
Fully reduced, non-sky subtracted uvw1 counts2 uvm2 Counts
Fully reduced, non-sky subtracted uvm2 counts3 uvw2 Counts Err
Uncertainty associated with uvw2 counts4 uvw1 Counts Err
Uncertainty associated with uvw1 counts5 uvm2 Counts Err
Uncertainty associated with uvm2 counts6 uvw2 Exposure
Exposure map for uvw2 image7 uvw1 Exposure
Exposure map for uvw1 image8 uvm2 Exposure
Exposure map for uvm2 image9 uvw2 Mask
Mask for uvw2 image10 uvw1 Mask
Mask for uvw1 image11 uvm2 Mask
Mask for uvm2 image
Aguado, D. S., Ahumada, R., Almeida, A., et al. 2019,ApJS, 240, 23Astropy Collaboration, Price-Whelan, A. M., Sip˝ocz, B. M.,et al. 2018, AJ, 156, 123Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981,PASP, 93, 5Battisti, A. J., Calzetti, D., & Chary, R.-R. 2016, ApJ, 818,13Battisti, A. J., Calzetti, D., & Chary, R. R. 2017, ApJ, 851,90Beers, T. C., Flynn, K., & Gebhardt, K. 1990, AJ, 100, 32Belfiore, F., Westfall, K. B., Schaefer, A., et al. 2019, AJ,158, 160Bentz, M. C., & Katz, S. 2015, PASP, 127, 67Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393Blanton, M. R., Kazin, E., Muna, D., Weaver, B. A., &Price-Whelan, A. 2011, AJ, 142, 31Blanton, M. R., Bershady, M. A., Abolfathi, B., et al. 2017,AJ, 154, 28Bolton, A. S., Schlegel, D. J., Aubourg, ´E., et al. 2012, AJ,144, 144Breeveld, A. A., Curran, P. A., Hoversten, E. A., et al.2010, MNRAS, 406, 1687Bundy, K., Bershady, M. A., Law, D. R., et al. 2015, ApJ,798, 7Burrows, D. N., Hill, J. E., Nousek, J. A., et al. 2000, inProc. SPIE, Vol. 4140, X-Ray and Gamma-RayInstrumentation for Astronomy XI, ed. K. A. Flanagan &O. H. Siegmund, 64–75Burstein, D., Faber, S. M., Gaskell, C. M., & Krumm, N.1984, ApJ, 287, 586Calzetti, D. 2013, Star Formation Rate Indicators, ed.J. Falc´on-Barroso & J. H. Knapen, 419Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ,533, 682Cappellari, M., & Copin, Y. 2003, MNRAS, 342, 345Cappellari, M., & Emsellem, E. 2004, PASP, 116, 138Charlot, S., & Fall, S. M. 2000, ApJ, 539, 718Drory, N., MacDonald, N., Bershady, M. A., et al. 2015,AJ, 149, 77Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2007,A&A, 469, 379—. 2009, MNRAS, 397, 1177Evans, P. A., Osborne, J. P., Beardmore, A. P., et al. 2014,The Astrophysical Journal Supplement Series, 210, 8Fabbiano, G. 2006, Annual Review of Astronomy andAstrophysics, 44, 323Faber, S. M., Friel, E. D., Burstein, D., & Gaskell, C. M.1985, ApJS, 57, 711 Garay, G., & Lizano, S. 1999, PASP, 111, 1049Greene, J. E., & Ho, L. C. 2007, ApJ, 670, 92Gunn, J. E., Siegmund, W. A., Mannery, E. J., et al. 2006,AJ, 131, 2332Hunt, L. K., & Hirashita, H. 2009, A&A, 507, 1327Husemann, B., Jahnke, K., S´anchez, S. F., et al. 2013,A&A, 549, A87Kauffmann, G., Heckman, T. M., Tremonti, C., et al. 2003,MNRAS, 346, 1055Kennicutt, R. C., & Evans, N. J. 2012, ARA&A, 50, 531Kennicutt, Jr., R. C. 1984, ApJ, 287, 116—. 1998, ARA&A, 36, 189Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T.2006, MNRAS, 372, 961Kim, K.-T., & Koo, B.-C. 2001, ApJ, 549, 979Law, D. R., Cherinka, B., Yan, R., et al. 2016, AJ, 152, 83Lehmer, B. D., Alexander, D. M., Bauer, F. E., et al. 2010,ApJ, 724, 559Martig, M., Bournaud, F., Teyssier, R., & Dekel, A. 2009,ApJ, 707, 250Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005,ApJL, 619, L1Molina, M., Ajgaonkar, N., Yan, R., et al. 2020, MNRAS,494, 4751O’Donnell, J. E. 1994, ApJ, 422, 158Osterbrock, D. E., & Ferland, G. J. 2006, Astrophysics ofgaseous nebulae and active galactic nucleiPanessa, F., Bassani, L., Cappi, M., et al. 2006, A&A, 455,173Peacock, J. A. 1983, MNRAS, 202, 615Poole, T. S., Breeveld, A. A., Page, M. J., et al. 2008,MNRAS, 383, 627Roming, P. W. A., Kennedy, T. E., Mason, K. O., et al.2005, SSRv, 120, 95Salim, S., Boquien, M., & Lee, J. C. 2018, ApJ, 859, 11S´anchez-Bl´azquez, P., Peletier, R. F., Jim´enez-Vicente, J.,et al. 2006, MNRAS, 371, 703Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ,500, 525Smee, S. A., Gunn, J. E., Uomoto, A., et al. 2013, AJ, 146,32Springel, V., Di Matteo, T., & Hernquist, L. 2005,MNRAS, 361, 776Steinhauser, D., Schindler, S., & Springel, V. 2016, A&A,591, A51Trager, S. C., Worthey, G., Faber, S. M., Burstein, D., &Gonzalez, J. J. 1998, The Astrophysical JournalSupplement Series, 116, 10