EDP Sciences
Free Access
Issue
A&A
Volume 549, January 2013
Article Number A31
Number of page(s) 13
Section Catalogs and data
DOI https://doi.org/10.1051/0004-6361/201219435
Published online 11 December 2012

© ESO, 2012

1. Introduction

During the past year, the simultaneous operation of ESA’s Herschel (Pilbratt et al. 2010) and Planck (Tauber et al. 2010; Planck Collaboration 2011a) missions has given us an unprecedented opportunity to cover one of the last few observational gaps in the far-infrared and sub-millimeter regions of the electromagnetic spectrum. Herschel is an observatory facility that covers the 55−671 μm spectral range, with angular resolution ranging between 6 and 35 arcsec (Pilbratt et al. 2010). Planck is a surveyor that is observing the whole sky in nine spectral bands between 350 μm and 1 cm with angular resolution ranging from 4.23 to 32.65 arcmin. Planck has two frequency channels close to the Herschel bands: the 545 and 857 GHz (550 and 350   μm) channels of the High Frequency Instrument (HFI). In this paper we will study the cross-correlation of the Planck Early Release Compact Source Catalogue (ERCSC) with the phase 1 of the catalogue of the Herschel Astrophysical Terahertz Large Area Survey (H-ATLAS, Eales et al. 2010).

The overlap in time and frequency between Herschel and Planck is not accidental: the two missions have been planned keeping in mind the added scientific value of a synergy between them (Planck Collaboration 2006). In addition to providing a broader spectral coverage of common sources, the combination of Planck and Herschel data will be beneficial in other respects. In particular, the much higher resolution and sensitivity of Herschel makes it well-suited for follow-up studies of Planck sources1, allowing us to assess the effects of source confusion in Planck channels. In some cases it will be possible to resolve individual detections by Planck into separated sources. More generally, Herschel will make it possible to quantify the boosting of Planck fluxes by the many faint sources within its beam. Moreover, Herschel data can be used to improve the foreground characterization, thus helping us to distinguish between genuine, possibly extragalactic, point sources and compact Galactic cirrus, and to provide more precise positions, essential for source identification at other wavelengths. This knowledge will be important for the interpretation of the all-sky Planck survey data. Note however that although Herschel’s resolution is much better than Planck’s, it is still highly likely that many of the Herschel 350 and 500 μm sources are also blends. H-ATLAS maps are in general not affected by source confusion, except in regions with considerable cirrus, but Herschel sources resolve to multiple MIPS (Multiband Imaging Photometer for Spitzer, Rieke et al. 2004) sources in many cases.

Also, a comparison between catalogues of galaxies observed with Herschel and Planck can be used as a check on the calibration of the two observatories, which is done in different ways. The absolute calibration of Planck fluxes should be better than 2% up to 353 GHz, where it is based on the CMB dipole, and  ≃ 7% in the two highest frequency channels (545 and 857 GHz), where it is based on a comparison with COBE/FIRAS (Planck HFI Core Team 2011; Zacchei et al. 2011)2. For comparison, the overall photometric accuracy of the Herschel/SPIRE instrument is conservatively estimated as  ≃7% (SPIRE Observers’ Manual v2.2 2010)3.

thumbnail Fig. 1

Positions of the ERCSC 857 GHz sources in the 350 μm H-ATLAS GAMA-09 field. The source at coordinates 0857+3 has not been included in this analysis because it lies on the poorly sampled edge of the field.

Open with DEXTER

thumbnail Fig. 2

Positions of the ERCSC 857 GHz sources in the 350 μm H-ATLAS GAMA-12 field.

Open with DEXTER

thumbnail Fig. 3

Positions of the ERCSC 857 GHz sources in the 350 μm H-ATLAS GAMA-15 field.

Open with DEXTER

thumbnail Fig. 4

Position of the ERCSC sources (black ovals) superimposed on the IRIS 100 micron map around the GAMA-09 H-ATLAS field.

Open with DEXTER

Owing to their different angular resolution and sensitivity, Planck and Herschel have almost complementary selection functions for extragalactic sources: Planck primarily detects nearby galaxies (although it might also detect extreme high-redshift objects), whereas most H-ATLAS galaxies lie around z ~ 1 (Eales et al. 2010). Moreover, Planck covers a broader spectral range (nine frequencies from 30 to 857 GHz), which is useful to improve the characterization and removal of foregrounds and to follow the spectral energy distributions (SEDs) of interesting objects, such as blazars, down to radio frequencies. With its all-sky coverage, Planck is ideal for detecting the rarest, most extreme (sub-)mm sources. In particular, it may detect the most luminous protoclusters of dusty galaxies, whose luminosities, integrated over the Planck beam, may be, at z ≳ 1, more than an order of magnitude higher than the mean luminosity of individual dusty galaxies at the same redshift (Negrello et al. 2005). The far superior Herschel resolution and point source detection capabilities will then allow us to establish the nature of candidate high-z protoclusters and to characterize the physical properties of those that are confirmed.

In this paper we present the results of a cross-correlation of the Planck ERCSC catalogue with the catalogue of H-ATLAS sources detected in the phase 1 fields (Dunne et al., in prep.). The common sources are described in Sect. 2. In Sect. 3 we compare the flux density estimates of both experiments. In Sect. 4 we study the contamination of the ERCSC subsample by looking for extended diffuse emission as a tracer of high-latitude cirrus. In Sect. 5 we discuss a very unusual source that may be a combination of a (maybe random) condensation of low redshift low-flux galaxies and a strongly lensed galaxy. Finally, in Sect. 6 we summarize our conclusions.

