The DustPedia Archive is now available here!

The Galaxy Sample

Our intention was to select a sample of nearby galaxies, that included a sub-set that were resolved by Herschel (D\(_{25}\) \(> 5\) arc min), but also sampled large scale structure and hence different galactic environments. We also wanted to primarily select objects according to their stellar mass, the best proxy for which is near infrared flux density. To satisfy these criteria the initial definition of the DustPedia sample was that it should consist of all galaxies observed by Herschel that lie at recessional velocities of \(< 3000\,\mathrm{km\,s}^{-1}\), with optical diameters \(>1\) arc min and a WISE 3.4\(\mu\)m signal-to-noise ratio (SNR) of \(>3\).


Fig. 1. - The spatial distribution of galaxies in Galactic coordinates but plotted so that for \(l>180^{\circ}\) the plotted value is \(l-360^{\circ}\). Blue: galaxies selected from WISE. Red: WISE galaxies with a Herschel detection. The Virgo and Fornax clusters are labeled along with Ursa Major where there is also a concentration of nearby galaxies.

The velocity restriction means that we include galaxies that are 'local' yet still reside in different environments. For example this selection includes the Virgo and Fornax clusters and galaxies in the super-galactic plane that consists of the Virgo southern extension and connecting galaxies from Virgo to the less rich Ursa Major cluster, see Fig. 1. The size constraint ensures resolved observations in some, but not all, Herschel bands and the WISE flux provides for our required stellar mass selection.

Our initial selection was from the LEDA database. We selected 4259 galaxies that had v\(_{\mathrm{rad}}\) \(< 3000\,\mathrm{km\,s}^{-1}\) and D\(_{25}\) \(> 1\) arc min, where v\(_{\mathrm{rad}}\) is the measured radial velocity either from an optical or 21cm spectrum and D\(_{25}\) is the major axis diameter measured at the 25th blue magnitude per sq arc sec isophote.

Selection CriterionNumber
LEDA + WISE\(_{5\sigma}\)4231
LEDA + WISE\(_{5\sigma}\) + PACS829
LEDA + WISE\(_{5\sigma}\) + SPIRE907
LEDA + WISE\(_{5\sigma}\) + PACS + SPIRE798
LEDA + WISE\(_{5\sigma}\) + PACS/SPIRE867

Table 1. Summary of the numbers of identified galaxies when applying the various DustPedia sample selection criteria.

Subsequently, small alterations to the WISE sample selection criteria and methods were found to be necessary. The WISE all-sky data release source catalogue actually only contains sources with SNR \(>5\) in at least one band. Given that the 3.4\(\mu\)m band is the most sensitive WISE band, and is also the band in which galaxies tend to be most luminous, it is almost always the band with the greatest SNR. Therefore no or very little information is available for sources with SNR \(<5\). Moreover, the observational depth achieved by WISE varies across the sky (ranging from 1:100 scans) as a result of observing strategy. The WISE All-Sky Release Explanatory Supplement1 states that at 3.4\(\mu\)m WISE has a \(95\%\) sky coverage at a \(5\sigma\) sensitivity limit of \(0.039\,\)mJy (\(19.91\) magAB, \(17.23\) magVega). It is this \(5\sigma\) selection criterion that we eventually used to select DustPedia galaxies. The WISE data is contained within the WISE ALL-Sky Data Release Catalogue (Cutri et al., 2012). When matched to the LEDA data the WISE selection criterion only very slightly reduced the sample size from 4259 to 4231.

It was also decided that studying the very large angular size Local Group galaxies M31 (and its companions), M33, LMC and SMC falls outside the scope of the project, given the very different nature of working with such extended systems. For that reason these galaxies were also excluded from our sample. We also excluded five galaxies that lie behind the LMC, which would make them particularly difficult to analyse.

Finally, we have interrogated the Herschel Science Archive (HSA) to look for photometer (not spectra) observations of our LEDA and WISE selected targets. The two relevant Herschel instruments are PACS (observations at 70, 100, 160\(\mu\)m) and SPIRE (observations at 250, 350, 500(\mu\)m).

The HSA does not yet allow users to ascertain whether or not an arbitrary position on the sky was observed. Therefore, to determine whether a particular object was observed required a rather brute force approach. Basic pipeline reductions of every Herschel photometer observation were downloaded in turn from the HSA and the coverage area of each observation was compared to the positions of the LEDA + WISE selected galaxies. Of the 4231 objects selected from LEDA and WISE, 907 have been observed by SPIRE, whilst 829 have been observed by PACS. The vast majority, 798, have been observed by both instruments; 109 only have SPIRE data, and 31 have PACS data only. A total of 938 have been observed by either PACS and/or SPIRE and this constitutes our primary DustPedia sample. Table 1 summarises the numbers of sources found when applying the various sample selection criteria.

In Fig. 2 we compare the physical properties of the galaxies selected using LEDA and WISE (blue) with those that additionally have a detection by Herschel (red). In the main the Herschel detected galaxies appear to broadly follow the distributions of the primary (LEDA and WISE) selected galaxies. However, using a Kolmogorov-Smirnov two sided test we can reject the hypothesis that the WISE(LEDA) and Herschel data are drawn from the same parent distributions. The only possible exception to this is the distribution of galaxy sizes, but even here the probability that they are drawn from the same distribution is only 0.53. This entirely foreseeable problem of course arises, because of the rather arbitrary way in which the various uncoordinated Herschel projects we have obtained data from, have sampled the local galactic population. To overcome this problem our intention is to use correlations obtained from the Herschel data to predict the properties of those galaxies not observed by Herschel. In this way we hope to apply much of our analysis to the entire 4231 WISE(LEDA) selected galaxies.

We are also able to obtain multi-wavelength data for the majority of the primary sample galaxies; GALEX (DR5, Bianchi et al., 2011) coverage is available for 697 (74\(\%\)), SDSS (DR9, Ahn et al., 2012) coverage is available for 651 (69\(\%\)), and UKIDSS (DR9, Lawrence et al., 2013) coverage is available for 419 (41\(\%\)), along with the all-sky coverage provided by DSS in the optical, 2MASS (Skrutskie et al., 2006) in the NIR, and WISE (Wright et al., 2010) in the MIR. At 22\(\mu\)m WISE has a 95\(\%\) coverage at a \(5\sigma\) sensitivity level of 5.1 mJy (8 magAB, 14.6 magVega). The majority of the sample galaxies have good 22\(\mu\)m detections; 637 (67\(\%\)) are brighter than the \(5\sigma\) level at 22\(\mu\)m, with an additional 106 galaxies detected at \(>\,3\sigma\) (total fraction 79\(\%\)). When considering only late-type galaxies, the 5 and \(3\,\sigma\) detection fractions at 22\(\mu\)m increase to 74\(\%\) and 83\(\%\) respectively.

apparent mag

a) WISE 3.4\(\mu\)m apparent magnitude.

ang diameter

b) Optical angular diameter.

measured radial velocity

c) Measured radial velocity.


d) Morphological type.

Fig. 2. - The above figures compare the properties of the galaxies selected from WISE (blue) with those that also have a Hershel detection (red).

Correlating the positions of the DustPedia sources with the Planck CCP (Planck Collaboration et al., 2014) using a matching radius of 5 arc min (the resolution of Planck at 550 and 350\(\mu\)m) returns matches for 461 (49\(\%\)) of the sample. However, the shorter Planck wavelengths have more detections; there at 450 matches with the Planck 350\(\mu\)m catalogue, whilst there are only 197 with the Planck 850\(\mu\)m catalogue.

DustPedia data is now available VIEW