PDF (15.67 MB) PDF
Preview PDF (15.67 MB)
A&A585,A41(2016) Astronomy DOI:10.1051/0004-6361/201527007 & (cid:13)c ESO2015 Astrophysics The Effelsberg-Bonn H Survey: Milky Way gas(cid:63) First data release B.Winkel1,2,J.Kerp1,L.Flöer1,P.M.W.Kalberla1,N.BenBekhti1,R.Keller2,andD.Lenz1 1 Argelander-InstitutfürAstronomie(AIfA),AufdemHügel71,53121Bonn,Germany 2 Max-Planck-InstitutfürRadioastronomie(MPIfR),AufdemHügel69,53121Bonn,Germany e-mail:[email protected] Received20July2015/Accepted26October2015 ABSTRACT Context.TheEffelsberg-BonnHSurvey(EBHIS)isanew21-cmsurveyperformedwiththe100-mtelescopeatEffelsberg.Itcovers thewholenorthernskyouttoaredshiftofz∼0.07andcomprisesH lineemissionfromtheMilkyWayandtheLocalVolume. Aims.Weaimtosubstitutethenorthern-hemispherepartoftheLeiden/Argentine/BonnMilkyWayHsurvey(LAB)withthisfirst EBHISdatarelease,whichpresentstheH gasintheMilkyWayregime. Methods. The use of a seven-beam L-band array made it feasible to perform this all-sky survey with a 100-m class telescope in a reasonable amount of observing time. State-of-the-art fast-Fourier-transform spectrometers provide the necessary data read-out speed, dynamic range, and spectral resolution to apply software radio-frequency interference mitigation. EBHIS is corrected for strayradiationandemploysfrequency-dependentflux-densitycalibrationandsophisticatedbaseline-removaltechniquestoensurethe highestpossibledataquality. Results.DetailedanalysesoftheresultingdataproductsshowthatEBHISisnotonlyoutperformingLABintermsofsensitivityand angularresolution,butalsomatchestheintensity-scaleofLABextremelywell,allowingEBHIStobeusedasadrop-inreplacement forLAB.Dataproductsaremadeavailabletothepublicinavarietyofforms.Mostimportant,weprovideaproperlygriddedMilky WayH columndensitymapinHEALPixrepresentation.TomaximizetheusefulnessofEBHISdata,weestimateuncertaintiesin theHcolumndensityandbrightnesstemperaturedistributions,accountingforsystematiceffects. Keywords.surveys–ISM:atoms–techniques:spectroscopic 1. Introduction which mapped the southern sky out to a radial velocity of ∼13000 kms−1 with the 64-m Parkes telescope. Limitations In 2009 we initiated a new northern-hemisphere H survey in the spectrometer meant that the velocity resolution is only with the 100-m telescope at Effelsberg, Germany, to succeed 18 kms−1, so too coarse to be useful for many studies of the theLeiden/DwingelooSurvey(LDS;Hartmann&Burton1997) Milky Way (MW) and its halo. Therefore, a second large-area donewiththe25-mDwingelootelescope.Outstandinginterms Parkes H survey was initiated in 2005: the Galactic All-Sky ofsensitivityandskycoveragecomparedtoanypriorendeavor, Survey (GASS; McClure-Griffiths et al. 2009; Kalberla et al. the LDS was later merged with the Instituto Argentino de 2010; Kalberla & Haud 2015). It recorded data with narrow Radioastronomía Survey (IAR; Arnal et al. 2000; Bajaja et al. bandwidth,hencehigherspectralresolution. 2005,forδ<−27◦.5)toformtheLeiden/Argentine/BonnSurvey AnotherrecentprojectistheGALFA-HIsurvey(Peeketal. (LAB; Bajaja et al. 1985; Kalberla et al. 2005) – the first full- 2011)currentlygoingonwiththeArecibo300-mdish.Thesheer skyMilkyWayH surveythatwascorrectedforstrayradiation sizeofthetelescopemakesGALFA-HIuniqueintermsofsen- (SR).Thehigh-qualitySRcorrectionmadetheLABsurveyone sitivity. For the Galactic velocity regime, a dedicated backend of the most important H data bases to date. The references to isdeliveringspectralresolutionupto0.2kms−1.Thedownside the seminal article by Kalberla et al. (2005) reveal the tremen- is the