2. Planck sources in the H-ATLAS phase 1 fields

thumbnail Fig. 5

SPIRE images at 350 μm showing postage stamps of the Planck ERCSC sources at 857 GHz in the H-ATLAS GAMA-09 field. The images are 400 arcsec wide and the red crosses indicate the ERCSC position for each source. The full ERCSC name of the objects has been abbreviated in the captions for conciseness.

Open with DEXTER

thumbnail Fig. 6

SPIRE images at 350 μm showing postage stamps of the Planck ERCSC sources at 857 GHz in the H-ATLAS GAMA-12 field. The images are 400 arcsec wide and the red crosses indicate the ERCSC position for each source.

Open with DEXTER

thumbnail Fig. 7

SPIRE images at 350 μm showing postage stamps of the Planck ERCSC sources at 857 GHz in the H-ATLAS GAMA-15 field. The images are 400 arcsec wide and the red crosses indicate the ERCSC position for each source.

Open with DEXTER

The Planck ERCSC (Planck Collaboration 2011b) provides an all-sky list of compact Galactic and extragalactic objects including stars with dust shells, prestellar cores, radio galaxies, blazars, infrared luminous galaxies, Galactic interstellar medium features, cold molecular cloud core candidates, and galaxy cluster candidates. The list contains more than 15 000 distinct objects,  ~60% of which are visible in the Planck highest frequency (545 and 857 GHz) channels that virtually overlap with the Herschel/SPIRE bands. A sufficiently wide area survey made with Herschel is bound to contain at least some of these sources.

The H-ATLAS is the largest area survey carried out by the Herschel Space Observatory (Pilbratt et al. 2010). It will cover  ~550   deg2 with PACS (Poglitsch et al. 2010) and SPIRE (Griffin et al. 2010) in five bands, from 100 to 500   μm. The phase 1 observations have surveyed a sufficiently wide area (134.55   deg2 at the moment of submitting this paper) to allow a preliminary, yet meaningful comparison with Planck ERCSC data.

The two highest frequency channels of the Planck/HFI practically overlap with the two lower frequency bands of Herschel/SPIRE. The 350 μm band almost coincides in central wavelength and bandwidth with the Planck/HFI 857 GHz channel. The 500 μm band and the Planck/HFI 545 GHz channel do not coincide exactly, but are close enough to consider a cross-check. The H-ATLAS phase 1 covers 134.55   deg2 divided into three regions (GAMA-09, GAMA-12, and GAMA-15).

2.1. H-ATLAS 350 μm counterparts of ERCSC 857 GHz sources

Within the H-ATLAS phase 1 fields there are 28 ERCSC sources detected by Planck at 857 GHz. The positions of the 28 sources in the H-ATLAS GAMA-09, GAMA-12, and GAMA-15 fields are shown in Figs. 13 respectively. Among these, there are no clusters of galaxies (detected through the Sunyaev-Zel’dovich effect) or cold cores. Figures 57 show postage stamp images of these 28 sources as viewed by Herschel at 350 μm in the GAMA-09, GAMA-12, and GAMA-15 fields, respectively. By inspecting the SPIRE images around the positions of the sources we find several different situations.

As many as 16 out of the 28 ERCSC objects do not have a consistent H-ATLAS counterpart, because H-ATLAS sources within the Planck beam are too faint to explain the flux density measured by Planck. Almost all (15) of them are flagged as extended or have a relatively high ( ≥ 0.125) cirrus flag in the ERCSC (11 have both properties). All but G266.26+58.99 (object #17 in Table 1 and (b) in Fig. 6 which, incidentally, is not flagged as extended and has a relatively low cirrus flag) are in the GAMA-09 field, which is more contaminated by Galactic emission than the other two GAMA fields. Figure 4 shows the positions of the ERCSC sources around the GAMA-09 field superimposed on the IRIS 100 micron map (Miville-Deschênes & Lagache 2005). The high correlation between ERCSC sources and the IRIS map reinforces the idea that the GAMA-09 ERCSC sources are likely to be mostly related to Galactic cirrus.

Of the other 12 sources, one (G263.84+57.55, #16 in Table 1 and object (a) in Fig. 6) is clearly resolved by Herschel into two relatively bright sources, the galaxy pair KPG289 (Karachentsev et al. 1976) formed by the galaxies NGC 3719 and NGC 3720, each with flux density  >250 mJy at 350 μm, while another object (G270.59+58.52, object #18 in Table 1, which is shown in panel (c) of Fig. 6) is resolved into an unusual condensation of low flux, probably high-redshift point sources. We will discuss this source in more detail in Sect. 5. Finally, 10 ERCSC sources can be clearly identified with single bright H-ATLAS sources at low redshift, including the edge-on spiral NGC 5746 (object (h) in Fig. 7) the spirals NGC 5690, NGC 5705 and NGC 4030, the peculiar asymmetric galaxy NGC 5713 (Dale et al. 2012), in addition to the above mentioned pair NGC 3719 and NGC 3720.

Table 1 lists the 28 sources with their 857 GHz flux densities taken from the ERCSC (Planck Collaboration 2011). The table gives the ERCSC source name, the RA and Dec coordinates, the flux densities and associated errors, and the EXTENDED and CIRRUS flags. Column 9 gives the 350   μm flux densities of the brightest H-ATLAS sources found inside the Planck beam4, provided that they have S350 ≥ 250 mJy. For these sources we give, in Col. 11, the spectroscopic or photometric redshifts taken from the H-ATLAS catalogue (Dunne et al., in prep.).

Table 1

ERCSC sources at 857 GHz and their H-ATLAS counterparts (inside a 4.23 arcmin radius circle around the ERCSC position) at 350 μm.

