Radio spectral study of the cluster of galaxies Abell 2255

21 Pages·2009·2.56 MB·English

A&A507,639–659(2009) Astronomy
DOI:10.1051/0004-6361/200912465 &
(cid:2)c ESO2009 Astrophysics
Radio spectral study of the cluster of galaxies Abell 2255
R.F.Pizzo1 andA.G.deBruyn1,2
1 KapteynInstitute,Postbus800,9700AVGroningen,TheNetherlands
e-mail:[email protected]
2 ASTRON,Postbus2,7990AADwingeloo,TheNetherlands
Received11May2009/Accepted31August2009
ABSTRACT
Context.Thestudyofthenonthermalcomponentsassociatedwiththeintraclustermedium(ICM)ingalaxyclustersisimportantin
understandingthehistoryandevolutionofclusters.
Aims.Spectralindexstudiesofhalos,relics,andradiogalaxiesprovideusefulinformationontheiroriginandconnectionwithmerger
processes.Moreover,theyrevealtheenergyspectrumoftherelativisticparticlesandthemagneticfielddistributioningalaxyclusters.
Methods.WepresentWSRTmulti-wavelengthobservationsofthegalaxyclusterAbell2255at25cm,85cm,and2m.Thespectral
indeximagesallowedustostudytheintegratedspectrumofhaloandrelicandtoinvestigatethephysicalpropertiesoftheBeaver
head-tailradiogalaxybelongingtothecluster.
Results.Intheradiohalo,thespectralindexissteeperatthecenterandflatteratthelocationsoftheradiofilaments,clearlydetected
at25cm.Intherelics,thespectralindexflattens,movingawayfromtheclustercenter.FortheBeaverradiogalaxy,thespectrum
severelysteepensfromtheheadtowardstheendofthetail,becauseoftheenergylossessufferedbytherelativisticparticles.Inthe
2mmap, which isthefirsthigh-sensitivity imagepresented intheliteratureat such along wavelength, anew Mpc-size emission
regionisdetectedbetweentheknownradiohaloandtheNWrelic.Notdetectingthisfeatureinthemoresensitive85cmobservations
impliesthatitmusthaveaverysteepspectrum(α≤−2.6).
Conclusions.Theobservationalpropertiesoftheradiohalosuggestthateitherwearelookingatasuperpositionofdifferentstructures
(filamentsintheforegroundplusrealhalointhebackground)seeninprojectionacrosstheclustercenterorthatthehaloisintrinsically
peculiar.ThenewlydetectedextendedregiontotheNWofthehalocouldbeconsideredasanasymmetricextensionofthehaloitself.
However,sinceradiohalosareknownintheliteratureasstructuresshowingaregularmorphology,thenewfeaturecouldrepresent
thefirstexampleofsteepMpc-sizediffusestructures(MDS),detectedaroundclustersatverylowfrequencies.Thespectrumofthe
initialpartofthetailoftheBeaver,detectedatthethreewavelengths,iswell-fittedbyasingleinjectionmodel.
Keywords.galaxies:clusters:general–galaxies:clusters:individual:Abell2255–galaxies:intergalacticmedium
1. Introduction opticalandX-raydomains,theseextendedfeaturesaredirectly
connected to the cluster history and evolution. Moreover, they
Galaxy clusters, which contains up to a few thousand galaxies
provideasignificanttestforseveraltheoriesconcerningtheori-
andconsiderableamountsofgas,arethemostmassivegravita-
gin of relativistic particlesin the ICM andparticle propagation
tionallyboundobjectsintheuniverse.Cosmologicalsimulations
inastrophysicalplasmas.
(Navarro et al. 1995) show that they are not static, but instead
grow and still form at the present epoch, as the result of sev- The origin of halos and relics is still matter of debate. It is
eralmergerevents(e.g.Evrard&Gioia2002). suggestedthatrelicsrelicsarerelatedtoshockseitherbyFermi-I
diffuseaccelerationofICMelectrons(Enßlinetal.1998;Keshet
Inanincreasingnumberofmassive,merging,andX-raylu-
minous galaxy clusters, large diffuse radio sources associated etal.2004b)orbyadiabaticenergizationoftherelativisticelec-
trons confined in fossil radio plasma (“ghosts”), released by a
withtheICMhavebeendetected.Theyrepresentthemostspec-
former active radio galaxy (Enßlin & Gopal-Krishna 2001). In
tacular aspectofcluster radioemission, andtheycannotbe as-
sociated with any individualgalaxy. These diffuse synchrotron the case of radio halos, it is required that the electrons are ei-
therre-accelerated(primarymodels:e.g.Tribble1993;Brunetti
radiosourcesarecharacterizedbyatypicalsizeofabout1Mpc,
lowsurfacebrightness(∼1μJyarcsec−2at20cm),andsteepra- et al. 2001; Petrosian 2001; Brunetti et al. 2004; Fujita et al.
dio spectrum(α ≤ −1, S(ν) ∝ να). Theyare classified (Feretti 2003; Cassano & Brunetti 2005) or continuouslyinjected over
theentireclustervolumebyhadroniccollisions(secondarymod-
& Giovannini 1996) as radio halos, if located at the center of
els:e.g.Dennison1980;Blasi&Colafrancesco1999;Dolag&
the cluster and not significantly polarized, and radio relics, if
Enßlin2000).
lying at the cluster periphery and showing high polarization
percentages (∼10%−30%). The synchrotron nature of their ra- Spectral index studies of halos and relics provide impor-
dio emission indicates cluster-wide magnetic fields of approx- tant information on their formation and their connection to
imately 0.1−1 μG, and of a population of relativistic electrons cluster merger processes. Given their large angular size and
withLorentzfactorγ(cid:6)1000. steep spectra, low-frequency radio observations are suitable
Theknowledgeofthephysicalconditionsinhalosandrelics for such investigations. There are only a few clusters for
is important for a comprehensive physical description of the whichspectralindexmapsofhaloshavebeenpublished:Coma
ICM. Since they are related to other cluster properties in the (Giovannini et al. 1993), A2163, A665 (Feretti et al. 2004),
ArticlepublishedbyEDPSciences
640 R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255
A3562(Giacintuccietal.2005),A2744,andA2219(Orrúetal. Table1.Thefluxcalibratorsusedduringthedatareduction.
