Optically Faint Radio Sources: Reborn AGN?
Mercedes E. Filho, Jarle Brinchmann, Catarina Lobo, Sonia Antón
AAstronomy & Astrophysics manuscript no. hawki˙astroph c (cid:13)
ESO 2018October 22, 2018
Optically Faint Radio Sources: Reborn AGN?
Mercedes E. Filho , Jarle Brinchmann , , Catarina Lobo , , and Sonia Ant´on , Centro de Astrof´ısica da Universidade do Porto, Rua das Estrelas, 4150–762 Porto, Portugal Leiden Observatory, University of Leiden, PO Box 9513, NL–2300 RA Leiden, The Netherlands Departamento de F´ısica e Astronomia, Faculdade de Ciˆencias da Universidade do Porto, Rua do Campo Alegre, 687,4169–007, Porto, Portugal Centro de Investiga¸c˜ao em Ciˆencias Geo-Espaciais, Faculdade de Ciˆencias da Universidade do Porto, Porto, Portugal SIM, Faculdade de Ciˆencias da Universidade de Lisboa, Lisboa, PortugalAccepted 2011. Received 2011; in original form 2011
ABSTRACT
We have discovered a number of relatively strong radio sources in the field-of-view of SDSS galaxy clusters whichpresent no optical counterparts down to the magnitude limits of the SDSS. The optically faint radio sources appearas double-lobed or core-jet objects on the FIRST radio images and have projected angular sizes ranging from 0.5 to1.0 arcmin. We have followed-up these sources with near-infrared imaging using the wide-field imager HAWK-I on theVLT. K s -band emitting regions, about 1.5 arcsec in size and coincident with the centers of the radio structures, weredetected in all the sources, with magnitudes in the range 17–20 mag. We have used spectral modelling to characterizethe sample sources. In general, the radio properties are similar to those observed in 3CRR sources but the optical-radioslopes are consistent with moderate to high redshift ( z <
4) gigahertz-peaked spectrum sources. Our results suggest thatthese unusual objects are galaxies whose black hole has been recently re-ignited but retain large-scale radio structures,signatures of previous AGN activity.
Key words.
Galaxies: active
1. Introduction
In the past fifteen years, deep and extensive radio obser-vations of the Hubble Deep Field (HDF; Richards et al.1999), and surveys like the Very Large Array (VLA) 8.4GHz survey (Fomalont et al. 2002),
Phoenix (Hopkins etal. 2003) and
Atlas (Norris et al. 2006) have uncovereda number of previously uncatalogued radio sources. Theseare characterized by flux densities that range from severalmicrojansky to hundreds of millijansky and by projectedangular sizes that can be as large as several megaparsec.Cross-correlation studies have shown that as many as 10-15% of the compact radio sources have faint or no opti-cal or infrared counterparts (Hopkins et al. 2003; Sullivanet al. 2004; Higdon et al. 2005, 2008; Middelberg et al.2008a; Garn & Alexander 2008; Huynh et al. 2010; Norriset al. 2011; Banfield et al. 2011; Zinn et al. 2011; see alsoMachalski et al. 2001; Rigby et al. 2007). While a significantfraction of the sub-millijansky radio population appears tobe faint star-forming galaxies (e. g. Haarsma et al. 2000),radio sources that are faint in the optical or infrared aregenerally consistent with high redshift ( z > =70 kms − Mpc − andΩ m =0.27 (Larson et al. 2011). a r X i v : . [ a s t r o - ph . C O ] O c t Filho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN?
2. Sample Selection
The SDSS maxBCG cluster catalog (13 823 clusters;Koester et al. 2007) was used as the seed catalogue forour study. The radio sources were selected in the followingmanner: – We first cross-correlated the cluster sample with theFIRST catalogue (White et al. 1997); – We then retained 291 clusters which contained at leastone FIRST object within 1 Mpc in projection from thebrightest cluster galaxy (BCG); – Visual inspection of the selected fields provided a sub-sample of radio galaxies with extended double-lobed orcore-jet radio morphology; – For these, when a secure SDSS identification was made,the existing spectroscopy or multi-band photometry wasused to optically characterise the source and assesswhether the source belongs to the cluster or not.During this process, we have identified eight radiosources with no optical SDSS counterpart, which indicatesthat their r AB -band magnitudes must be over 22 mag. Theradio sources are located in the fields-of-view of clusterswith redshifts ranging from 0.17 to 0.28. The radio sourcesare further characterized by their radio-loudness (as definedby a large radio flux density relative to the optical and NIR;see Section 5.2 and 5.3), arcsec-scale FR II-type or core-jetradio morphology and relatively strong FIRST flux densi-ties in the range 1 mJy < F . <
80 mJy.Table 1 contains the radio data for the sample sources.Radio sources are identified by the sequential number ofthe cluster field to which they belong in projection, asit appears in the maxBCG cluster catalogue (Koester etal. 2007). Radio source components are identified by theposition and nature relative to the overall radio sourcestructure. For the only core-jet radio source in our sam-ple (maxBCG 3131; Fig. 1), we have designated the ra-dio emission regions North Comp and South Comp 1 and2 (Table 1). In addition, maxBCG 2596 is the only radiosource with a radio core detection (Nucleus; Table 1). Wenote that the radio properties of the optically faint radiosources are dominated by the extended (lobe) radio emis-sion.Table 2 contains redshift information on the maxBCGclusters containing our sample sources in their field-of-view.