2.2. H-ATLAS 500 μm counterparts of ERCSC 545 GHz sources

There are 14 ERCSC sources detected at 545 GHz that lie in the H-ATLAS phase 1 fields. Among them, 13 are also in the sample of 28 sources detected by the ERCSC at 857 GHz that we described in Sect. 2.1. The remaining object (PLCKERC545 G230.17+32.05, with ID #29 in Table 2) is in the GAMA-09 region and has a high CIRRUS flag. Like at 857 GHz, more than half of the sources (8 out of 14, 7 of them being in the GAMA-09 field) do not have a plausible H-ATLAS counterpart. Table 2 lists the 14 sources, giving the H-ATLAS ID and redshifts for the six sources that have a 500 μm counterpart with S500 ≳ 250 mJy.

Table 2

ERCSC sources at 545 GHz and their H-ATLAS 500 μm counterparts inside a 4.47 arcmin radius circle around the ERCSC position.

3. Photometry

Comparing the flux density estimations of the ERCSC and the SPIRE phase 1 catalogues is not straightforward. To correctly compare ERCSC and SPIRE photometric estimations it is necessary to take into account that

  • Herschel has better angular resolution than Planck.It is possible that an ERCSC sourcecan be resolved into several sources by Herschel.

  • The wavelengths of the Planck bands do not correspond exactly to the wavelenghts of their Herschel counterparts.

  • Both catalogues have been obtained by using different detection and photometry extraction algorithms.

The effect of the different angular resolutions can be corrected, at least to first order, by integrating the SPIRE flux densities over the larger Planck beam area and weighting by the Planck beam response. The effect of the different central wavelenghts can be taken into account by means of SED colour correction, as will be described in Sect. 3.2. A review of the technical details of the ERCSC and SPIRE flux density estimates used in the catalogues is beyond the scope of this work. ERCSC photometry is described in Planck Collaboration (2011b) and Aatrokoski et al. (2011); SPIRE photometry is described in Rigby et al. (2011); both catalogues have passed strict internal and external validation. For pointlike sources, we assume that the flux density estimates that are given in both catalogues are accurate, within the calibration uncertainties of their experiments. However, it must be noted that the different ways in which the photometry of extended sources is obtained in the ERCSC and in the H-ATLAS phase 1 catalogues can affect the comparison. The H-ATLAS catalogues use aperture photometry scaled to the optical size of the sources (Rigby et al. 2011). The Planck ERCSC contains several types of photometric measurements for each source; the default aperture photometry information is listed in the ERCSC “FLUX” column, but other photometric measurements may be more appropriate in certain circumstances. As described in the Planck Explanatory Supplement to the ERCSC (Aatrokoski et al. 2011), for extended sources it may be better to use the Gaussian-fit photometry listed in the ERCSC “GAUFLUX” column instead of a fixed aperture photometry. Since a significant fraction of the sources in our sample at 857 GHz are flagged as extended in the ERCSC, we have compared the H-ATLAS flux densities to both the fixed aperture photometry (“FLUX”) and the Gaussian-fit (“GAUFLUX”) ERCSC flux densities. We find that at the lowest fluxes FLUX and GAUFLUX work similarly well, while at bright fluxes GAUFLUX is more consistent with Herschel photometry (maybe because many bright ERCSC sources are extended). Therefore, for the remainder of this paper, we will use GAUFLUX when referring to Planck photometry.

3.1. ERCSC-857 GHz and H-ATLAS 350 μm flux densities

According to the technical specifications of their respective instruments, the Planck/HFI 857 GHz and the Herschel/SPIRE 350 μm channels have almost exactly the same central wavelength and roughly the same bandwidth. This makes it easy to directly compare the flux density estimations of the ERCSC sources present in the H-ATLAS fields. To take into account the different angular resolutions of Herschel and Planck, we calculated an effective Herschel 350 μm flux density by summing the flux densities, corrected for the effect of the Planck beam, of the sources listed in the H-ATLAS phase 1 catalogue around the ERCSC positions. We assumed a circular, Gaussian beam of FWHM = 4.23 arcmin (Planck Collaboration 2011b).

Figure 8 shows the Planck 857 GHz flux densities compared to Herschel 350   μm flux densities for those ERCSC sources that have at least one SH − ATLAS ≥ 250 mJy. All these sources lie in the GAMA-12 or GAMA-15 fields. According to Aatrokoski et al. (2011), the ERCSC flux densities below  ≃1.3 Jy are boosted by the well-known selection bias (sources sitting on top of positive noise plus confusion plus Galactic emission fluctuations that dominate the contribution to the measured flux densities are more likely to be detected). Excluding the large edge-on spiral NGC 5746, which will be discussed below, for sources with H-ATLAS 350   μm flux densities  ≥1.5 Jy we see a good agreement between Herschel and Planck flux densities. We find  ⟨ SH − ATLAS − SERCSC ⟩  ≃ 0.1 Jy. The relative flux density difference for the same sources, defined as (1)is  | ϵ |  = 3%, smaller than the calibration uncertainty of both Herschel and Planck. However, the relative difference of the individual sources shows a large ( ~ 20%) scatter, probably due to the small size of the sample. The mean difference between the ERCSC and H-ATLAS positions for sources brighter than 1.5 Jy is 0.42 arcmin, with a dispersion of 0.17 arcmin. For the assumed beam shape, this may account for a  ≃ 5% underestimate of ERCSC flux densities.

thumbnail Fig. 8

ERCSC 857 GHz flux densities compared with the 350 μm flux densities of the brightest H-ATLAS sources inside the Planck beam (squares). The dashed line indicates the SATLAS = SERCSC identity. The filled circles show the summed flux densities, weighted with a Gaussian beam centred on the ERCSC position and with FWHM = 4.23 arcmin (Planck Collaboration 2011b). Only the ERCSC sources with at least one Herschel counterpart with flux density SH − ATLAS > 0.25 Jy are shown in this plot (see text for further details). The outlier with ERCSC flux density  ~10 Jy is the edge-on spiral galaxy NGC 5746 (object #28 in Table 1, see discussion in the main text). The outlier with ERCSC flux density  ~2 Jy is the G12H29 (source #18 in Table 1), to be discussed in Sect. 5.