2007). In agreement with primary models, in all these objects
the spectral index maps show a patchy structure with, in some Wavelength Name Flux α
cases, a steepening of the spectrum with increasing distance
(Jy)
from the center to the edge. The detailed spectral index distri- 25cm CTD93 5.2 –0.5
butionisknownfortherelicsofA3667(Rottgeringetal.1997), 85cm 3C295 63 –0.6
theComacluster(Giovanninietal.1991),S753(Subrahmanyan 2m 3C295 95 –0.6
etal.2003),A2744andA2219(Orrúetal.2007),andA2345and
A1240(Bonafedeetal.2009).Apartfromoneofthetworelics
ofA2345,inalltheotherexamplesthespectralindexflattensgo-
ingfromtheregionsclosertotheclustercentertotheouterrim. 2. Observationsanddatareduction
This is consistent with the presence of electron re-acceleration
TheobservationswerecarriedoutwiththeWSRTinthreebroad
inanexpandingmergershock.
frequency ranges with central wavelengths of about 25 cm,
Other important members of clusters in the radio domain
85 cm, and 2 m. The array consists of fourteen 25 m dishes
aretheradiogalaxies.Theiremissionoftenextendswellbeyond
onaneast-westbaselineandusesearthrotationtofullysynthe-
the opticalboundariesof the radiogalaxies,outto hundredsof
sizetheuv-plane.Tenofthetelescopesareonfixedmountings,
kilo-parsec,andhenceitis expectedthattheICM wouldaffect
144 m apart; the four remaining dishes are movable along two
theirstructure.Thisinteractionisprovedbytheexistenceofra- railtracks.Theinterferometercanobserveindifferentconfigu-
diogalaxiesshowingdistortedstructures(tailedradiosources). rations that differ from each other in the distance between the
Spectralstudiesofradiogalaxieswithdifferentmorphologiesat lastfixedtelescopeandthefirstmovableone(RT9−RTA).Inthe
differentfrequenciesprovideanexcellentopportunitytotestthe
array,thebaselinescanextendfrom36mto2.7km.Atthehigh
ageingoftheelectrons,undersomestandardassumptions. declinationofA2255thearraydoesnotsufferfromshadowing.
A2255is a nearby(z = 0.0806,Struble & Rood 1999)and
The pointing center of the telescope, as well as the phase
rich cluster, which has been studied in several bands. ROSAT
center of the array, was directed towards RA = 17h13m00s,
X-rayobservationsindicatethatA2255hasrecentlyundergone Dec = +64◦07(cid:7)59(cid:7)(cid:7), which is the center of A2255. The time
a merger (Burns et al. 1995; Feretti et al. 1997; Davis et al.
samplingis 30sin the 25cm and85cm datasets,and10 sfor
2003).RecentXMM-Newtonobservationsrevealedtemperature
the 2 m observations.Thisis generallysufficientto sample the
asymmetries of the ICM suggesting that the merger happened
phase fluctuationsof the ionosphereand to avoid the smearing
∼0.15 Gyr ago, probablyalong the E −W direction(Sakelliou
of sources at the outer edge of the field. During each observa-
& Ponman 2006). Optical studies of A2255 revealed the pres-
tion,twopairsofcalibrators,onepolarizedandoneunpolarized,
enceofkinematicsubstructuresintheformofseveralassociated
have been observed for 30 min each. Table 2 summarizes the
groups(Yuanetal.2003).Thisresult,togetherwiththelargera-
observationalparameters.
tioofvelocitydispersiontoX-raytemperature(6.3keV;Horner
The technical details about the datasets will be discussed
2001)indicatesanon-relaxedsystem.
in the next sub sections. Here we give a short overview about
Whenstudiedatradiowavelengths,A2255showsthepres-
ence of a diffuse radio halo (located at the center of the clus- the mainstepstakenforthedatareduction.Thedatawere pro-
cessed with the WSRT-tailored NEWSTAR reduction package
ter) and of a relic (at the cluster periphery), together with a
following mostly standard procedures: automatic interference
largenumberofembeddedhead-tailradiogalaxies(Harrisetal.
flagging, self calibration, fast Fourier transform imaging, and
1980).High resolution radio observationsat 21 cm shown that
CLEANdeconvolution(Högbom1974).Furtherflaggingbased
the radio halo has a rectangular shape and a surface bright-
on the residual data after self calibration and model subtrac-
nessincreasingfromthecentertotheedge(Ferettietal.1997).
At this location 3 highly polarized (∼20%−40%) bright fila- tion was done after each self calibration iteration. An on-line
Hammingtaperwasusedtolowerthedistantspectralsidelobe
ments perpendicular to each other are detected (Govoni et al.
level(Harris1978).Thefinalanalysiswas doneusing onlythe
2005).Sensitivelow-frequencyobservationscarriedoutwiththe
oddchannels,becauseeachHammingtaperedoutputchannelis
Westerbork synthesisradiotelescope (WSRT) at 85 cm proved
alinearcombinationofitsdirectneighborsanditself.
the presence of a large number of low-surface brightness ex-
At 85 cm and 2 m, the data were flux-calibrated using
tendedfeaturesaroundthecluster,thatarelikelyrelatedtoLarge
3C295,forwhichweadoptedafluxof63Jyand95Jyatcentral
ScaleStructure(LSS)shocks(Pizzoetal.2008).Inthisworkwe
frequenciesof 346 MHz and 148 MHz, respectively.At 25 cm
combinethesedatawithobservationsat25cmandat2minorder
the flux scales have been set using the flux calibrator CTD93,
tostudythespectralindexpropertiesofthestructuresbelonging
for which we adopted a flux of 5.2 Jy at 1220 MHz. The flux
toA2255.