3. Near-Infrared Observations and Data Reduction
The unidentified radio sources were followed-up with NIRimaging using the wide-field imager HAWK-I on the ESOUT4 of the VLT. The proposal, with reference 081.A-0624(A), was awarded a total of 2.7 hours observing timeduring period 81. The total integration time on-source wasabout 1 000 seconds. The data were reduced using theHAWK-I data reduction pipeline (version 1.4.2), which in-cludes recipes for darks, flats, zero-point computation, de-tector linearity, illumination, additional calibration, distor-tion corrections, jittering and image stitching. The reducedimages were then astrometrically calibrated using the TwoMicron All Sky Survey (2MASS) catalogue and tools pro-vided in the Graphical Astronomy and Image Analysis Tool(GAIA; version 4.3–0). The astrometric calibration is goodto better than one arcsec, which is sufficient for our pur-poses. Aperture photometry was performed on the images using the GAIA se xtractor program. K s -band magnitudes(in the Vega magnitude system) were estimated using zero-points given in the corresponding quadrant of the stan-dard star images and assuming no significant color termwith respect to 2MASS (ESO HAWK-I Science Verification,November 7, 2007: Note on Photometry ).Results of the NIR observations are presented inTable 3. NIR sources are identified by the sequential num-ber of the cluster field to which they belong in projection, asit appears in the maxBCG cluster catalogue (Koester et al.2007). NIR counterparts of the centers of the radio struc-tures are designated as components ”N”. Typically the sizesof these central NIR components, as provided by GAIA se xtractor aperture photometry, are ∼ ∼
4. SDSS and Stripe 82 Data
The SDSS (York et al. 2000) is a survey covering more than35% of the sky providing deep, photometric observations in5 bands ( u , g , r , i and z ; AB magnitude system) of about500 million objects and spectra for more than 1 millionsources.We have used the SDSS Data Release 7 (DR7;Abazajian et al. 2009) to attempt to identify opticalcounterparts of the radio and NIR components of oursources. Identification was based on the coincidence ofthe positions to less than 1 arcsec, followed by visualinspection of the images. None of the radio components(Table 1) or NIR emission regions coincident with thecenters of the radio structures (components N; Table 3)were detected in the SDSS, providing an r AB -band upperlimit of ∼
22 mag. However, several NIR sources near orwithin (in projection) but not associated with (see Section5.1) the FIRST radio sources, were identified.Stripe 82 is a ∼
300 deg region along the celestial equa-tor, that has been imaged by a number of different col-laborations (with the same set-up and in the same bands)over one hundred times, providing coadded optical data twomagnitudes deeper than the single-epoch SDSS data.Three of our fields are located within the Stripe 82 re-gion, namely maxBCG 3131, maxBCG 6167 and maxBCG8495. For these fields, we have looked for possible op-tical counterparts to our NIR components. Identificationwas based on visual inspection of all possible counterpartswithin 5 arcsec of the NIR source. For each of the threefields, we have detected the faint optical Stripe 82 counter-parts of the NIR components coincident with the centersof the radio structures (components N; Table 3). We havealso identified the Stripe 82 optical counterparts of severalother NIR sources near or within (in projection) but notassociated with (see Section 5.1) the FIRST radio sources.Stripe 82 optical counterparts were also sought directlyfor the extended radio components. In this procedure, Stripe 82 yielded matches within 5 arcsec of the positionsof the radio components in maxBCG 3131 and maxBCG6167. For maxBCG 3131, radio components North Compand South Comp 1 coincide with the position of opticalsources; these are rather faint ( r AB -band magnitudes of22.7 and 23.7 mag) extended sources. As for maxBCG 6167,only its North Lobe radio component flags a cross-match inStripe 82; this is an extremely faint extended object withan ill-determined r AB -band magnitude of 24.5 ±
5. Discussion
The FIRST, NIR, SDSS and Stripe 82 (when available) r AB -band images of the fields containing the samplesources are presented in Fig. 1–4. HAWK-I observationswere successful, as NIR emission, coincident with thecenters of the radio structures (components N; Table 3),was detected in all the targets. In the particular case ofmaxBCG 2596, the central NIR emission source (compo-nent N; Table 3) coincides with the radio core (Nucleus;Table 1), while in maxBCG 3131, the central NIR source(component N; Table 3) falls between the radio componentsNorth Comp and South Comp 1 (Table 1). We have alsoinvestigated all NIR sources near or within (in projection)but not associated with the FIRST radio sources. In thefollowing, we discuss the HAWK-I sources individually. maxBCG 2596 In this object we have identified the NIRcounterpart (component N; Table 3) to the radio core(Nucleus; Table 1) at the center of the radio structure,which is the brightest NIR component in our sample, butstill falls below the ∼
22 mag r AB -band limit of the SDSS.We have also detected NIR sources (components W and E;Table 3) coincident with the termination of the radio lobes(components East and West Lobe, Table 1), the latter ofwhich display extended NIR morphology. The NIR E andW components are detected in the SDSS image, wherethey are identified with extended, galaxy-type objects –J010202.12-103258.3 and J010159.33-103311.2 (Table 4)– with somewhat similar r AB -band magnitudes of ∼ ∼ z =0.09 and 0.19; Table 4). Because radio lobes do notproduce significant amounts of optical emission and due tothe redshift discrepancy between the radio source ( z =0.56;Table 7; see also Section 5.3) and the correspondingoptical/NIR components ( z =0.09 and 0.19; Table 4),we conclude that this is a chance alignment. However,according to their redshift, the NIR E component lies inthe cluster foreground, whereas the SDSS counterpartidentified with the NIR W component has a photometricredshift ( z =0.19; Table 4) consistent with both the photo-metric redshift of the cluster field determined by Koester et al. (2007) and the spectroscopic redshift of the BCGassociated with the same cluster ( z ∼ g − r colour index typical of anearly-type galaxy (Fukugita et al. 1995) at the clusterdistance ( z ∼ maxBCG 3131 The HAWKI-I image shows a complexNIR morphology. All of the NIR components can beidentified on the Stripe 82 images with r AB -band magni-tudes ranging from 19 to 24 mag (Table 4). The centralNIR source (component N; Table 3), coincident with thecenter of the radio structure (between North Comp andSouth Comp 1 components; Table 1) is the faintest, witha very blue g − r =0.35 colour index, consistent with theoptical colour of a quasar (Richards et al. 2001) at theestimated redshift of our radio source ( z =1.03; Table 7;see also Section 5.3). This result is confirmed by thephotometric redshift estimate for the Stripe 82 detection( z =1.57; Table 7; see also Section 5.3). There is diffuseNIR emission to the East (components E1, E2 and E3;Table 3), coincident with the outline of the Northernradio component (component North Comp; Table 1).From careful inspection of the Stripe 82 images, it canbe seen that components E1 and E2 are two resolvedcomponents of the same edge-on foreground galaxy (rel-ative to the cluster in this field), located at a redshift of z ∼ r AB -band 21.8 mag SDSS extendedgalaxy-type source J003448.99-002130.9 (Table 4), locatedat a photometric redshift of z =0.58 (Table 4), higherthan the one attributed to the cluster dominating thefield-of-view ( z =0.27; Table 2), but closer than our radiosource estimation ( z =1.03; Table 7; see also Section 5.3).Extended NIR galaxy-type emission is also present dueSouth (components S1 and S2; Table 3) of the Southernradio components (components South Comp 1 and 2;Table 1) and identified with r AB -band ∼ z =0.46; Table 4) than the one attributed to the cluster( z =0.27; Table 2), but still not coinciding, by far, with theestimated redshift of our radio source ( z =1.03; Table 7;see also Section 5.3). Furthermore, there are points ofaligned Southeast-Northwest NIR emission perpendicularto the orientation of the radio structure – components D1and D2 (Table 3). There is a r AB -band ∼