Open with DEXTER

The two remarkable outliers in Fig. 8 are sources #18 and #28 in Table 1. Source #18, G270.59+58.52 (object (c) in Fig. 6), flagged as extended in the ERCSC and with SH − ATLAS = 352 mJy, will be discussed in Sect. 5. Source #28 (object (h) in Fig. 7) is identified as NGC 5746, a large edge-on spiral that is clearly resolved as a very extended source by SPIRE but is not flagged as extended by the ERCSC. We believe that the discrepancy between the flux densities for this object reported by H-ATLAS and the ERCSC is due to the very different angular resolution of Herschel and Planck and to the different way in which background subtraction has been performed by the catalogue-making pipelines of the two experiments. In particular, if aperture photometry is applied to the raw (not background-subtracted) SPIRE 350 micron map, a flux density of  ~10 Jy is obtained for this object, which is more consistent with the 857 GHz value. A revised version of the phase I H-ATLAS catalogue, which deals more carefully with very extended sources, solves this discrepancy (L. Dunne, priv. comm.).

3.2. ERCSC-545 GHz and H-ATLAS 500 μm flux densities

The comparison between the Planck 545 GHz and the SPIRE 500 μm flux densities is less straightforward. The central frequency of the 500 μm channel is  ~600 GHz, significantly higher than that of the nearest Planck channel (545 GHz). A colour correction is thus necessary. From the mean SED of IRAS PSCz galaxies determined by Serjeant & Harrison (2005), we find  ⟨ S600 / S545 ⟩  ≃ 1.35. For the comparison with Planck we scaled down the H-ATLAS flux densities by this factor, except for the rich clump of low-flux galaxies, for which we adopted a correction factor of 1.1 for the strongly lensed galaxy at z = 3.259 and of 1.3 for the surrounding H-ATLAS sources, assumed to be at z ≤ 1. These factors were calculated using the SED of H-ATLAS J142413.9+022304, a strongly lensed sub-mm galaxy at z ≈ 4.24 (Cox et al. 2011), for the strongly lensed galaxy and the SED of SMM J2135-0102 (“The Cosmic Eyelash”; Ivison et al. 2010; Swinbank et al. 2010), that Lapi et al. (2011) found to work well for many high-z H-ATLAS galaxies. We denote the colour-corrected flux by the symbol S.

Figure 9 compares the Planck 545 GHz flux density with that of the brightest source inside the Planck beam (filled squares) and with that obtained summing the flux densities of H-ATLAS sources within the Planck beam, corrected for the effect of the beam response (FWHM = 4.47 arcmin) function and the colour correction (filled circles). Only the Planck ERCSC 545 GHz sources that have at least one H-ATLAS counterpart with (not colour-corrected) SH − ATLAS ≳ 250 mJy are shown in the plot. Again NGC 5746 and the clump around G270.59+58.52 stand out for their high ERCSC-to-H-ATLAS flux density ratios. Excluding these, we find, after the colour correction,  Jy, with a dispersion of  ≃ 0.25 Jy for sources with  Jy. The corresponding relative difference is  ⟨ ϵ ⟩  =  −7.5% with dispersion σϵ = 13.5%. This result, however, was obtained from a very small sample of five sources and cannot be considered to be statistically meaningful.

thumbnail Fig. 9

ERCSC flux densities at 545 GHz compared with the colour-corrected 500 μm flux densities of the brightest H-ATLAS sources inside the Planck beam (squares). The dashed line indicates the SATLAS = SERCSC identity. The filled circles show the colour-corrected summed flux densities, weighted with a Gaussian beam centred on the ERCSC position and with FWHM = 4.47 arcmin (Planck Collaboration 2011b). Only the ERCSC sources with at least one Herschel counterpart with (not colour-corrected) flux density SH − ATLAS > 0.25 Jy are shown in this plot. The outlier with ERCSC flux density  ~ 3.5 Jy is the edge-on spiral galaxy NGC 5746. The outlier with ERCSC flux density  ~ 1.5 Jy is the G12H29 source to be discussed in Sect. 5.

Open with DEXTER

4. Contamination by Galactic cirrus

The GAMA-09 field is more contaminated by Galactic emission than the other two GAMA fields (Bracco et al. 2011). None of the 15 ERCSC 857 GHz sources, in this field with SERCSC in the range 1.2 − 3.5 Jy, have a plausible Herschel counterpart. The summed flux densities of the faint Herschel sources within the Planck beam fall well short of accounting for the ERCSC flux densities. A visual inspection of Fig. 5 clearly reinforces the idea that the ERCSC sources in this region of the sky are not associated with bright Herschel galaxies. All but two of the ERCSC sources in the GAMA-09 have a cirrus flag  ≥ 0.125 and all but two (different from the previous two objects) are labelled as extended in the ERCSC. It is thus likely that most of the flux density within the Planck beam comes from Galactic cirrus. The situation is much better in the GAMA-12 and GAMA-15 fields. In the former, only one (out of five) 857 GHz ERCSC source does not have an H-ATLAS counterpart with SH − ATLAS > 250 mJy. Somewhat surprisingly, this source is not labelled as extended in the ERCSC and has a relatively low cirrus flag (0.0625). One of the two ERCSC 545 GHz sources in the same field behaves in a similar way: no H-ATLAS counterpart with SH − ATLAS > 250 mJy, not labelled as extended, low cirrus flag (0.03125). Fortunately, all the eight 857 GHz and the five 545 GHz ERCSC sources in the GAMA-15 field do have a consistent H-ATLAS counterpart.