limited area on the sky that is accessible to the Arecibo douslegacyvalueoftheLAB:oneofitsmostfrequentusesisto telescope. correcthigh-energyobservationsforgalacticforegroundextinc- In late 2008 a seven-beam 1.4-GHz (L-band) receiver was tion.However,thecoarseangularresolutionofLABmaycause installed at the 100-m telescope at Effelsberg. Attached to the significant uncertainties when it is applied to evaluating the in- seven-feed array are state-of-the-art digital FFT-type spectrom- tensityattenuationofunresolvedsources. eters (FFTS, Stanko et al. 2005; Klein et al. 2012), allowing to With the development of 21-cm multibeam receivers in the observe Galactic and extra-galactic H simultaneously for the late1990s,itfinallybecamefeasibleforthe100-mclasssingle- firsttimewithsufficientspectralresolution.TheFFTSnotonly dishobservatoriestosurveysignificantportionsoftheskywith offersgreatspectralresolution(∼1kms−1)butalsoallowshigh- theresultingbetterangularresolutioninreasonableamountsof speed (∼1 s) storage of spectra. Both properties are beneficial observing time. The first such project was the very success- forremovingtime-dependentradiofrequencyinterference(RFI) ful H Parkes All-Sky Survey (HIPASS, Barnes et al. 2001), fromthedataduringpost-processing. (cid:63) EBHISMilkyWayHIdataisonlyavailableattheCDSvia With this new receiving system, we conducted a major anonymousftptocdsarc.u-strasbg.fr(130.79.128.5)orvia H survey of the sky north of declination δ (cid:38) −5◦, called http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/585/A41 Effelsberg-BonnHSurvey(EBHIS).Toallowforanearlyfirst ArticlepublishedbyEDPSciences A41,page1of22 A&A585,A41(2016) 1 2 h 9h 15h 6h 18h 60◦ 22 3h 30◦ 21h 212cm])− [HI N ( g 20lo 0◦ 0 h Fig.1.EBHISHcolumndensitymap,asintegratedoverthevelocityinterval−600≤v ≤600kms−1. lsr data release, survey observations were divided into two runs, recentGalacticH all-skysurveys,LABandGASS,anddiscuss eachmappingthenorthernskycompletelywithanetintegration currentlimitationsofthedata.InSect.5westudyensembleun- time of about 35 s per beam. In April 2013 we finished first- certainties of the observed column density and brightness tem- coverage observations. This data has now been processed and perature distributions. Section 6 explains how EBHIS data are ismadeavailabletothescientificcommunity.Meanwhile,mea- made accessible to the astronomical community, most impor- surements for the second coverage are ongoing. Once finished, tantlytheall-skyH columndensitymap(Fig.1).Weconclude EBHISsensitivitywillimprovebyabout30%. withasummaryandoutlook. The main science goals of EBHIS have already been dis- cussed in detail in Kerp et al. (2011). Here, we concentrate on 2. Surveydescription assessing the quality of the final data products and on the way Observations of the first coverage were performed in on-the- thesewillbemadeavailabletotheastronomicalcommunity. fly mapping mode, scanning in right ascension, α, along lines In Sect. 2 we briefly recapitulate the properties of EBHIS. of constant declination, δ. Since the parallactic angle changes The data processing pipeline was explained in detail in Winkel duringeachscanline,theseven-feedarray1 needstoberotated etal.