scale at 2 m is still uncertainto a levelthatwe estimate of 5%
This paper is organizedas follows. Section 1 describes the
and1%at25cmand85cm.Wecomputedthefluxdensitiesof
main steps of the data reductionand discusses a few issues re-
thecalibratorsacrosstheentirebandsassumingaspectralindex
latedtolow-frequencyradioobservations.InSect.3,wepresent
ofα = −0.5andα = −0.6forCTD93and3C295respectively
thefinalmapsofA2255atthethreewavelengthsandinSect.4
(seeTable1).
weshowthespectralindexanalysisforthehaloandtherelics.In
Sect. 5, we analyze the physicalpropertiesof the Beaver radio
galaxy and we test the ageing processes of the radiating elec-
2.1.25cmdataset
tronsalongthetail.Wediscusstheimplicationsofourresultsin
Sect.6andsummarizeourworkinSect.7. At25cm,weobservedA2255for4×12hwith4differentcon-
ThroughoutthispaperweassumeaΛcosmologywithH = figurationsoftheWSRT.Bysteppingthe4movabletelescopes
0
71 kms−1Mpc−1, Ωm = 0.3, and ΩΛ = 0.7. All positions are at 18 m increments, from 36 m to 90 m, we pushed the grat-
given in J2000 coordinates. The resolutions are expressed in inglobedueto theregular18mbaselineincrementtoa radius
RA×Dec.AtthedistanceofA2255,1(cid:7)correspondsto90kpc. of∼1◦.
R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255 641
Table2.Observationsoverview.
Wavelength Frequencyrange Calibrators ObservationID RT9–RTA Startdate(UTC) Enddate(UTC)
(MHz) (m)
10702691 36 2007/06/1817:00:15 2007/06/1904:58:50
3C286,CTD93 10702898 54 2007/06/2816:20:55 2007/06/2904:19:25
25cm 1159–1298MHz 3C147,3C138 10702958 72 2007/07/0316:01:15 2007/07/0403:59:45
10703066 90 2007/07/1015:33:45 2007/07/1103:10:35
10602227 36 2006/05/0919:36:20 2006/05/1007:35:20
10602239 48 2006/05/1019:32:20 2006/05/1007:31:20
DA240,3C295, 10602259 60 2006/05/1119:28:20 2006/05/1207:27:20
85cm 310–380MHz PSR1937+21, 10602224 72 2006/05/1619:08:40 2006/05/1707:07:40
3C48 10602372 84 2006/05/1918:57:00 2006/05/2006:56:00
10602340 96 2006/05/2218:45:10 2006/05/2306:44:10
10702727 36 2007/06/2116:48:15 2007/06/2204:47:05
10703226 48 2007/07/1914:58:05 2007/07/2002:56:55
DA240,3C295, 10703186 60 2007/07/1715:05:55 2007/07/1803:04:45
2m 115–175MHz PSR1937+21, 10703004 72 2007/07/0515:53:15 2007/07/0603:52:05
3C48 10703056 84 2007/07/0915:37:25 2007/07/1003:36:15
10703099 96 2007/07/1215:25:35 2007/07/1303:24:25
Atthiswavelength,thereceivingbandisdividedinto8con- disturbances in low-frequency observations (e.g. Cotton et al.
tiguous slightly overlapping sub-bands of 20 MHz centered at 2004;Intema2007),butitcanbealsoappliedathigherfrequen-
1169,1186,1203,1220,1237,1254,1271,and1288MHz.Each cies, when phase errors due to the differential tropospheric ef-
sub-bandiscoveredby64channelsin4cross-correlationstore- fectsarealsopresentinhighdynamicrangeimages.
coverallStokesparameters.
The peeling scheme involves self calibration on individual
Becauseoflimitationsofthesoftwareinhandlingfileslarger
brightsources.It producesphase correctionsper arrayelement
than 2.15 GB, the data reduction was done for each frequency
for several viewing directions, repeatedly using the following
bandindependently.Thus,theoriginaldatasethasbeensplitin
recipe:
8 sub-datasets,eachcontainingthe4×12h ofdata,butjustin
onefrequencyband.
BecauseofstrongRFI,2outoftheoriginal8frequencysub- – subtractionofallbutthebrightestsourcefromUVdata,us-
bandshave notbeen used for imaging.Moreover,the presence ingthebest modelandcalibrationavailable;
ofastrongoff-axissourcecausedphaseerrorsaffectingthecen- – severalroundsofphaseonlyselfcalibrationandimagingon
tralfieldofthefinalimage.Thepeelingprocedurethatweused thebrightestsource;
to solve for those problems was successful only in the lower – subtractionofthebrightestsourcefromtheoriginalUVdata,
3frequencybands,wheretheproblematicsourceisbrighter(see usingmodelandcalibrationfromthepreviousstep.
Sect.2.2).Therefore,thetotalamountofdatathatwecoulduse
forimagingwas∼35%.At25cm,theresolutionis14(cid:7)(cid:7)×15(cid:7)(cid:7).