23 mag Stripe 82stellar-like identification of component D1 – J003448.41-002136.0 (Table 4) – and a ∼
21 mag SDSS identificationof component D2 – J003447.71-002126.9 (Table 4) – anextended galaxy-type object with a photometric redshiftof z ∼ z =0.27; Table 2), but at a similar redshift tocomponent E3 ( z =0.58; Table 4). There is also evidencefor faint NIR emission (component L; Table 3) within (inprojection) the Northern radio component (componentNorth Comp; Table 1), which has been identified with a r AB -band ∼
23 mag stellar-like object in Stripe 82 (Table 4). maxBCG 6167
Like its radio emission, the NIR emissionassociated with this source is faint. We detect centralNIR emission (component N; Table 3) coincident with the
Filho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? center of the radio structure. An optical counterpart hasbeen detected in Stripe 82 – J002452.65-005201.7 (Table 4)– as a galaxy-type source with an r AB -band magnitude ofabout 25 mag (Table 4). This is the faintest radio, NIRand optical (Stripe 82) source in our sample. Its extreme g − r blue colour of − ∼ z =1.13; Table 7; see also Section 5.3). This result isconfirmed by the photometric redshift estimate for theStripe 82 detection ( z =1.00; Table 7; see also Section 5.3). maxBCG 8495 The NIR image shows a ”double nuclear”emission region (components Na and Nb; Table 3) atthe center of the radio structure. A similarly extendedgalaxy-type morphology is seen in the Stripe 82 image,although Stripe 82 identifies a sole, quite faint, component– J004359.47+001230.4 (Table 4) – with an r AB -bandmagnitude of about 24 mag and g − r =1.31 (Table 4).Given its red colour and the estimated redshift of theradio source ( z =0.96; Table 7; see also Section 5.3), wecan conclude that the Stripe 82 optical counterpart has a g − r colour index suggestive of a late-type spiral galaxy(Fukugita et al. 1995), which is coherent with the diskyappearance in the NIR. The red colour generally excludes aquasar host (Richards et al. 2001) at the estimated redshiftof our radio source ( z =0.96; Table 7; see also Section 5.3).These results are confirmed by the photometric redshiftestimate for the Stripe 82 detection ( z =0.88; Table 7; seealso Section 5.3). There is also a small patch of unidentifiedNIR emission (component D; Table 3) near (in projection)the center of the radio source. maxBCG 10942 There is a central NIR emission region(component N; Table 3) which is coincident with the centerof the radio structure and some unidentified NIR emissionto the West (component W; Table 3) of the Northern radiolobe (component North Lobe, Table 1), in projection. maxBCG 11079
This source shows a complex, faint NIRmorphology. There is evidence for central NIR emission(component N; Table 3) coincident with the center ofthe radio structure, which has a disk-like appearance.There are also peaks of unidentified NIR emission within(in projection) and/or near the edge of the Southern(components S1, S2 and S3; Table 3) and Northern radiolobes (component L; Table 3). maxBCG 11390
There is a clear NIR detection (compo-nent N; Table 3) at the position of the center of the radiostructure and an unidentified NIR Eastern (component E;Table 3) source. maxBCG 11780
There is an unidentified region of NIRemission within (in projection) the Northern radio lobe(component L; Table 3) and at the position of the centerof the radio structure (component N; Table 3).All the NIR images reveal central emission, relative to theradio structure (components N; Table 3), that we inter-pret as thermal host-related emission (see Section 5.2). TheNIR magnitudes of these sources range from 17–20 mag(Table 3) and NIR sizes, provided by GAIA se xtractor aperture photometry, are typically ∼ We present, in Fig. 5 (see also Table 6), a plot of the NIRemission versus the FIRST flux density. With the excep-tion of maxBCG 8495 component N (Table 3; see Section3), the plotted NIR flux (in the Vega magnitude system)corresponds to GAIA se xtractor aperture photometry ofthe NIR components coincident with the radio structurecenters (components N; Table 3). Total radio flux densi-ties were calculated as the sum of the flux densities of allthe FIRST radio components (Table 1). We note in pass-ing that the extended radio emission dominates the totalradio flux density in all of our sources even when we detecta radio core (maxBCG 2596) or core-jet (maxBCG 3131)source. We have estimated a radio ”core” flux density using sao image ds9 (version 5.1) funtools on circular regionsof approximately the same size ( ∼ × rms (root mean square, as a measure of thenoise level) given in the corresponding FIRST images.We have run two correlation tests on the data – thegeneralized Kendall’s τ correlation coefficient estimation(BHK) and the correlation probability by Cox’s propor-tional hazard model (Cox-Hazard), as implemented inIRAF (version 2.14.1; Table 5). Both the BHK and Cox-Hazard tests show a >
50% (radio core flux density) and >
80% (total radio flux density) probability that there is nocorrelation between the FIRST and NIR emission. At suchscales and given the radio morphology, the FIRST radioemission is sampling synchrotron emission associated withthe AGN. The lack of correlation between the radio and theNIR emission therefore reflects the thermal nature of theNIR source, likely arising from the host galaxy. This is incontrast to 3CR galaxies (Spinrad 1985), which commonlyshow non-thermal NIR central regions associated with theAGN (Baldi et al. 2010).The infrared-radio correlation, represented by the pa-rameter q , is an indicator of the dominant emission mech-anism in a galaxy (Yun et al. 2001); an AGN contributionserves, in general, to lower the value of q . Empirically it hasbeen shown that starburst-dominated galaxies at redshiftsbelow 2 typically show a ratio of about q ∼ q ∼ q = log F IR F . and IR (infrared) is measured at 24 µ m and a combinationof 60 and 100 µ m, respectively. A clear enhancement of theradio emission relative to a pure starburst galaxy is gener-ally attributed to the presence of an AGN. We note that fortypical galaxy starbursts, AGN and composite SEDs (e.g.Huynh et al. 2010), the difference between the 24 µ m and2.2 µ m flux may be as high as two orders of magnitude.Therefore, in the very extreme case, for a starburst-drivengalaxy q ∼ − K s -band. We have proceeded to estimate the infrared-radio cor-relation using the K s -band magnitude of the NIR N com-ponents (Table 3) and the radio ”core” and total radio fluxdensities, as defined above. The results are contained inTable 6. Our sources show values that are generally consis-tent with the presence of an AGN ( q < At first glance, the radio morphologies (Fig. 1–4) sug-gest that our optically faint radio sources may be sim-ilar to the radio sources found in the Third CambridgeRevised Catalogue of Radio Sources (3CRR) catalog . Wepresent in Fig. 6–8 plots of the redshift, LLS, radio powerand K - (gigahertz-peaked spectrum sources and 3CRR) or K s -band (our sources) emission. For reasons to be illus-trated below, we also include the gigahertz-peaked spec-trum source (GPS) sample of Labiano et al. (2007). TheLLS for the 3CRR sources have been obtained from thelargest angular size (LAS) measurements and for the GPSsources the LLS has been obtained from O’Dea & Baum(1997), when a measurement is available. The quoted radiopowers are not k -corrected (Table 7). Plotted radio powersfor the 3CRR are for the radio cores, while for the GPSsources these are the total radio powers (Labiano et al.2007). For our optically faint radio sources, we use the ra-dio ”core” powers (Table 7) and the total (FIRST) radiopowers (Table 7) as defined in Section 5.2. For simplic-ity we have assumed a flat radio spectrum to normalizethe radio powers to 5 GHz. K -band magnitudes for theGPS sources were obtained by interpolation for a subset ofthese sources with NED NIR photometry. K -band magni-tudes for the 3CRR sample are emission-line- and aperture-corrected magnitudes for a subset of 3CRR sources with K -band observations . K s -band magnitudes for our sam-ple sources were obtained for the central NIR components(components N; Table 3) coincident with the centers of theradio structures. All K - and K s -band magnitudes are inthe Vega system. We expect a small colour term between K s - and K -band: | K - K s | < z > http://3crr.extragalactic.info/cgi/database are unknown, we adopt a similar approach to Huynh et al.(2010): we compare our sample data with a sparse set ofSED templates with NIR and radio data available in NEDor in the literature. We have found that our data required aconsiderably larger set of SED templates than that used byHuynh et al. (2010). We have compared our source SEDswith SED templates from the Third Cambridge Catalogueof Radio Sources (3CR; 169 sources; Spinrad 1985), a GPSsample (32 sources; Labiano et al. 2007) and galaxies thatspan a large range of mid-infrared galaxy classifications(Spoon et al. 2007; their Fig. 1): M82 – a star-forming FR Igalaxy (class 2C), Arp 220 – a Seyfert-type UltraluminousInfrared Galaxy (ULIRG; class 3B), Mrk 231 – a dustyAGN-dominated (Seyfert 1) ULIRG (class 1A), Mrk 1501 –a Seyfert 1.2 flat-spectrum radio source, 3C 305 – a Seyfert2 FR I galaxy, Mrk 668 – a Seyfert 1.5 GPS source, 3C 273– a radio-loud quasar and 3C 295 – a narrow-line FR IIradio galaxy.Our approach is not to rigorously derive photometricredshift estimates, but rather to assess what class of objectsour sample galaxies are more consistent with and what theimplied redshift range is. To that end, we commence bycomparing the radio-to- K or - K s -band luminosity. Fig. 9shows the results of this exercise. The quoted NIR flux (inthe Vega magnitude system) corresponds to that of the cen-tral host galaxy (components N; Table 3), coincident withthe centers of the radio structures. The radio flux densityfor our optically faint radio sources is the total FIRST radioflux density (Table 7) as defined in Section 5.2, while forthe remaining sources, the total radio flux density is the 1.4GHz radio flux density as reported in NED. The horizontaldotted lines show the measured flux ratio between the radioand NIR for our sample sources. On top of this (solid anddashed lines) we show the radio-to-NIR ratio for a rangeof different sources, with starburst towards the bottom andluminous radio galaxies towards the top. The large dotswith error bars show the average and 68% spread aroundthe mean for the 3CR (dashed) and GPS (solid) samples.Fig. 9 clearly shows that the majority of our samplesources possess lower radio-to-NIR ratios than the 3CRsample, but much larger than starbursts, except perhapsat extremely high redshifts. However, the latter scenario isruled out because it would imply unrealistically high lumi-nosities in both the radio and optical regime. Viewed as aclass, our sample sources are more consistent with havingcharacteristics similar to that of GPS sources at z <
4. Westress that the heterogeniety of the GPS sample does notalter significantly the results presented in Fig. 9. Whetherwe use only the QSOs ( z >
1) or the entire GPS sample,the result is robust; the median ratio in each redshift binof the GPS sample will always fall below the 3CR sample.For each template SED that the horizontal dotted linesof our sources intersect, we can obtain a redshift estimatefor our sample sources. We do this by first selecting thosetemplate SEDs/redshifts that reproduce the infrared-radioslope to within 10% (approximately consistent with the er-ror estimates). However, it is obvious from Fig. 9 that usingthe radio-to-NIR ratio alone does not provide a unique red-shift. We therefore reject any template SED/redshift solu-tion that would predict optical fluxes (SDSS or Stripe 82)brighter than the 1 σ upper limits.The heterogeneity of the SEDs and the limited obser-vational data does, however, complicate our attempts toestimate rigorous redshifts for our sample sources. We do Filho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? note that Mrk 668 traces fairly closely the mean trend ofthe GPS sources (Fig. 9). We have therefore adopted thisSED template in order to obtain representative redshifts forour sample sources. This might be misleading for individ-ual targets but it is a reasonable approach given that oursample do seem to show typical infrared-radio slope val-ues consistent with the mean GPS sample. The estimatedredshifts obtained through this procedure are included inTable 7. We stress that by using a template GPS source likeMrk 668 provides only an illustrative estimate of the red-shift; there is a degeneracy in the redshift determination.The most robust aspect is that the very lowest redshiftsolutions are excluded when we include the optical upperlimits.We can obtain an independent check on our redshift es-timates by exploiting the fact that three of the NIR sources(components N corresponding to maxBCG 3131, 6167 and8485) appear to have faint counterparts on the Stripe 82images. By combining the u , g , r , i and z (AB magnitudesystem) photometry from Stripe 82 with our NIR photom-etry, we have estimated photometric redshifts using version2.2 of the Le Phare photometric redshift code (Arnouts etal. 1999; Ilbert et al. 2006). The resulting redshift estimatesare included in Table 7. As can be seen, these are in verygood agreement with the estimates obtained using Mrk 668as a template. If we compare these radio source redshift es-timates (Table 7) with those of the clusters and respectiveBCGs (Table 2), we can conclude that, although originallyidentified in the field-of-view of the clusters, the opticallyfaint radio sources in our sample are not members of theSDSS clusters listed in Table 2.An alternative interpretation for the nature of the op-tically faint radio sources is that they are predominantlysteep spectrum radio sources, where the steep spectrumemission of the lobes dominates the radio flux. This wouldfit most observational results but the main obstacle lies inthe radio-to-NIR ratio. While there exists a few steep spec-trum radio sources such as 3C 305 (Fig. 9) that have radio-to-NIR ratios lower or comparable to our sources, most haveconsiderably higher ratios. More generally, double-lobed ra-dio quasars from the Hough & Readhead (1989) sample lie,on average, well above our sample in the radio-NIR dia-gram. Even if we use 3C 305 as a template, this results ina large discrepancy between the redshift inferred from theradio-to-NIR ratio ( z >