(2010);however,severalstepswerefurtherimproved,such as RFI flagging, baseline-fitting, and correction for stray radia- 1 Onedual-circularcentralfeedandsixdual-linearoff-axisfeedsina tion, which we present in Sect. 3. Section 4 comprises the data hexagonallayout.Theseparationoftheoff-axisfeedsfromtheoptical qualityevaluationofEBHIS.WecompareEBHISwiththetwo axisis15(cid:48). A41,page2of22 B.Winkeletal.:TheEffelsberg-BonnHSurvey:MilkyWaygas accordingly to ensure a regular scan pattern of the offset feeds. Table1.ComparisonofbasicparametersofrecentHsurveysofthe Furthermore,thefeedarrayisrotatedbyanadditional19◦.1with MilkyWay. respecttothescanningdirectionsuchthatthescantracksofthe individualfeedshaveequalseparations. LAB GASS GALFA∗ EBHIS The sky north of δ > −5◦ was divided into 25 deg2 sec- δ Full ≤1◦ −1◦...38◦ ≥–5◦ tionsforatotalof915individualmaps.Thepolarcapδ >∼ 85◦ ϑ 36(cid:48) 16.(cid:48)1 4.(cid:48)0 10.(cid:48)8 wasobservedintheGalacticcoordinateframe,becausenearthe |vfwh|m ≤460† ≤470 ≤750 ≤600 kms−1 equatorialpole,theAz–Elmountofthe100-mtelescopewould ∆lvsr 1.03 0.82 0.18 1.29 kms−1 significantlyaffectthepossiblemappingspeedintheequatorial δv 1.25 1.00 0.18 1.44 kms−1 system.HereafterwefollowtheEffelsbergobservatoryparlance, σ 80 57 325 90 mK where one observation/map is called “scan”, consisting of sev- σrnmorsm 89 57 140 108 mK rms eral“subscans”(theindividualscanlines). Γ 0.132 0.649 10.52 1.434 KJy−1 mb Strong observing time constraints were applied for each field: NHlim 3.9 2.5 6.1 4.7 1018cm−2 1. The southeastern quadrant is preferred in azimuth to maxi- SHlim 16.1 2.1 0.3 1.8 Jykms−1 mizetheangulardistancetoamobile-communicationbroad- Notes. The table quotes the declination range, δ, angular resolution casttowerinthevicinityofthetelescope(atAz≈290◦). ϑ ,velocityinterval,v ,channelseparation,∆v,spectralresolution, 2. Weneedtoavoidlargedifferentialchangesofparallactican- fwhm lsr δv,brightnesstemperaturenoiselevel,σ ,aswellasthenormalized rms gle during a scan line because the rotation speed of the re- noiselevel,σnorm,thedatawouldhaveatacommonspectralresolution rms ceiverboxatthe100-mdishislimited. of1kms−1.Themainbeamsensitivity,Γ ,istheconversionfactorbe- mb 3. The angular distance to the Sun has to be kept as large as tweenbrightnesstemperatureandfluxdensity,i.e.,T =Γ S.Thetwo B mb possibleduringdaytimeobservationstoavoid“solarripples” bottomrowsquotetheoretical5σdetectionlimits(velocity-integrated in the spectral baselines (see also Barnes et al. 2005, and intensity)ofanobjectwithaGaussianprofileof20kms−1 linewidth Sect. 4.4.1). These are most likely caused by reflections of (FWHM).(∗)Numbersgivenarefortheshallowestmode.Effectivein- solaremissionoffthetelescopestructure.Theexactpathsof tegration time varies across the survey area. Fields observed in com- reflectionaredifficulttopredict,anditisthereforenearlyim- mensal mode with AGES (Minchin et al. 2007, σnrmorsm = 33 mK) and ALFALFA(Giovanellietal.2005;Haynesetal.2011,σnorm=60mK) possibletoavoidsolarinterferencebymeansofobservation rms are much deeper (J. Peek, priv. comm.). (†) Northern (LDS) part only scheduling.Becauseofthat,severalfieldshadtobeobserved goesouttov ≤400kms−1. againforoursurvey. lsr Thefast-Fouriertransform(FFT)spectrometersallowstoringa (FWHM)ofw = 20kms−1.Likewise,alimitingcolumnden- full spectrum every 500 ms, which is an advantage to identify- 50 ing time-variable RFI. Fast read-out is also beneficial, because sity,NHlim,wascalculatedwithoutass√umingpointlikeness.Both the dump time restricts the maximum mapping speed because quantities,NHlim andSHlim,scalewith w50andσnrmorsm. eachindependentbeamareaontheskyneedstobesampledwith at least two data points along the scanning direction (Shannon 3. Datareductionrevisited 1949).Toallowforareasonablyfastcompletionofthefirstsky coverage,wechosearelativelyhighscanspeedof240(cid:48)(cid:48)persec- InthefollowingwereportonmajormodificationsoftheEBHIS ond,pushingtheazimuthalenginesofthe100-malmosttotheir datareductionsoftwareovertheproceduresreportedinWinkel limitsforsourcesaboveanelevationof60◦.Wespreadtheob- etal.