Weappliedthisalgorithmtothe25cmdataset,whichistheonly
oneaffectedbyoffaxiserrorsthataffecttheanalysisatthecen-
2.2.Off-axiserrorsin25cmmaps
ter of the maps. Also our maps at 85 cm and at 2 m showed
off axis errors, but they were confined in areas far away from
High dynamic range imaging is seriously limited by the phase
the field center. Therefore no peeling procedurewas needed at
stabilityoftheatmosphere(troposphereandionosphere),which
these wavelengths. At 25 cm the phase errors were associated
causes the presence of spiky patternssurroundingthe brightest
off-axis sources within the field of view of the interferometer. withtheoffaxissource4C+64.21,locatedatRA=17h19m59s,
This pattern is due to the instantaneous fan beam response of Dec = +64◦04(cid:7)40(cid:7)(cid:7). In Fig. 1, we show the resultof the proce-
the WSRT, which rotates clockwise from position angle +90◦ dure.Withoutpeeling,4C+64.21showsadistortedshape,with
to+270◦duringthe12hsynthesistime. a positive spike going towards east and a long negative spike
directedtowardstheoppositedirection,wherethecenterofthe
The selfcal algorithm tries to minimize the time variations
cluster is located. The source is also surroundedby ring struc-
between data and the model of the sky, applying only one av-
eragedtime-dependentcorrectionintheuvplane.Ifinthefield tures,whichwereleftafterthedeconvolution.Afterpeeling,the
ofviewtherearebrightoff-axissourcesthatseemtomovedur- source is basically removed.The flux below which the sources
ingtheobservation,duetodifferentialtroposphericrefractionor havebeensubtractedforthisprocedureis∼1Jy.
rapidionosphericphaseinstabilities,theselfcalwillnotbeable Thepeelingprocedurehasbeenappliedineachofthe6fre-
to correct simultaneously for the phase errors associated with quencybandswecoulduseat25cm(Sect.2.1).Unfortunately,
them.Thus,thefinalmapwillshowartefactsaroundthesources, it turned out to be satisfactory just for the three lower fre-
assecondarylobes,orinthesourceitself,whichwillappeardis- quencybands, where the selfcal correctionsfor the steep spec-
torted. The only way to removethe off-axis errorsin the maps trum4C+64.21canbebetterdetermined,giventhehighersig-
is to “peel” each problematic source out of the dataset, using nal to noise ratio of the data. Thus, we decided to use only
its own direction-dependentselfcal corrections. This technique these3frequencybandstomakethefinalimageforthefurther
has been recentlydevelopedto solve for the ionosphericphase analysis.
642 R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255
Fig.1. Grey scale images of 4C +64.21 at 25 cm before (left panel) and after (right panel) applying the peeling procedure. The resolution is
14(cid:7)(cid:7)×15(cid:7)(cid:7).Leftpanel:thesourceshowsphase-relatederrors,likeringstructuresandanegativeandapositivespikealongtheeast-westdirection.
Rightpanel:justasmallresidualisleftatthesourcelocation.
2.3.85cmdataset selfcalibrated.Attheend,theimageswerecombined,weighting
themfortheinputnumberofvisibilities.
Tofullyimagetheprimarybeam,sixarrayconfigurationswere Due to serious problems with the correlators of the WSRT
used for the observations, with the four movable telescopes during the nightbetween the 17th July and the 18th July 2007
steppedat12mincrements(i.e.halfthedishdiameter)andthe (RT9−RTA = 60 m), the final data reduction was done using
shortestspacingrunningfrom36mto96m.Thisprovidescon- 5 × 12 h of observation. Moreover, given serious ionospheric
tinuousuv-coveragewithinterferometerbaselinesrangingfrom instabilities during the last night of observations(RT9−RTA =
36mto2760m. 96 m),we decidedto notincludethis12hoursrunin imaging.
The receiving band is covered by 8 sub-bands of 10 MHz Thetotalamountofflaggeddatawas∼35%.At2m,theresolu-
centered at 315, 324, 332, 341, 350, 359, 367, and 376 MHz. tionis163(cid:7)(cid:7)×181(cid:7)(cid:7).
Eachsub-bandiscoveredby128channelsin4cross-correlations
torecoverallStokesparameters.The85cmdatasetwassplitin
16sub-datasets,eachcontaining6×12hofobservation,butin 2.5.Afewissuesrelatedtolow-frequencyobservations
onefrequencyhalf-bandonly. The low-frequency radio sky is very bright and populated by
The imaging was done using 5 out of the 6 nights of ob- strong radio sources, such as Cas A, Cyg A, Vir A and Tau A,
servation,becauseofthepoorphasestabilityofthe ionosphere whose flux densitiesvaryfrom 1000−10000 Jy. Evenfar from
duringthethirdnight(RT9–RTA= 60m).Thetotalamountof thefieldcenter,the(relatively)highdistantsidelobelevelsofthe
flaggeddatawas∼25%.At85cmwavelength,theresolutionis primarybeam(“only”−30to−40dB)keepsthesesourcesvery
54(cid:7)(cid:7)×64(cid:7)(cid:7). bright, giving rise to significant side lobes in the final images.
Solarflarescanalsoaffectthedata,causinginterferencesinthe
shortbaselinesand/orgratingringsintheimagedcentralfield.
2.4.2mdataset(LFFE-bandobservations) Duringtheeditingandimagingprocessesatlowfrequency,
we had to take care of Cas A, Cyg A and the Sun. Cas A lies
At 2 m, A2255 was observed for 6 × 12 h to fully image the ataprojecteddistanceof∼40◦ fromthecluster,CygAat∼30◦
primarybeam. and the Sun at ∼85◦. At 85 cm and at 2 m, we used modelsof
TheLFFE(Low-FrequencyFrontEnd)attheWSRTiscov- CasAandCygAandwesubtractedthemfromthedataduring
eredby8frequencysub-bandsof2.5MHzcenteredat116,121, imaging. To remove the interferences generated by the Sun in
129,139,141,146,156,and162MHz.Eachsub-bandisdivided the 85 cm dataset, we flagged the short baselines in the hour
in512channels. anglesinwhichtheSunwasabovethehorizon(+30◦ ≤ HA ≤
Given the large size of the final dataset (∼250GB) andthe +90◦).At2m,theSunwashighintheskyformostpartofthe
limitationsinthemaximumsizeofaNEWSTARdataset,theob- observation.Inthiscase,weproduceddifferentcleaningmodels
servations have been split in 240 sub-datasets, each containing of this source for the different nights and we subtracted them
dataforasinglenightofobservationandwithalimitednumber duringthefinalimaging.