Although the statistics are poor, these findings may indicate that all ERCSC sources with a cirrus flag  ≥ 0.125 are cirrus dominated, even if they are not labelled as extended, as is the case for two 857 GHz and three 545 GHz GAMA-09 sources. Of the three 857 GHz sources labelled as extended but with a cirrus flag  < 0.125, two are probably cirrus dominated, while the third is the composite high-z lensed galaxy plus low-z clump (see Sect. 5). All 545 GHz sources labelled as extended are probably cirrus dominated, even if the cirrus flag is  <0.125.

We can then tentatively conclude that a cirrus flag  ≥ 0.125 or the “extended” label are good indicators of cirrus dominance, although their presence (or absence) does not always guarantee that a source is (or is not) dominated by cirrus.

thumbnail Fig. 10

Clump around G12H29 as seen by Herschel at 250 (left), 350 (centre) and 500 μm (right). The field of view is 600 arcsec wide and is centred at the position of the 857 GHz ERCSC source PLCKERC857 G270.59+58.52.

Open with DEXTER

thumbnail Fig. 11

Herschel three-colour image of the region around HATLAS J114637.9-001132 (alias G12H29). Colours are: blue – 250 μm; green – 350 μm; red – 500 μm. The colour scale was chosen so that objects that are red/white are likely to lie at z ~ 3. The brightest such object in this image is the z = 3.259 likely lensed galaxy HATLAS J114637.9-001132. The circle is centred on the Planck source and is the size of the Planck 857 GHz beam. (This figure is available in color in the electronic form.)

Open with DEXTER

5. The strongly lensed Herschel/Planck source H-ATLAS J114637.9-001132 at z = 3.26

The ERCSC object # 18, PLCKERC857 G270.59+58.52, shown as observed by Herschel in panel (c) of Fig. 6 and in Fig. 10, has some unusual characteristics that make it interesting and worthy of further study. Eleven out of the 12 matches between ERCSC sources found at 857 GHz and H-ATLAS sources with SH − ATLAS > 250 mJy are associated with nearby quiescent galaxies detected in the optical. One source, however, has no bright optical counterpart. Instead, it is associated with a clump of sources with low Herschel flux densities (as can be better appreciated in Fig. 11) grouped around a bright Herschel source, HATLAS J114637.9-001132, alias G12H29, whose flux peaks at 350 μm. Its Planck colour is also unusual, as can be seen in Fig. 12 where G12H29 is indicated by a red dot. This source was already a target for spectroscopic sub-mm follow-up of candidate lensed galaxies (see e.g. Negrello et al. 2010). A CO spectroscopic redshift of z = 3.259 has been obtained for this source (Harris et al. 2012; Van der Werf et al., in prep.). Recent LABOCA data do suggest the presence of other sources in the same clump with SPIRE-to-870 flux ratios that match those of the z = 3.26 source, providing indirect evidence for other sources associated with it at the same redshift (Clements et al., in prep.).

The SDSS DR7 (Abazajian et al. 2009) shows a dense clump just on top of G12H29. Fu et al. (2012) have studied G12H29 and the SDSS clump in detail and determined the lensing nature of HATLAS J114637.9-001132 during a Keck laser guide star adaptive optics imaging (J and Ks-band) programme of bright Herschel 500   μm sources from H-ATLAS. The Ks-band image shows complex filamentary structures that do not appear at J-band. Observations with the SubMillimeter Array (SMA) reveal two 880 μm sources with flux densities of 31 mJy and 27 mJy, separated by 3′′. In their work, Fu et al. (2012) show that the observations can be nicely explained by a lens model in which the lens is a rather complex system located at z ~ 1. The photometric redshifts of the optical sources around the lensed object have been determined using SDSS ugriz + UKIDSS YJHK photometry using the publicly available photo-z code EAZY (Brammer et al. 2008). The Keck K-band data were not used for the photo-z calculation. For the two central lensing galaxies, their blended photometry indicates a photo-z of 1.076 (the 68% confidence interval is 0.982 to 1.305), which is substantially higher than the SDSS DR8 photo-z (0.71). For a more detailed discussion on the lens, the reader is referred to Fu et al. (2012).

Apart from the SDSS clump already described, there is an unusually rich condensation of other low-flux H-ATLAS objects within the Planck beam centred on G12H29. Several of these sources have red Herschel colours. We have determined photometric redshifts for these objects using the SMM J2135-0102 SED, that was shown to work quite well for z ≥ 1 (González-Nuevo et al. 2012; Lapi et al. 2011). We find that four out of the six H-ATLAS sources that are located inside a circle of radius  arcmin (Planck Collaboration 2011b) centred on G12H29 have 0.75 < zphot < 1.25. We checked that other photo-z methods (e.g. Clements et al., in prep.) produce answers that are broadly consistent with those from the Lapi et al. (2011) method. However, these photo-z are calculated using only the three SPIRE bands and are therefore uncertain. For those H-ATLAS objects that can be matched to SDSS galaxies, we recalculated the photo-z using the optical data. Our results seem to indicate that most of these clump sources have lower redshifts  ⟨ z ⟩  < 0.3 than estimated using SPIRE data alone.

To test if the overdensity we observe around HATLAS J114637.9-001132 is statistically significant, we randomly selected 1000 Herschel sources and counted the number of galaxies in circles with the same radius σ857. The median number of neighbours is 1, with a standard deviation of 1.2. None of the sources in the control sample has a number of neighbours equal to or larger than H-ATLAS J114637.9-001132.