(2010). served bandwidth of 100 MHz over 16384 channels, yielding a spectral resolution of 6.1 kHz. For the Milky Way data re- lease, only the velocity range of |v | ≤ 600 kms−1 is consid- 3.1. ImprovedRFIflagging lsr ered,whichisaboutthefrequencyintervalfrom1417.5MHzto 3.1.1. Automatedflaggingalgorithm 1423.5MHz. After gridding, the median noise level in the data cubes is Flöer et al. (2010) described how to make the best use of the approximately90mKbutdifferssignificantlyfromfieldtofield factthatallsevenfeedsofthereceiverareexposedtothesame (see Sect. 5.1). This is mainly a consequence of different ele- RFIenvironment.FollowingtheirapproachimprovesRFIrejec- vation angles during the observations causing different levels tionefficiencysignificantly.WebrieflyreviewthebasicRFIflag- of stray radiation from the ground and atmosphere. A second gingworkflow.Ourautomatedflaggingpipelineisoptimizedfor importanteffectistheseasonalchangesinambienttemperature threedistincttypesofRFIencounteredattheEffelsberg100-m (compareWinkeletal.2010). telescope: In Table 1 we compile important survey parameters of 1. Near-constantnarrow-bandspikes,typicallyaffectingoneor EBHISincomparisonwiththemostrelevantotherrecentsingle- twospectralchannels; dishMilkyWayH surveysLAB,GASS,andGALFA-HI.Apart 2. Intermittent broad-band events affecting a hundred up to from the T noise level, which is usually quoted for full spec- B thousandsofspectralchannels; tral resolution data, we also provide a normalized noise value, 3. Extremely strong RFI caused by the L3 mode of the σnorm,scaledtoavelocityresolutionof1kms−1.Thepurenoise rms GPSsatellitesystem. level,however,onlyquantifiestheinstrument’sabilitytodetect gas, which entirely fills the observing beam. For the many un- To detect the first two types of RFI, we use very similar detec- resolvedorpoint-likeobjectsinthesurveys,abettersensitivity tionstrategies.Becausewehave14independentmeasurements proxyisthevelocity-integratedfluxdensity,SHint.Inthetablewe (7feedswith2polarizationchannelseach)oftheRFIenviron- quote numbers for the 5σ flux-density detection limit, SHlim, of ment at any given time, we use coincidence flagging to distin- anunresolvedobjectwithaGaussianprofilehavingalinewidth guish man-made interference from astronomical signals. Here A41,page3of22 A&A585,A41(2016) weassumethatasignalbeingpresentinmorethanonefeedat T [K] sys thesametimedoesnothaveanastronomicalorigin,becauseall 25 30 35 40 45 50 sevenfeedsarepointedatdifferentlocationsonthesky.Thisas- sumptionisviolatedifanastronomicalsourceisveryextended, 40 asfordiffuseHemissionfromtheMilkyWaydisk.Wethere- 30 forealsousematchedfilteringadaptedtothetypicalappearance p m ofRFIinatime-frequencyplot.ThisenablesareliableRFIde- u 20 D tection with the exception of the innermost part of the bright 10 Galactic emission, where contributions at all spatial frequen- cies are present that cannot be reliably suppressed by matched 0 filtering. 