of channels. Therefore, for a single spacing and for each sub-
band,weworkedwith5datasets.Toproducethefinalmap,we
2.6.Fluxscaleat2mwavelength
processedthe240sub datasetsseparately,self calibratingthem
with differentmodels and doing the imaging and the deconvo- The flux scale at low frequencies is not very accurately deter-
lutionsteps240times.Wedidnotobtainagoodimagethrough mined. Until a more definitive flux scale is in place (this is
thenormalselfcalprocedure.Thebestimagehasbeenobtained being developed for LOFAR) we use the radio source 3C295
usingasimplifiedselfcalibrationwhereweonlysolvedforthe as the primary WSRT flux calibrator at low frequencies. We
phase slope acrossthearray.Inthis case, onlythe phaseswere have adopted a flux density at 150 MHz of 95 Jy for 3C295
R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255 643
Table3.Parametersoftheobservationsandofthefinalfullresolutionmaps.
Wavelength Fullresolution Largestdetectablestructure σ a σ b σ c σ d
(cid:7)(cid:7) (cid:7) F mJy/tbheam mJcyon/fbceentaerm mJoybs/cbenetearm
2m 163×181 191 5% 0.85 3 3.1
85cm 54×64 81 1% 0.05 0.25 0.25
25cm 14×15 24 1% 0.010 0.008 0.012
aFluxscaleuncertainty.
bThermalnoise,constantoverthemaps.
cConfusionnoise.
dObservednoiseatthepointingposition.
and a power-law spectral index of −0.6 in the frequencyrange observeasource,thelatter isalwaysthesamefora givenPSF,
from 115−175MHz. We believe this numberto be accurate to i.e.itrepeatsitselfeverytimeweobservethesamefield.
about5%. In Table 3, we list the estimated σ andσ for the three
th conf
A majorcomponentofthe system noise (receiver+ sky)at observingwavelengths.Reliable estimates ofthe thermalnoise
low frequencies is due to our Galaxy. The System Equivalent havebeenobtainedfrompolarizationimages(Q,U,V)andnar-
Flux Density (SEFD) of the telescopes at 150 MHz is about rowspectralbands.TheconfusionlimitforWSRTat21cmwas
8000 Jy in the Galactic areas where 3C295 happens to be lo- determined by Garrett et al. (2000) to be ∼5 μJy. Because the
cated, but it rises to well over 10000 Jy in the Galactic plane. PSF of the WSRT dependson the declination,this valuerefers
Because the WSRT receivers operate with an automatic gain toDec=+6◦.Atadifferentwavelength(λ )itisgivenby
2
control (AGC) system before the analog-to-digitalconverter,it (cid:2) (cid:3)
λ 2.75
cthoentvianruioabulselyinmpueatsleuvreeslst.hUentfootratlupnoawteelyr,tmooasltloowftchoerrtiemcteiotnhseftoor- σλ2 =σλ1 λ2 , (1)
1
tal powersdetectors(whichintegratethepoweroverthe whole
2.5MHzsub-band)arecorruptedbyRFIsowecannotautomat- wnehnetre(−σ2λ.715is) ttahkeecsoinnftuosiaocncoliumntitthateλs1pe=ctr2a1l icnmdeaxnddetpheendexenpcoe-
icallycorrectthecorrelationcoefficientsforthevariationsinsys-
(−0.75)andthedifferentsynthesizedbeam(−2).
temnoise.TheLFFEbandisfullofmostlyimpulsiveandnarrow
Assuming thatσ and σ are uncorrelated,the observed
bandRFIcomingfromairplanes,satellites,andmobileusersas th conf
noiseinthefinalmapasafunctionofthedistancefromthecen-
wellaselectronichardwarewithinthebuildingwhichislocated
terofthefieldisgivenby:
halfwaythearray.Atthehighspectralresolution(10kHz)pro-
(cid:4)
videdbythebackendmostofthisRFIcanbeexcised.However,
σ (r)= σ2 +σ2 (r). (2)
the total power data must be manually inspected for suitable obs th conf
stretchesofpowerlevelmeasurements.Thesedataformtheba-
InFig.2,weplottheobservednoiselevels,uncorrectedforthe
sisforamanualcorrectionofthefluxscale.
primarybeam,asafunctionofthedistancefromthefieldcenter
For A2255 the total power ratio between the cluster region
forthefinal2m,85cm,and25cmfullresolutionmaps.Because
and the 3C295field is 1.15±0.05at 141 MHz. Followingthe
the classical confusionnoise will be attenuated by the primary
transfer of the complex gain correction determined for 3C295
beam,theobservednoisedecreasesfromthecentertotheedges
we have therefore applied an additional correction of a factor
of the field, where it approaches the thermal noise level. This
1.15tothevisibilitydata.
trend,moreclearinthelow-frequencymaps,isaproblemwhen
tryingtoassessthesignificanceofthecentralextendedfeatures
in the contour maps. In principle, we should consider as de-
2.7.Errorsandnoiseinthefinalimages
tection limit the noise observed at the center of the field and
Knowingthevalueoftheerrorinanimageisessentialindeter- plotthecontoursstartingfrom,forexample,3timesthisvalue.
miningthereliabilityoftheimageandofthederivedparameters. However, it is worth noting that in this area we are confusion
Theobservederrors(σ )inamapmainlyconsistof: limited (this is a situation similar to that found in deep optical
obs
imagesofgalaxies,wherethecentralfluctuationsareoftendom-
– thethermalnoise(σth),whichisduetothestochasticerrors, inatedbythelargenumberofstarswithinaresolutionelement).