Following a bootstrap-type argument and realizing that we can divide the survey area into 30 257 independent cells of size σ857 around H-ATLAS sources (ordered downwards in flux density at 350 microns), we counted the number of H-ATLAS sources inside each one of these cells. We find that the fraction of cells as populated as the G12H29 clump, or more, is 9.915  ×  10-4 (30 cells). Using a Bayesian approach and the binomial distribution (Wall & Jenkins 2003), we find that the 99.99% confidence interval for that fraction is [0.4422  ×  10-4, 1.8736  ×  10-3] (narrowest interval that includes the mode and encompasses 99.99% cases, S. Andreon, priv. comm.). We conclude that the overdensity observed around G12H29 is statistically significant, but given our limited knowledge about the redshift distribution of the clump objects it is unclear whether the clump is a real association of objects at the same redshift or a random alignment of galaxies at very different distances. Our current data seem to favour this last interpretation (a low-redshift clump at z < 0.3 plus the lens SDSS clump at z ~ 1 plus the lensed galaxy at z = 3.259), but the large uncertainties of photo-z estimates make it impossible to rule out other possibilities at this point.

thumbnail Fig. 12

Planck colours of detected objects in the H-ATLAS phase 1 regions compared to model SED colour tracks for two star-forming and one normal spiral galaxy template. The template SEDs go from z = 0 to z = 4.5, with the dots along the lines spaced by Δz = 0.5. The large dots show the ERCSC sources within the H-ATLAS fields; the 1 sigma error bars for these sources are also included in the plot. A colour code has been assigned based on our classification using the Herschel images: orange dots correspond to sources we classify as galaxies, whereas black dots are classified as Galactic cirrus. H-ATLAS J114637.9-001132 (alias G12H29) is shown in red. For comparison, all ERCSC sources with  | b |  > 20° and with 3σ detections or better are also shown as purple points. Seven of the eight isolated purple points in the lower left corner of the plot correspond to blazars identified in the ERCSC (López-Caniego et al. 2012).

Open with DEXTER

The large Planck beam means that there is considerable potential for source confusion to affect the colours. The H-ATLAS survey has detected at  ≥ 5σ in the Herschel 350 μm band (equivalent to the Planck 857 GHz channel), sources with a total flux density of SH − ATLAS = (0.52 ± 0.01) Jy (taking into account the effect of a Gaussian beam with FWHM = 4.23 arcmin centred on the Planck position) to be compared with the ERCSC flux density SERCSC = 2.15  ±  0.83 Jy. Again allowing for the effect of the Planck beam (with FWHM = 4.47 arcmin in this case), the summed contributions of  ≥ 5σ H-ATLAS sources at 500   μm is SH − ATLAS = (0.44 ± 0.02) Jy. Applying a colour correction of a factor of 1.1 for G12H29 and 1.3 for the other sources (assumed to be at redshift  ≤ 1), we obtain  Jy, compared with SERCSC = (1.36 ± 0.61) Jy. This suggests that Planck measurements are boosted by a positive background fluctuation, which may also account for the fact that only one (out of nine) H-ATLAS sources with flux densities at 350 μm  ≃  300 mJy and z ≥ 1, over the entire H-ATLAS phase 1 area, is associated with a Planck detection. Indeed, background fluctuations of the order of  ≥ 2 sigma are needed to make those sources detectable by Planck. The probability of such an event is so extremely low (just a few per cent for Gaussian fluctuations) that even a single occurrence can be considered a stroke of luck. Since the fluctuations are dominated by confusion and have a strong super-Gaussian tail, the frequency of these fluctuations is substantially higher than expected from Gaussian statistics. However, the probability that this fluctuation happens to be by chance on top of a strongly lensed galaxy is tiny (but hard to quantify without knowing the statistics of fluctuations). Therefore the most likely scenario is the one in which H-ATLAS J114637.9-001132 is associated with a clump of sources, most of which fall below the H-ATLAS detection limit, but their total integrated flux is seen as a positive fluctuation by Planck, due to its relatively large beam.

Figure 12 shows the distribution of F545 / F353 vs. F857 / F545 colours for the ERCSC sources in the H-ATLAS fields (black, orange, and big red dots) and the rest of ERCSC sources with  | b |  > 20° and with 3σ detections or better (small purple dots), compared to model SED colour tracks for two star-forming and one normal spiral galaxy templates going from z = 0 to z = 4.5 with a z = 0.5 interval. The black dots correspond to the sources we have classified as Galactic cirrus, whereas orange dots denote sources that are thought to be truly extragalactic. The red dot corresponds to G12H29. Its isolated position in the diagram suggests that there will be very few other objects like this in the ERCSC5. However, the large photometric errors in Fig. 12 make it hard to extract strong statements from it.

6. Conclusions

A cross-correlation of the Planck ERCSC with the catalogue of Herschel-ATLAS sources detected in the phase 1 fields, covering 134.55   deg2, has highlighted several problems that need to be solved to correctly interpret the data from the Planck sub-mm surveys.

  • Contamination by diffuse Galactic emission is a serious problem,as demonstrated by the fact that even in a region ofmoderate Galactic emission (GAMA-09) all857 GHz ERCSC sources seem to be associatedwith cirrus. Therefore, to estimatee.g. the number counts of extragalactic sources,it is crucial to carefully select regions of lowGalactic emission. A cirrus flag  ≥ 0.125 and the “extended” flag, as defined in Planck Collaboration (2011b), are remarkably effective in picking up probable cirrus, but are not 100% reliable.