3) and that obtained from the LePhare photometric redshift estimate (Table 7).The result is that we have large-scale, relatively pow-erful radio sources similar in radio properties to 3CRRsources, but whose optical-radio slopes are consistent withthose found in GPS sources. GPS sources are compact ( ≤ . ≥
25 W Hz − ) radio sourceswith a well-defined peak ( ∼ ∼ − years), with periods of activity ( ∼ years) andperiods of dormency (Franseschini, Vercellone & Fabian1998). Large-scale radio lobes are produced when the ∼ arnouts/lephare.html AGN jets deposit energy into intergalactic medium (IGM)”cocoons”. If the nuclear activity is halted, the lumi-nosity of the radio lobes drops and the radio spectrumbecomes steeper due to radiation and expansion losses.However, such lobe structures are long-lived, fading awayon timescales of about 10 years (Komissarov & Gubanov1994), comparable to the timescale of the nuclear activityitself. The re-ignition of the jet-forming activity can occurvia internal instability of the accretion disc (Natarajan &Pringle 1998) or gas accretion through an interaction ormerger (Barnes & Hernquist 1996), potentially producinga GPS spectrum typical of a young radio source. If the AGNrestart occurs within a few 10 years, the large-scale radiolobes will still be visible.The above theory predicts a large number of such”double-double” radio sources – 3C 219 (Clarke &Burns 1991; Clarke et al. 1992), 4C 26.35 (Owen &Ledlow 1997), 3C 445 (Kronberg, Wiebelinski & Graham1986; Leahy et al. 1997), 1245+676 (Marecki et al.2003), B1834+620 (Schoenmakers et al. 2000a, 2000c),B0925+420, B1240+389 and B1450+333 (Schoenmakers etal. 2000a) are just a few examples. Some of these sourcesalso show GPS-type spectra – e.g. 0108+388 (Baum et al.1990), 1245+676 (Marecki et al. 2003; Ant´on et al. 2004;Bondi et al. 2004), B1834+620 (Schoenmakers et al. 2000a,2000c). However, GPS sources with large-scale radio emis-sion appear to be rare (Stanghellini et al. 1990, 2005).When compared to these classical ”double-double” ra-dio sources (Schoenmaker et al. 2000a), our optically faintradio sources have, on average, similar projected FIRSTLLS and total FIRST radio powers (Table 7; Fig. 7–8).The similarities occur also with the 3CRR sample in termsof radio morphology, radio power and FIRST LLS (Table 7;Fig. 1–4 and 7–8), although our sources appear, on aver-age, fainter in the K -band (Table 3; Fig. 8). Comparedto the GPS sample (Labiano et al. 2007; see also O’Dea1998 and Stanghellini et al. 2005), our sources have, on av-erage, larger projected FIRST LLS, lower total (FIRST)radio powers (Table 7; Fig. 7–8) and fainter K -band mag-nitudes (Table 3; Fig. 8).The way to reconcile the GPS-type optical-radio slopewith the large-scale radio structure in our sample sourcesis to assume the total FIRST emission as an upper limit tothe radio power of the hypothetical GPS. The fact that wedo not detect significant radio core emission in the FIRSTimages is not deterministic; the moderate redshift of thesources combined with the low signal-to-noise ratio of theFIRST images may mask out the presence of such a ra-dio core. In this scenario, our sources may be regarded asre-ignited AGN, with the large-scale structure a relic of aprevious cycle of black hole activity.Thus based on the present evidence we would argue thatwhile interpreting the optically faint radio sources as steepspectrum radio sources would be possible in some cases,overall our sample appear to show systematically betteragreement with a ”double-double” radio source scenario inthe radio-to-NIR diagram (Fig. 9). To robustly distinguishbetween these two scenarios we do, however, require furtherobservational data for our sources. Deep, high-resolutionradio observations should assess the presence of small-scaleradio structures, signatures of recently re-instigated AGNactivity.These results are somewhat distinct from previous stud-ies of radio sources with compact radio morphology that are ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 7 faint in the optical or infrared (Higdon et al. 2005, 2008;Garn & Alexander 2008; Jarvis et al. 2009; Huynh et al.2010; Norris et al. 2011; Banfield et al. 2011). Both theobserved infrared-radio correlation and SED modelling re-sults suggest that these radio sources are mainly radio-loudAGN at z >
6. Conclusions
In this paper we have presented new, deep near-infrared im-ages of a small sample of optically faint radio sources found,in projection, in low redshift galaxy cluster fields. Thesenew observations have allowed us to detect near-infraredemitting host galaxies with K s ∼ ∼ Filho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN?
Table 1. The Radio Data.
Col. 1: The radio source component corresponding to the SDSS maxBCGcluster field: identification number is the sequential number (Koester et al. 2007). Col. 2: NVSS 1.4 GHz,45 (cid:48)(cid:48) resolution radio position (J2000). Col. 3: NVSS total flux density. Col. 4: FIRST 1.4 GHz, 5 (cid:48)(cid:48) resolutionradio position (J2000). Col. 5: FIRST radio peak flux density. Col. 6: FIRST integrated radio flux density. maxBCG RA Dec NVSS RA Dec FIRST Peak FIRST Int(J2000) (mJy) (J2000) (mJy/beam) (mJy)(1) (2) (3) (4) (5) (6)2596 Total 01:02:00.36 -10:33:03.1 68.9 . . . . . . . . .2596 Nucleus . . . . . . 01:02:00.498 -10:33:04.95 3.80 13.942596 East Lobe . . . . . . 01:02:01.726 -10:32:58.02 9.95 19.602596 West Lobe . . . . . . 01:01:59.682 -10:33:08.81 20.43 32.183131 Total 00:34:48.07 -00:21:31.2 36.8 . . . . . . . . .3131 North Comp . . . . . . 00:34:48.268 -00:21:27.67 16.42 24.913131 South Comp 1 . . . . . . 00:34:47.665 -00:21:36.57 3.75 4.213131 South Comp 2 . . . . . . 00:34:47.226 -00:21:38.28 3.61 4.026167 Total 00:24:52.61 -00:52:05.6 7.6 . . . . . . . . .6167 North Lobe . . . . . . 00:24:52.733 -00:51:54.90 2.05 3.846167 South Lobe . . . . . . 00:24:52.332 -00:52:14.76 1.92 1.878495 Total 00:43:59.54 +00:12:31.2 43.5 . . . . . . . . .8495 North Lobe . . . . . . 00:44:00.178 +00:12:43.26 10.81 17.818495 South Lobe . . . . . . 00:43:58.808 +00:12:16.19 9.94 14.4010942 Total 02:17:57.39 -09:18:22.2 75.2 . . . . . . . . .10942 North Lobe . . . . . . 02:17:57.475 -09:17:54.85 7.52 15.0610942 South Lobe . . . . . . 02:17:57.362 -09:18:28.86 26.94 50.4911079 Total 02:01:28.37 -08:19:55.8 173.2 . . . . . . . . .11079 North Lobe . . . . . . 02:01:28.629 -08:19:47.17 72.38 86.0011079 South Lobe . . . . . . 02:01:28.173 -08:20:05.50 61.92 77.6311390 Total 00:03:28.78 -11:12:55.1 10.8 . . . . . . . . .11390 North Lobe . . . . . . 00:03:28.709 -11:12:49.20 4.68 5.5411390 South Lobe . . . . . . 00:03:28.860 -11:13:02.36 3.90 4.3411780 Total 00:05:57.07 -09:09:01.1 21.7 . . . . . . . . .11780 North Lobe . . . . . . 00:05:57.469 -09:08:48.91 3.69 9.8111780 South Lobe . . . . . . 00:05:56.725 -09:09:08.36 8.25 12.19
Table 2. SDSS maxBCG ClusterRedshifts.