40 Toprocesseachindividualobservation,wefirstaveragethe 30 two orthogonal polarizations for each feed. Since RFI is often p m stronglypolarized,eitherbyreflectionorintrinsically,thereare u 20 D caseswhereanRFIeventisonlypresentinasinglepolarization 10 channel.Keepingthepolarizationsseparatewouldthenweaken our assumption that all feeds are exposed to the same RFI en- 0 1.430 1.425 1.420 1.415 1.410 vironment.Whilereducingthesensitivitygainedfromrequiring √ Frequency [GHz] signal coincidence, we still gain a factor of 2 in sensitivity fromtheaveraging.Afteraveraging,thesubscansofeachobser- Fig.2. Narrow-band flagging result. Top panel: spectrogram of raw vation are processed independently, allowing multicored CPUs data,bottompanel:same,butwithflagsvisualizedasverticallines. tobeexploitedforprocessingmultiplesubscansinparallel. Todetectnarrow-bandRFI,weaverageeachsubscanintime, yielding seven spectra per subscan. Because narrow-band RFI canreducethefidelityofthedataoveralargepercentageofthe eventsaretypicallyconfinedtofewerthanthreechannelsatfull 100-MHzband(about50%). spectral resolution, we remove any large scale spectral compo- Thedataaffectedbybroad-bandandGPSinterferenceisex- nentbysubtractingamedian-filteredversionofeachspectrum. cludedfromfurtherprocessing,i.e.,flagged.Typically,thisdoes Toflagacertainchannelascontaminated,weusecombinatorial not limit the sensitivity of the Galactic survey, because broad- thresholding. Instead of marking a channel as containing RFI bandinterferencemostlyoccursatthelow-frequencyendofthe once its level exceeds a certain threshold t , we require that a 1 100-MHz-wideobservingband.Also,GPSinterferenceusually channelexceedsalowerthresholdt inNfeedssimultaneously. N doesnotaffecttheradialvelocities−600<v <+600kms−1. AssumingthenoiseinthedataisGaussianwithstandarddevia- LSR Because narrow-band interference in EBHIS data is numerous tionσ,thethresholdsfort thathavethesamestatisticalsignif- N and often has near constant intensities, we attempt to subtract icanceasanindividualthresholdt arecalculatedbysolvingthe 1 the RFI contribution instead of completely flagging an affected equation channel. This removes most of the narrow-band interference in 1−Φ(t1)=[1−Φ(tN)]N (1) theGalacticsurveywiththeexceptionofafewnarrow-bandsig- where Φ(x) is the cumulative distribution function of the stan- nals that mix with the bright H emission from the Milky Way disk. Here, a reliable subtraction is not possible and the inter- dardnormaldistribution.Herewealsomakeuseofthefactthat ference is left untouched. From our perspective, this is better allRFIeventshavepositiveintensities. than flagging the spectral channels in question completely be- Using this type of combinatorial thresholding, we can im- cause leftover narrow-band RFI signals have amplitudes (typi- posethecoincidencerequirementandwiththatmakebetteruse of the available data to optimize our detection efficiency. We cally (cid:46)0.5 K) that are small compared to the bright MW disk choose a lowest required coincidence level of N = 4. To flag emission. acertainchannel,wecomparethedatafromallfeedstoallsuc- In the final data cubes, the Doppler correction causes fre- cessivethresholdsforN =4...7andflagthechannelifthecrite- quency shifts of narrow-band RFI signals. In the topocentric rionismetforanyN.AnexamplefortheRFIdetectionquality frametheyaremostlyfixedinfrequency(afewkHzchangedur- is provided in Fig. 2. It shows a time–frequency plot (spectro- inghours),whereasafterapplyingtheLSRcorrection,shiftsof gram) of one subscan of raw data (one feed) in the top panel. up to two spectral channels occur. Therefore, residual narrow- Thebottompanelshowsthesamedatabutwithallnarrow-band band RFI can appear as “wave-like” patterns when visually events,detectedbytheRFIflagger,marked. inspecting consecutive channel maps in the data cube (see To detect broad-band interference, we apply a three-point