  • We find a good, essentially linear, correlation between ERCSC flux densities at 857 GHz and SPIRE flux densities at 350 μm above SERCSC ≃ 1.5 Jy. We also find a good correlation between ERCSC flux densities at 545 GHz for sources SERCSC ≥ 1 Jy and SPIRE flux densities at 500 μm, after a colour correction has been applied to SPIRE flux densities to take into account the different central wavelengths of the bands. Excluding the large edge-on disk galaxy NGC 5746, whose H-ATLAS flux density is different form the ERCSC values probably due to resolution and background subtraction systematic effects, we find  ⟨ SH − ATLAS − SERCSC ⟩  ≃ 0.1 Jy at 857 GHz and  Jy at 545 GHz. ERCSC flux densities are affected by flux boosting and have  ≥ 30% uncertainties below  ~1.3 Jy. The relative difference between Herschel and Planck flux densities is compatible with these error levels and the calibration uncertainties of both experiments. The mean difference between ERCSC and H-ATLAS positions for 857 GHz sources brighter than 1.5 Jy is 0.42 arcmin, with a dispersion of 0.17 arcmin, confirming the accuracy of ERCSC positions.

  • In addition to the contamination from Galactic thermal dust emission, the Planck sub-mm surveys are limited by confusion from faint sources within the beam, as expected (e.g. Negrello et al. 2004; Fernandez-Conde et al. 2008). A significant contribution to confusion fluctuations is clustering (Planck Collaboration 2011c). Occasionally, the confusion fluctuations may be dominated by a single protocluster of star-forming galaxies (Negrello et al. 2005). Alternatively, the confusion fluctuations may be caused by apparent clustering due to a random alignment of galaxies at different redshifts. We presented evidence suggesting that at least one object with an anomalous contamination from confusion fluctuations has been detected within the H-ATLAS phase 1 fields. This source is a mixture of a strongly lensed galaxy at z = 3.26 surrounded by a statistically significant overdensity of faint galaxies detected by SPIRE. This unusual overdensity of faint galaxies plus an excess of confusion fluctuations at the same position has made it possible for the ERCSC to detect a high-redshift lensed galaxy that would otherwise be below the Planck detection limit. The available information is insufficient to reliably estimate the redshift of the galaxies in this clump, although there are some indications that at least the lensing galaxies are at z ~ 1. Upcoming PACS photometry and near-IR follow-up of the galaxies in this clump will allow us to better constrain the photometric redshifts of its galaxies. If the rest of the galaxies of the clump could be shown to be at the same redshift, this would represent an example showing the power of combining Planck and Herschel data. Such a combination may open a new window for the study of cluster evolution, since the main searches carried out so far at similar redshifts have selection functions that are different to that of sub-mm surveys. X-ray and Sunyaev-Zel’dovich (SZ) effect surveys preferentially find massive and evolved structures, dominated by passive early type galaxies; the body of work on SZ-detected clusters is growing quickly (Planck Collaboration 2011d, 2012; Williamson et al. 2011; Story et al. 2011; Stanford et al. 2012; Stalder et al. 2012) and maybe in a near future a number of objects similar to those discussed here will be directly observed. Optical/nearIR cluster-finding algorithms, depending on what detection technique is used, can be biased to red evolved galaxies (e.g. red sequence fitting) but this is not always the case (van Breukelen et al. 2006). The sub-mm selection could find clusters with a high level of star-formation activity, thus shedding light on this poorly known phase of their evolution (Michałowski et al. 2010).

  • Although our statistics are too poor to reach definite conclusions, simple source-blending seems to be a less frequent problem: in only one case has an ERCSC 857 GHz source been resolved by Herschel into two similarly bright objects, and in general, the contribution of lower luminosity H-ATLAS sources to the flux density measured by Planck was minor (see Fig. 8).

In conclusion, we find that the higher sensitivity and higher angular resolution H-ATLAS maps provide key information for the interpretation of candidate sources extracted from Planck sub-mm maps. The phase 1 survey considered in this paper represents  ~ 1 / 4 of the full H-ATLAS (550   deg2). Therefore, the results presented here will be substantially improved once the H-ATLAS survey is completed.

Of special interest is the possibility that Planck may be able to sample the tail of the distribution of high-z over-densities, providing unique information about both the early evolution of large-scale structure and galaxy formation and evolution in high-density environments.


1

For comparison, the resolution at the SPIRE 350 μm band is FWHM = 29.4 arcsec, whereas the ERCSC nominal beam at 857 GHz is 4.23 arcmin.

2

This refers to the absolute calibration of the instrument. For individual sources, however, the uncertainty in the photometry is affected by other factors such as the uncertainty on the beam shape and the possibly extended nature of the source, leading to errors that can be as large as 30%, as cited by the Explanatory Supplement to the Planck ERCSC.

3

The SPIRE Observers’ Manual is available from the Herschel Science Centre: herschel.esac.esa.int/Docs/SPIRE/pdf/spire_om.pdf

4

Unless otherwise stated, the beam is a circle with radius .

5

Another interesting feature of this diagram, somewhat beyond the focus of this paper, are the eight isolated purple dots that appear in the lower left part of the plot: seven of them correspond to blazars identified in the ERCSC (López-Caniego et al. 2012).

Acknowledgments

The Herschel-ATLAS is a project with Herschel, which is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA. The H-ATLAS website is http://www.h-atlas.org/. D.H. and M.L.C. acknowledge partial financial support from the Spanish Ministerio de Ciencia e Innovación project AYA2010-21766-C03-01 and the Consolider Ingenio-2010 Programme project CSD2010-00064. D.H. also acknowledges the Spanish Ministerio de Educación for a José Castillejo’ mobility grant with reference JC2010-0096 and the Astronomy Department at the Cavendish Laboratory for their hospitality during the elaboration of this paper. The Italian group has been supported in part by ASI/INAF agreement No. I/009/10/0 and by INAF through the PRIN 2009 “New light on the early Universe with sub-mm spectroscopy”. F.J.C. acknowledges partial financial support from the Spanish Ministerio de Ciencia e Innovación project AYA2010-21490-C02-01.