comingfromtheGalaxy,receiversandotherelectronicsused Therefore,inordertoshowthesignificanceofthefeaturesinour
inthearray; final contourmaps, we startplotting the contoursfrom 3 times
– theconfusionnoise(σconf),whichactuallyconsistsofthree the noiseobservedat the edgeof the fully imagedfield. To as-
contributions: sess the significance of the detected structures with respect to
1. the “normal”sidelobenoise,whicharisesfromthesum the background fluctuations, we advice the reader to compare
of all the sidelobes responses to the very large number thecontourmapsat25cm,85cm,and2mwiththeirgreyscale
ofsourcesvisiblewithinthefieldofview.Thisnoisecan versioninFig.3.
beminimizedbydeconvolution;
2. the “classical” source confusion, which is associated
withthenumberofsourceswithinthesamebeam; 2.8.Fluxmeasurementuncertainties
3. the“error”sidelobenoise,whichresultsfromcalibration
Because of the errors present in the final images, the flux and
errorsandnon-isoplanaticity. spectralindexestimatesareaffectedbyuncertainties.Theerror
associatedwiththefluxdependson:
Itisimportanttoappreciatethedifferencebetweenthermaland
confusion noise. While the former is different every time we – theobservednoiseinthefinalmaps(σ );
obs
644 R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255
Fig.2.Observednoiseleveluncorrectedfortheprimarybeamasafunctionofdistancefromthefieldcenterforthe2m(leftpanel),85cm(middle
panel),and25cm(rightpanel)maps.Thedashedlinerepresentstheestimatedthermalnoiselevel,againuncorrectedfortheprimarybeam.
– the error due to the negativebowl, which arises aroundthe 3.1.25cmmap
extended structures because of the missing short spacings
(σ ). This severely affects the non full resolution maps The 25 cm map is presented in Fig. 4. The image, which has
bowl
only(see Sect.2.9).Inourcase,itsvaluewasestimatedby a noise level of 11 μJy at the edge of the field, shows the well
determiningthemeanvalueofthenegativebowlaroundthe knownextendedhalo,locatedattheclustercenter,therelic,that
centralmaximumintheantennapattern; lies 10(cid:7) to the northeast from it, and 3 extended radio galax-
– the uncertainty related with the flux of the calibrator (σ ). ies at a very large distance from the center of the cluster: the
c
Thisisascaleerrorwhichapplytothewholemap. Embryo and the Beaver lie at ∼1.6 Mpc from the cluster cen-
ter,whiletheBeanliesatmorethan3.5Mpc(quotednamesare
Undertheassumptionthatthesethreeuncertaintiesareuncorre-
takenfromHarrisetal.1980).Sincemorerelicfeaturesarede-
lated,thefinalerroronthefluxis
(cid:4) tectedatlowfrequencyaroundA2255(Sect.3.2),fromnowon
wewillrefertotherelicasNE(north-east)relic.Zoominginto
σ = σ2+σ2 +σ2 . (3)
F c obs bowl the central region of A2255,we notice the presence of 4 addi-
Theuncertaintyonthespectralindexisgivenby tionalextendedclusterradiogalaxies:theGoldfish,theDouble,
(cid:5) the original TRG, and the Sidekick. The positions of the 7 ra-
(cid:2) (cid:3) (cid:2) (cid:3)
1 σ 2 σ 2 dio galaxies are listed in Table 4. Each of them has an optical
σα = ln(ν /ν ) FF1 + FF2 , (4) counterpartthatbelongstothecluster(Miller&Owen2003).
2 1 1 2
The halo has a rectangular shape and shows a filamentary
whereF ±σ andF ±σ arethefluxdensitiesandthecor-
1 F1 2 F2 structure,whichisinagreementwithprevious21cmVLAob-
respondingerrorsatfrequencyν andν ,respectively.
1 2 servations (Govoni et al. 2005). Given the high sensitivity of
our new observations, halo and NE relic look more extended
2.9.Makingmapsat25cm,85cm,and2mforaspectral anddirectlyconnectedbyaradiobridge,whichextendstowards
indexanalysis the north-eastand seems associated with two features; the for-
mer is located at RA = 17h13m13s, Dec = +64◦12(cid:7)54(cid:7)(cid:7) and
Tomakethefinalspectralindexmaps,wetaperedthedatatothe belongs to the NE relic, the latter lies at RA = 17h14m18s,
samelowresolutionofthe2mdataset(163(cid:7)(cid:7)×181(cid:7)(cid:7)),andwecut Dec = +64◦16(cid:7)10(cid:7)(cid:7). They are labeled respectively C1 and C2
theuvdatawiththesameminimumuvcoverage(144λ),which in Fig. 4. The physicalparametersof halo and NE relic are re-
isdeterminedbythe25cmdata. portedinTable5.
The missing short spacings in the low resolution 2 m and
We detect a very low surface brightness feature at location
85 cm maps created a negative bowl around the central radio RA = 17h12m12s, Dec = +64◦30(cid:7)08(cid:7)(cid:7), that is associated with
halo. To remove as much as possible of the negative bowl, we
one of the two filaments of the NW relic detected at 85 cm
cleanedverydeepthehaloemission.At2m,thefinalmapwas
(Sect.3.2).Sinceitliesinaprettyemptyregionoftheradiosky,
obtained including in imaging only the three frequency bands
a possible association of this feature with a collection of point
leastaffectedbyRFI.Finally,themapshavebeencorrectedfor
sourcesorwithacentralradiogalaxyseemsunlikely.
the total power primary beam of the WSRT, which can be ap-
proximatedby
G(ν,r)=cos6(cνr), (5) 3.2.85cmmap
where c is a constant ∼0.064, ν is the observing frequency The85cmmapofA2255ispresentedinFig.6.Withnoiselevels
inMHzandristheradiusfromthepointingcenterindegrees. ranging between 0.08 mJy/beam to 0.25 mJy/beam,limited by
classicalconfusionnoise(seeSect.2.7andTable3),itimproves
over previousimaging at close wavelength(Feretti et al. 1997)
3. Results
byafactorof20.Theoverlayoftheradiomapwiththeredband
A greyscale versionof thefinalfullresolutionmapsat25 cm, DigitizedSkySurvey(DSS)opticalimageispresentedinFig.5.