Col. 1: SDSS maxBCGcluster field: identification number isthe sequential number (Koester et al.2007). Col. 2: The photometric red-shift of the SDSS cluster, as deter-mined for a sample of red sequencemember galaxies (Koester et al. 2007;their Table 1). Col. 3: The spectro-scopic redshift of the brightest clustergalaxy when available (Koester et al.2007; their Table 1). maxBCG z photo , cluster z spec , BCG (1) (2) (3)2596 0.18095 0.188503131 0.27005 0.248606167 0.16745 0.162698495 0.22955 0.2187610942 0.26195 . . .11079 0.28085 . . .11390 0.23495 . . .11780 0.20795 0.21456ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 9
Table 3. The Near-Infrared Data.
Col. 1: The NIR source componentcorresponding to the SDSS maxBCG cluster field: identification number is thesequential number (Koester et al. 2007). Col. 2: Near-infrared position (J2000)from the GAIA aperture photometry se xtractor fit (except for maxBCG 8495component N). Col. 3: Ellipticity of the fitted ellipse (except for maxBCG8495 component N). Col. 4: GAIA se xtractor aperture photometry K s -bandmagnitude and error (except for maxBCG 8495 component N). Magnitudesare in the Vega system. Col. 5: Note on the component: N – componentcoincident with the center of the radio structure; SDSS – SDSS detection; S82– Stripe 82 detection. maxBCG RA Dec Ellipticity K s Note(J2000) (mag)(1) (2) (3) (4) (5)2596 N 01:02:00.677 -10:33:02.46 0.056 17.57 ± ± ± ± ± ± ± a ± a ± ± ± ± ± b ± b ± c ± ± ± ± ± ± ± ± ± ± ± ± ± a Stripe 82 images show that components E1 and E2 are two resolved componentsof the same edge-on galaxy. b The maxBCG 8495 Na and Nb components refer to the ”double nuclear” emissionregion detected in the K s -band. c The maxBCG 8495 N component refers to GAIA manual aperture photometry ofa region ( ∼ Table 4. The Optical Data.
Radio (top) or NIR (bottom) component positions. Col. 1: SDSS maxBCG clusterfield and radio or NIR component: identification number is the sequential number (Koester et al. 2007). Col. 2:FIRST or NIR component position (J2000). Col. 3: SDSS or Stripe 82 (S82) detection. Col. 4: IAU designation ofthe SDSS or Stripe 82 counterpart. Col. 5: Separation between the position of our radio or NIR component andthe SDSS or Stripe 82 detection. Col. 6: SDSS or Stripe 82 r AB -band (model) magnitude, corrected for Galacticextinction, and respective error. Magnitudes are in the AB system. Col. 7: g − r colour index. Col. 8: Photometricredshift and error (only available for SDSS detections). Col. 9: Type of SDSS or Stripe 82 source: G – galaxy; S –star. maxBCG RA Dec Survey IAU ID θ r AB g − r z phot Source(J2000) (”) (mag) (mag) type(1) (2) (3) (4) (5) (6) (7) (8) (9)2596 Nucleus 01:02:00.498 -10:33:04.95 . . . . . . . . . . . . . . . . . . . . .2596 East Lobe 01:02:01.726 -10:32:58.02 . . . . . . . . . . . . . . . . . . . . .2596 West Lobe 01:01:59.682 -10:33:08.81 . . . . . . . . . . . . . . . . . . . . .3131 North Comp a ± a ± a ± ± ± ± ± ± ± ± ± ± ± b ± b ± ± ± ± ± ± ± ± ± ± ± c ± a The Stripe 82 identification of an optical source at the position of the radio component is a chance alignment. b Stripe 82 images show that components E1 and E2 are two resolved components of the same edge-on galaxy. c The maxBCG 8495 N component refers to GAIA manual aperture photometry of a region ( ∼ ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 11 Table 5. Correlation Statistics.
Col. 1: Statistical corre-lation method: BHK – generalized Kendall’s τ correlation co-efficient; Cox – correlation probability by Cox’s proportionalhazard model. Col. 2 and 3: Independent and dependent vari-able: NIR pertains to the K s -band fluxes of the NIR com-ponents (components N) coincident with the radio structurecenters (except for maxBCG 8495); FIRST Tot pertains tothe FIRST total radio as the sum of the flux density of all theFIRST components; FIRST Core pertains to the radio ”core”flux density estimated using funtools on circular regions ofapproximately the same size ( ∼ Method Variable Variable Result(1) (2) (3) (4)BHK FIRST Tot NIR Kendall’s τ = 0.7143Z value = 1.237prob = 0.2160FIRST Core NIR Kendall’s τ = 0.4000Z value = 0.490prob = 0.6242NIR FIRST Tot Kendall’s τ = 0.7143Z value = 1.237prob = 0.2160NIR FIRST Core Kendall’s τ = 0.4000Z value = 0.490prob = 0.6242Cox FIRST Tot NIR χ = 0.036prob = 0.8491FIRST Core NIR χ = 0.319prob = 0.5724Cox NIR FIRST Tot χ = 0.052prob = 0.8192NIR FIRST Core χ = 0.065prob = 0.79902 Filho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? Table 6. The Infrared-Radio Correlation.