References

All Tables

Table 1

ERCSC sources at 857 GHz and their H-ATLAS counterparts (inside a 4.23 arcmin radius circle around the ERCSC position) at 350 μm.

Table 2

ERCSC sources at 545 GHz and their H-ATLAS 500 μm counterparts inside a 4.47 arcmin radius circle around the ERCSC position.

All Figures

thumbnail Fig. 1

Positions of the ERCSC 857 GHz sources in the 350 μm H-ATLAS GAMA-09 field. The source at coordinates 0857+3 has not been included in this analysis because it lies on the poorly sampled edge of the field.

Open with DEXTER
In the text
thumbnail Fig. 2

Positions of the ERCSC 857 GHz sources in the 350 μm H-ATLAS GAMA-12 field.

Open with DEXTER
In the text
thumbnail Fig. 3

Positions of the ERCSC 857 GHz sources in the 350 μm H-ATLAS GAMA-15 field.

Open with DEXTER
In the text
thumbnail Fig. 4

Position of the ERCSC sources (black ovals) superimposed on the IRIS 100 micron map around the GAMA-09 H-ATLAS field.

Open with DEXTER
In the text
thumbnail Fig. 5

SPIRE images at 350 μm showing postage stamps of the Planck ERCSC sources at 857 GHz in the H-ATLAS GAMA-09 field. The images are 400 arcsec wide and the red crosses indicate the ERCSC position for each source. The full ERCSC name of the objects has been abbreviated in the captions for conciseness.

Open with DEXTER
In the text
thumbnail Fig. 6

SPIRE images at 350 μm showing postage stamps of the Planck ERCSC sources at 857 GHz in the H-ATLAS GAMA-12 field. The images are 400 arcsec wide and the red crosses indicate the ERCSC position for each source.

Open with DEXTER
In the text
thumbnail Fig. 7

SPIRE images at 350 μm showing postage stamps of the Planck ERCSC sources at 857 GHz in the H-ATLAS GAMA-15 field. The images are 400 arcsec wide and the red crosses indicate the ERCSC position for each source.

Open with DEXTER
In the text
thumbnail Fig. 8

ERCSC 857 GHz flux densities compared with the 350 μm flux densities of the brightest H-ATLAS sources inside the Planck beam (squares). The dashed line indicates the SATLAS = SERCSC identity. The filled circles show the summed flux densities, weighted with a Gaussian beam centred on the ERCSC position and with FWHM = 4.23 arcmin (Planck Collaboration 2011b). Only the ERCSC sources with at least one Herschel counterpart with flux density SH − ATLAS > 0.25 Jy are shown in this plot (see text for further details). The outlier with ERCSC flux density  ~10 Jy is the edge-on spiral galaxy NGC 5746 (object #28 in Table 1, see discussion in the main text). The outlier with ERCSC flux density  ~2 Jy is the G12H29 (source #18 in Table 1), to be discussed in Sect. 5.

Open with DEXTER
In the text
thumbnail Fig. 9

ERCSC flux densities at 545 GHz compared with the colour-corrected 500 μm flux densities of the brightest H-ATLAS sources inside the Planck beam (squares). The dashed line indicates the SATLAS = SERCSC identity. The filled circles show the colour-corrected summed flux densities, weighted with a Gaussian beam centred on the ERCSC position and with FWHM = 4.47 arcmin (Planck Collaboration 2011b). Only the ERCSC sources with at least one Herschel counterpart with (not colour-corrected) flux density SH − ATLAS > 0.25 Jy are shown in this plot. The outlier with ERCSC flux density  ~ 3.5 Jy is the edge-on spiral galaxy NGC 5746. The outlier with ERCSC flux density  ~ 1.5 Jy is the G12H29 source to be discussed in Sect. 5.

Open with DEXTER
In the text
thumbnail Fig. 10

Clump around G12H29 as seen by Herschel at 250 (left), 350 (centre) and 500 μm (right). The field of view is 600 arcsec wide and is centred at the position of the 857 GHz ERCSC source PLCKERC857 G270.59+58.52.

Open with DEXTER
In the text
thumbnail Fig. 11

Herschel three-colour image of the region around HATLAS J114637.9-001132 (alias G12H29). Colours are: blue – 250 μm; green – 350 μm; red – 500 μm. The colour scale was chosen so that objects that are red/white are likely to lie at z ~ 3. The brightest such object in this image is the z = 3.259 likely lensed galaxy HATLAS J114637.9-001132. The circle is centred on the Planck source and is the size of the Planck 857 GHz beam. (This figure is available in color in the electronic form.)

Open with DEXTER
In the text
thumbnail Fig. 12

Planck colours of detected objects in the H-ATLAS phase 1 regions compared to model SED colour tracks for two star-forming and one normal spiral galaxy template. The template SEDs go from z = 0 to z = 4.5, with the dots along the lines spaced by Δz = 0.5. The large dots show the ERCSC sources within the H-ATLAS fields; the 1 sigma error bars for these sources are also included in the plot. A colour code has been assigned based on our classification using the Herschel images: orange dots correspond to sources we classify as galaxies, whereas black dots are classified as Galactic cirrus. H-ATLAS J114637.9-001132 (alias G12H29) is shown in red. For comparison, all ERCSC sources with  | b |  > 20° and with 3σ detections or better are also shown as purple points. Seven of the eight isolated purple points in the lower left corner of the plot correspond to blazars identified in the ERCSC (López-Caniego et al. 2012).

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.