85cmand2misshowninFig.3.Inthefollowingsubsections, This image clearly shows the extension of the radio emission
wedescribethefeaturesdetectedateachwavelength. comparedtotheopticalgalaxies.
R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255 645
Fig.3.GreyscaleimagesofA2255at25cm(topleftpanel),85cm(toprightpanel),and2m(bottompanel).Allimagescoverthesameareaof
thesky.Theresolutionsare14(cid:7)(cid:7)×15(cid:7)(cid:7)(25cm),54(cid:7)(cid:7)×64(cid:7)(cid:7)(at85cm)and163(cid:7)(cid:7)×181(cid:7)(cid:7)(at2m).Themapsarenotcorrectedfortheprimarybeam.
The halo, the NE relic, and the radio galaxies belongingto NW relichasa morecomplexmorphology.We candistinguish
theclusteraredetected.Thecentralradiohalolooksmuchmore 2 filaments, labeled NW1 and NW2 in Fig. 6. NW1 points to-
complexthaninpreviousimagesatthesamefrequency(Feretti wardstheclustercenterandhasalengthof∼8(cid:7)×1(cid:7),whileNW2
etal.1997)anditismoreextendedthanat25cm,inparticular is∼13(cid:7)×1(cid:7)andisperpendiculartoNW1.TheSWandNWrelics
towardstheSandSW.Wenoticethatthesouthernregionofthe haveintegratedfluxdensitiesof∼17mJyand∼61mJy,respec-
haloisdirectlyconnected,inprojection,tothetailoftheBeaver tively. The physicalpropertiesof the newly detected structures
radiogalaxy,whichhasdoubleditslengthtoalmost1Mpcbe- and their origin have been investigated by Pizzo et al. (2008),
tween25cmand85cm.FeatureC2inthe25cmmap(seeFig.4) whosuggestedaconnectionwithLSSshocks.
isnowmoreprominentanditlooksdirectlyassociatedtotheNE Other extended diffuse low surface brightness features are
relic.ThephysicalparametersofhaloandNErelicat85cmare detectedtotheeast(RA = 17h16m57s,Dec = +64◦18(cid:7)45(cid:7)(cid:7))and
listedinTable5. tothewest(RA=17h07m08s,Dec=+63◦59(cid:7)36(cid:7)(cid:7))ofthecluster
center. Their nature is still unclear. Moreover, the map shows
The high sensitivity of our observations allow us to detect
positiveandnegativefluctuationsona scale of0.5to 1degree,
twonewextendedfeaturesataprojecteddistanceof2Mpcfrom
whicharelikelyduetoourGalaxy.
theclustercenter.Thenew“relics”arelocatedNWandSWof
thecenteroftheclusterandprevious21cmimagesofA2255re-
vealedthattheyaregenuinefeaturesandnotacollectionofdis- 3.3.2mmap
crete sources(Pizzo & de Bruyn2008).From now on, we will
call them the NW (north-west) and SW (south-west) relic, re- The2mmapisshowninFig.7.Thenoiserangesbetween2mJy
spectively.Theyhavedifferentshapes.TheSWrelicappearslike and3mJyandislimitedbyclassicalconfusionnoiseintheinner
afilamentofabout8(cid:7) inlengthand2(cid:7) inwidth.Ithasthesame partofthemap(seeSect.2.7andTable3).
orientationoftheknownNErelic,butislocatedontheopposite Aroundthecentralradiohalo,theknownradiogalaxiesbe-
sidefromtheclustercenterandatadoubledistancefromit.The longing to the cluster are still detected. The diffuse emission
646 R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255
Bean
B
A
Embryo
Beaver
A B
NE
Relic
C2
Halo
C1
Bridge
Fig.4.Toppanel:bigfieldgreyscaleimageofA2255at25cm.TheradiogalaxiesBean,Embryo,andBeaverarevisible,atlargedistancefrom
theclustercenter.Theresolutionis14(cid:7)(cid:7)×15(cid:7)(cid:7).Thenoiselevelattheedgeofthefullyimagedfieldis11μJy/beam(seeFig.2).Bottomleftpanel:
zoomintothecentralclusterregion.Wecandistinguishhalo,NErelic,bridgeconnectionbetweenthemandtheradiogalaxiesGoldfish,Double,
TheoriginalTRG,andSidekick.C1andC2seemtobeanextensionoftheradiobridge(seetext).Thecontoursare–0.03,0.03,0.06,0.12,0.24,
0.48,0.96,2,4,8,16,32mJy/beam.Bottomrightpanel:zoomintotheregionoftheclusterwherethenewNWrelichasbeendetectedat85cm
(Sect.3.2).At25cm,theonlyfilamentpointingtowardstheclustercenter(NW1)isdetected,butatverylowlevel(1σ).Herethecontoursare
–0.01(grey),0.01,0.02,0.04,0.06,0.08,0.16,0.32,0.64,1.2,2.4mJy/beam.Themapsarenotcorrectedfortheprimarybeam.
Table4.TheextendedradiogalaxiesofA2255. Thelargestdetectablestructureinour85cmobservationsis∼1◦,
which meansthat nondetecting this extendedfeatureat 85 cm
Name RA(J2000) Dec(J2000) cannot be due to uv plane coverageissues. Instead, it supports
Bean 17h15m31s +64◦39(cid:7)28(cid:7)(cid:7) themorelikelyhypothesisthatthenewemissionregionisasteep
Beaver 17h13m19s +63◦48(cid:7)16(cid:7)(cid:7) spectrum feature, whose nature is differentfrom the NW relic.