Col. 1: The radio and NIR source(components N) corresponding to the SDSS maxBCG cluster field: identification num-ber is the sequential number (Koester et al. 2007). Col. 2: NIR magnitude and errorof the central components N, from GAIA se xtractor aperture photometry (except formaxBCG 8495). Magnitudes are in the Vega system. Col. 3: NIR flux and error ofthe central components N, from GAIA se xtractor aperture photometry (except formaxBCG 8495). Col. 4: FIRST radio ”core” flux density estimated using funtools on circular regions of approximately the same size ( ∼ maxBCG NIR NIR FIRST Core First Tot q core q tot (mag) (mJy) (mJy) (mJy)(1) (2) (3) (4) (5) (6) (7)2596 17.57 ± ± ± ± ± ± ± ± ± ± ± a ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a The maxBCG 8495 NIR magnitude and flux refer to GAIA manual aperture photometryof a region ( ∼ Table 7. Redshift and Redshift-Dependent Parameters.
Col. 1: The radiosource corresponding to the SDSS maxBCG cluster field: identification number isthe sequential number (Koester et al. 2007). Col. 2: Redshift estimate assumingMrk 668 as a template. Col. 3: Redshift estimate using the Le Phare photometricredshift code and Stripe 82 photometry in conjunction with our NIR data. Col. 4:The logarithm of the ”core” rest-frame monochromatic radio luminosity. The radio”core” power is estimated using funtools on circular regions of approximatelythe same size ( ∼ k -correction. Col. 5: The logarithm of the total FIRST rest-framemonochromatic radio luminosity. The radio power is estimated from sum of theintegrated flux densities of all the FIRST components. We have not included a k -correction. Col. 6: FIRST projected largest angular size. Col. 7: FIRST projectedlargest linear size. maxBCG z Mrk668 z Le Phare log P core , . log P tot , . LAS LLS(W Hz − ) (W Hz − ) (”) (Mpc)(1) (2) (3) (4) (5) (6) (7)2596 0.56 . . . 24.21 25.93 42 0.273131 1.03 1.57 +1 . − . +3 . − . . . . 26.63 36 0.308495 0.96 0.88 +0 . − . . . . 26.20 39 0.3110942 1.71 . . . 25.32 27.14 51 0.4411079 3.18 . . . 26.10 28.19 39 0.3011390 1.28 . . . . . . 26.00 24 0.2111780 0.05 . . . 21.37 23.12 36 0.04ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 13 maxBCG 2596 E W NE E a s t Lobe N u c l eu s W e s t Lobe maxBCG 3131
D1 D2LE1E2E3 S1S2 NE N o r t h C o m p S ou t h C o m p S ou t h C o m p2 Fig. 1.
From left to right, inverted greyscale image of the NIR emission, FIRST greyscale emission with superimposedradio contours, the first two levels of the FIRST contour plots superimposed on the NIR field and likewise superimposedon the SDSS and the Stripe 82 (maxBCG 3131) r AB -band field. The field numbers are the sequential numbers of SDSSmacxBCG cluster fields to which the radio sources belong, as they appear in Koester et al. (2007), letters/numbers arethe NIR components and lobe designations refer to the radio components. The inlays contain the NIR N component(Table 3 and 4), coincident with the center of the radio structure. Radio contours are: (top) 0.001, 0.002, 0.004, 0.008mJy; (bottom) 0.001, 0.002, 0.004, 0.008 mJy. maxBCG 6167 NE North LobeSouth Lobe
Fig. 2.
From left to right, inverted greyscale image of the NIR emission, FIRST greyscale emission with superimposedradio contours, the first two levels of the FIRST contour plots superimposed on the NIR field and likewise superimposedon the SDSS field and the Stripe 82 (maxBCG 6167 and 8495) r AB -band field. The field numbers are the sequentialnumbers of SDSS macxBCG cluster fields to which the radio sources belong as they appear in Koester et al. (2007),letters/numbers are the NIR components and lobe designations refer to the radio components. The inlays contain theNIR N component (Table 3 and 4), coincident with the center of the radio structure. Radio contours are: (top) 0.0004,0.0008, 0.0016 mJy (in log); (bottom) 0.0005, 0.0010, 0.0020, 0.0040, 0.0060, 0.0080 mJy. ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 15 maxBCG 10942 W NE North LobeSouth Lobe maxBCG 11079
S1S2S3L NE North LobeSouth Lobe
Fig. 3.
From left to right, inverted greyscale image of the NIR emission, FIRST greyscale emission with superimposedradio contours, the first two levels of the FIRST contour plots superimposed on the NIR field and likewise superimposedon the SDSS r AB -band field. The field numbers are the sequential numbers of SDSS macxBCG cluster fields to whichthe radio sources belong, as they appear in Koester et al. (2007), letters/numbers are the NIR components and lobedesignations refer to the radio components. The inlays contain the NIR N component (Table 3 and 4), coincident withthe center of the radio structure. Radio contours are: (top) 0.001, 0.002, 0.004, 0.008, 0.016 mJy; (bottom) 0.008, 0.0016,0.0032, 0.0064, 0.0128, 0.0256, 0.0512 mJy. maxBCG 11390 NE NE North LobeSouth Lobe maxBCG 11780 L NE North LobeSouth Lobe
Fig. 4.
From left to right, inverted greyscale image of the NIR emission, FIRST greyscale emission with superimposedradio contours, the first two levels of the FIRST contour plots superimposed on the NIR field and likewise superimposedon the SDSS r AB -band field. The field numbers are the sequential numbers of SDSS macxBCG cluster fields to whichthe radio sources belong, as they appear in Koester et al. (2007), letters/numbers are the NIR components and lobedesignations refer to the radio components. The inlays contain the NIR N component (Table 3 and 4), coincident withthe center of the radio structure. Radio contours are: (top) 0.0004, 0.0008, 0.0016, 0.0032 mJy (in log); (bottom) 0.0005,0.0010, 0.0020, 0.0040, 0.0080 mJy. ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 17 Fig. 5.
The NIR versus 1.4 GHz FIRST flux density. TheNIR flux pertains to the K s -band fluxes of the NIR com-ponents (components N; Table 3) coincident with the radiostructure centers. The NIR magnitudes are in the Vega sys-tem. The radio core flux density (top) pertains to the radio”core” flux density (Table 6) estimated using funtools oncircular regions of approximately the same size ( ∼ × rms ,where rms is the root mean square, a measure of the noiselevel in the FIRST image. Fig. 6.
The redshift distribution of the 3CRR (solid line)and GPS (dashed line) sample.
Fig. 7.
The redshift versus largest linear size (LLS) for the3CRR (crosses), GPS (triangles) and our sample (circles).The LLS for the GPS sources were obtained from O’Dea &Baum (1997), when a measurement was available. ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 19
Fig. 8.
Clockwise from the top left: The redshift versus the 5 GHz power for the 3CRR (crosses), GPS (triangles) andour sample total (closed circles) and core (open circles); the largest linear size (LLS) versus the 5 GHz power for the3CRR (crosses), GPS (triangles) and our sample total (closed circles) and core (open circles); the redshift versus the K -(GPS and 3CRR) or K s -band (our sample) magnitude for the 3CRR (crosses), GPS (triangles) and our sample (circles);the K - (GPS and 3CRR) or K s -band (our sample) magnitude versus the 5 GHz power for the 3CRR (crosses), GPS(triangles) and our sample total (closed circles) and core (open circles). The radio measurements for the 3CRR sampleare for the radio cores and for the GPS sample it is the total radio power. For our sample we include the radio ”core”power (open circles; Table 7) estimated using funtools on circular regions of approximately the same size ( ∼ K -band magnitudes are aperture- and emission-line-corrected measurements fora subset of 3CRR sources. For the GPS sample, the K -band magnitudes were obtained by interpolation from NED NIRphotometric points. NIR magnitudes for our sample are for the central NIR component, coincident with the center ofthe radio structure (components N; Table 3). The NIR magnitude system is Vega.
7. Acknowledgments
References
Abazajian K. N., Adelman-McCarthy J. K., Ag¨ueros M. A., AllamS. S., Allende Prieto C., et al. , 2009, ApJS, 182, 543Ant´on, S., Browne, I. W. A., March˜a, M. J. M., Bondi, M., & Polatidis,A. 2004, MNRAS, 352, 673Appleton, P. N. et al. 2004, ApJS, 154, 147Arnouts, S., Cristiani, S., Moscardini, L., Matarrese, S., Lucchin, F.,et al. 1999, MNRAS, 310, 540Baldi, R. D. et al. 2010, ApJ, 725, 2426Banfield, J. K., George, S. J., Taylor, A. R., Stil, J. M., Kothes, R.,Scott, D. 2011, arXiv:1103.4228Barnes, J. E. & Hernquist, L. 1996, ApJ, 471, 115Baum, S. A., O’Dea, C. P., Murphy, D. W. & de Bruyn, A. G. 1990,A&A, 232, 19Bondi, M., March˜a, M. J. M., Polatidis, A., Dallacasa, D.,Stanghellini, C., & Ant´on, S. 2004, MNRAS, 352, 112Clarke, D. A. & Burns, J. O. 1991, ApJ, 369, 308Clarke, D. A. et al. 1992, ApJ, 385, 173Fomalont, E. B., Kellermann, K. I., Partridge, R. B., Windhorst, R.A. & Richards, E. A. 2002, AJ, 123, 240Fanaroff, B. L. & Riley, J. M. 1974, MNRAS, 167, 31Franceschini, A., Vercellone, A. & Fabian, A. C. 1998, MNRAS, 297,817Fukugita M., Shimasaku K., Ichikawa T. 1995, PASP 107, 945 [Garn, T. & Alexander, P. 2008, MNRAS, 391, 1000 Ilbert, O., Arnouts, S., McCracken, H. J., Bolzonella, M., Bertin, E.,et al. 2006, A&A, 457, 841Haarsma, D. B., Partridge, R. B., Windhorst, R. A. & Richards, E.A. 2000, ApJ, 544, 641Higdon, J. L., Higdon, S. J. U., et al. 2005, ApJ, 626, 58Higdon, J. L., Higdon, S. J. U., Willner, S. P., Brown, M. J. I., Stern,D., Le Floc’h, E. & Eisenhardt, P. 2008, ApJ, 688, 885Hopkins, A. M., Afonso, J., Chan, B., Cram, L. E., Georgakakis, A.& Mobasher, B. 2003, AJ, 125, 465Hough, D. H. & Readhead, A. C. S. 1989, AJ, 98, 1208Huynh, M. T., Norris, R. P., Siana, B. & Middelberg, E. 2010, ApJ,710, 698Jarvis, J. M, Teimourian, H., Simpson, C., Smith, D. J. B., Rawlings,S. & Bonfield, D. 2009, MNRAS, 398, L83Koester, B. P. et al. 2007, ApJ, 660, 239Komissarov, S. S. & Gubanov, A. G. 1994, A&A, 285, 27Kronberg, P.P., Wielebinski, R. & Graham, D. A. 1986, A&A, 169,63Labiano, A., Barthel, P. D., O’Dea, C. P, de Vries, W. H., Perez, I. &Baum, S. A. 2007, A&A, 463, 97Larson, D. et al. 2011, ApJS, 192, article id. 16Leahy, J. P. et al. 1997, MNRAS, 291, 20Ledlow, M. J., Owen, F. N. 1996, AJ 112, 9Machalski, J., Jamrozy, M. & Zola, S. 2001, A&A, 371, 445Marecki, A., Barthel, P. D., Polatidis, A. & Owsianik, I. 2003, PASA,20, 16Middelberg, E. et al. 2008a, AJ, 135, 1276Natarajan, P. & Pringle, J. E. 1998, ApJ, 506, 97Norris, R. P. et al. 2006, AJ, 132, 2409Norris, R. P. et al. 2011, arXiv:1105.0960O’Dea, C. 1998, PASP, 110, 4930O’Dea, C. P. & Baum, S. A. 1997, AJ, 113, 148Owen, F. N. & Ledlow, M. J. 1997, ApJS, 108, 41Richards, E. A. 1999, ApJ, 513, 9Richards, G. T. et al. 2001, AJ 121, 2308Rigby, E. E., Snellen, I. A. G. & Best, P. N. 2007, MNRAS, 380, 1449Schoenmakers, A. P. et al. 2000a, MNRAS, 315, 371Schoenmakers, A. P. et al. 2000c, MNRAS, 315, 381Spinrad, H., Marr, J., Aguilar, L. & Djorgovski, S. 1985, PASP, 97,932Spoon, H. W. W., et al. 2007, ApJ, 654, 77Stanghellini, C., Baum, S. A., O’Dea, C. P. & Morris, G. B. 1990,A&A, 233, 379Stanghellini, C., O’Dea, C. P., Dallacasa, D., Cassaro, P., Baum, S.A., Fanti, R. & Fanti, C. 2005, A&A, 443, 891Sullivan, M., Hopkins, A. M., Afonso, J., Georgakakis, A., Chan, B.,Cram, L. E., Mobasher, B. & Almeida, C. 2004, ApJS, 155, 1White, R. L., Becker, R. H., Helfand, D. J. & Gregg, M. D. 1997,ApJ, 475, 479York, D. G. et al. 2000, AJ, 120, 1579Yun, Min S., Reddy, Naveen A. & Condon, J. J. 2001, ApJ, 554, 803Zinn, P. -C., Middelberg, E. & Ibar, E. 2011, arXiv:1104.0564 ilho, Brinchmann, Lobo & Anton: Optically Faint Radio Sources: Reborn AGN? 21 R a d i o / N I R r a ti o BCG2596 BCG3131 BC G BCG10942BCG11079 BCG11390BCG11780 BCG8495