Double 17h13m28s +64◦02(cid:7)50(cid:7)(cid:7) We note that point sources might make an important contribu-
Embryo 17h15m05s +64◦03(cid:7)42(cid:7)(cid:7) tiontothenewextendedemissionfeature.Totestitsnature,we
Goldfish 17h13m04s +64◦06(cid:7)56(cid:7)(cid:7) removedthe contributionofpointsourcesdetectedat 25cm in
TRG 17h12m23s +64◦01(cid:7)46(cid:7)(cid:7) this area, assuming a nominalspectral indexof α = −0.7. The
Sidekick 17h12m16s +64◦02(cid:7)14(cid:7)(cid:7) totalfluxsubtractedintheregionis635mJy.Theresult,shown
in Fig. 8, confirms the genuine diffuse nature of the feature.
The source at location RA = 17h10m37s, Dec = +64◦30(cid:7)24(cid:7)(cid:7)
associated with A2255 seems to be rather complex. The radio isduetoablendof2pointsources,ascanbeseenathigherfre-
haloisextendedtowardsNWandisconnectedtotheNWrelic. quency(Fig.6). Giventheir steep spectrum(α ∼ −1),they are
R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255 647
Fig.5.CompositeimagesofA2255obtainedfromsuperposingtheradioandopticalimages.TheWSRT85cmradiomap(ingrey)forthetotal
field(centralpanel),NWrelic(bottomleftpanel),andSWrelic(bottomrightpanel)areshownoverlaidontheredbandDigitizedSkySurvey
image(blackandwhite).Theradioimagehasaresolutionof54(cid:7)(cid:7)×64(cid:7)(cid:7)anditisnotcorrectedfortheprimarybeam.
still visible after the subtraction. We computed the upper limit ThephysicalparametersofhaloandNErelicat2marelisted
for the spectral index of this feature using maps at 2 m and at in Table 5.We notethatitis difficulttodeterminetherealsize
85 cm restored with the same resolution (163(cid:7)(cid:7) × 181(cid:7)(cid:7)), cov- andthebordersofthehaloandtheNErelic,thereforetheinte-
ering the same uv range and in which we subtracted the point gratedfluxdensitiesreportedinthetablealsoreflectthisuncer-
sourcesdetectedat25cm.Inthe2mmap,thefeaturehasapeak tainty.Theangularsizesofthetwostructuresareassumedtobe
brightnessof28mJy/beam,whileinthe85cmmapwecanonly thesamethanat85cm.
give an upper limit, considering 3 times the noise of this map TheNWrelicisvisibleat2m,whiletheSWone,whichhas
(σobs = 1 mJy/beam), i.e. 3 mJy/beam. As a result, we obtain a surface brightness lower than the NW relic at 85 cm, is not
that the newly detected featureshould have a spectrum steeper detected (S < 6 mJy). This is mainly due to the confu-
SWrelic
than−2.6. sion limit in the central area of the map, where the noise level
Because of the low resolution of the image, the NE relic is3mJy.
seems to be embedded in the halo emission. The feature as- Otherextendedfeaturesare detectedto the E,NWandSW
sociated with it, already detected at 85 cm and tentatively at oftheclustercenterandatverylargeprojecteddistancefromit.
25 cm (C2 in Fig. 4, bottom left panel), is now more promi- ThefirstoneislocatedatRA=17h17m25s,Dec=+64◦05(cid:7)15(cid:7)(cid:7),
nent. However, we notice that, in this case, part of it could be thesecondoneatRA= 17h09m29s,Dec= +64◦36(cid:7)44(cid:7)(cid:7) andthe
due to the radio source located at RA = 17h14m04s, Dec = third oneat RA = 17h06m10s, Dec = +63◦18(cid:7)42(cid:7)(cid:7). Comparing
+64◦16(cid:7)10(cid:7)(cid:7),clearlydetectedatboth25cmand85cm. the2mmapwiththefullresolution85cmand25cmimages,it
648 R.F.PizzoandA.G.deBruyn:RadiospectralstudyoftheclusterofgalaxiesAbell2255
NW1
NW
NE
NW2
Relic
Halo
Fig.6. Contour map of A2255 at 85 cm.
SW The resolution is 54(cid:7)(cid:7) ×64(cid:7)(cid:7). The noise level
at the edge of the fully imaged field is
∼0.1mJy/beam (seeFig.2). Thecontours are
–0.3(grey),0.3,0.6,1.2,2.4,4.8,9.6,20, 40,
80,160mJy/beam.Thedottedellipsesindicate
thestructuresalreadyknownthroughprevious
1 Mpc studies,asthehaloandtheNErelic,whilethe
solidellipsesrefertothenewlydetectedrelics.
Werefertothetextfor adiscussionabout the
positivefeatures at theeast sideof the cluster
center.Themapisnotcorrectedfortheprimary
beam.
NW
NE
Relic
Halo
Fig.7.ContourmapofA2255at2m.Theres-
olutionis163(cid:7)(cid:7) ×181(cid:7)(cid:7).Thenoiselevel at the
edgeofthefullyimagedfieldis∼2mJy/beam
(see Fig. 2). The contours are –0.007 (grey),
0.007, 0.014, 0.028, 0.056, 0.1, 0.2, 0.4, 0.8,
1.6, 3.2 Jy/beam. The radio halo is more ex-
tended towards NW than at 85 cm and is di-
rectly connected to the NW relic. The dot-
1 Mpc tedlinesindicatethestructuresalreadyknown
through previous studies, as the halo and the
NErelic,whilethesolidlinesrefertothenewly
detectedrelics.Themapisnotcorrectedforthe
primarybeam.