Sect.4.2). medianfilterinthetimedomainofthedatatosuppressallper- sistent signals. We furthermore smooth the data in the spectral 3.1.2. ManualRFIflagging domain with a Gaussian filter adapted to the typical extent of thebroad-bandevents.Withtheprepareddataweagainperform The RFI described is present in all observations. Apart from combinatorialthresholdingacrossthesevenfeeds. these regular signals, there are additional glitches that are not SincetheGPSL3modeisextremelybrightandradiatesal- easily recognized by automated algorithms. To ensure consis- ways at a center frequency of 1381.05MHz, we simply com- tently high data quality, each observation was also inspected paretherootmeansquare(rms)ina1MHzwindowaroundthis manually. A simple graphical user interface was developed to frequencytothermsinneighboringfrequencies.Ifthermsdif- compute spectrograms (images of the time-frequency plane). fersbyafactoroftwo,weflaga10-MHz-widewindowaround The user can quickly iterate over the different subscans, feeds, this frequency. This removes the brightest part of the interfer- andpolarizationchannelstosearchforRFIandotherdefectsin ence, but GPS L3 features extended spectral side lobes, which thedata.Bymouse-clickingonaspectraldump,anRFIflag(bad A41,page4of22 B.Winkeletal.:TheEffelsberg-BonnHSurvey:MilkyWaygas 40 1.095 30 1.080 p m Du 20 1.065 10 1.050 Central Feed L Offset Feed X 0 1.035 Pν fi › / 40 P 1.020 30 mp 1.005 u 20 D 0.990 10 Central Feed R Offset Feed Y 0.975 0 1.430 1.425 1.420 1.415 1.430 1.425 1.420 1.415 1.410 Frequency [GHz] Frequency [GHz] Fig.3.BroadbandRFIevent,whichisonlyvisibleinasinglespectraldump(31).Theleftpanelsshowthetwopolarizations(LandR)ofthe centralfeed,andrightpanelscontainoffsetfeeddata(XandY).Forimprovedvisualization,dataisbinnedinfrequency(32-fold,∆ν=195kHz), andeachspectraldumpisdividedbythemedianspectrum.TheMilkyWayH emissionlineappearsslightlydisplacedfromitsrestfrequency owingtotheLSRDopplercorrection,whichwasnotappliedhere. spectrum) is generated in a data base. To improve the RFI-to- 5 40 noiseratio,dataisbinnedinfrequency(usually32-fold,giving a channel width of 195 kHz). To remove the bandpass shape, 4 30 whichwouldotherwisedominatethespectrogram,eachspectral dAusmidpeiseffdeivcitdiesdthbayttthheemMeidlkiaynWspaeyctermumissoiofnthleinceuirsreonntlsyusbhsocwann. ump 20 3 Pν ›fi D 2 P/ asaresidualsignal.ForfieldsclosertotheGalacticdisk,where theMilkyWayishighlystructured,thiscanmakeitdifficultto 10 1 distinguishfaintbroad-bandRFIfromastronomicalfeatures. In the following we briefly discuss the three most fre- 0 0 quenttypesofbroad-bandinterferencethatwereencounteredin 1.10 40 EBHISMilkyWaydata.Figure3showsanexampleofawide- bandburstthatispolarized,varieswithfrequency,andcouples 1.05 differently into the individual feeds. It lasted less than 500 ms 30 and affected the full 100 MHz band. Another glitch that is ex- mp 1.00 Pν fi tending over the full 100 MHz frequency range is plotted in u 20 › Fig.4(toppanel).Here,mostoftheaffectedspectrumwasmore D P/ negativethanthemedianspectrum,whichismostlikelycaused 10 0.95 by a high-intensity out-of-band RFI event, saturating the front end.Intermodulationproductsaredetectedalloverthespectrum, 0 0.90 for instance, at 1425 MHz. The bottom panel of Fig. 4 depicts 1.430 1.425 1.420 1.415 1.410 twoshortburstsofevents,whichareafew100-kHzwide,super- Frequency [GHz] posed on Milky Way emission. In this example, RFI and MW can be distinguished well, but closer to the Galactic plane this Fig.4. Examples of RFI. Top panel: a very intense out-of-band event becomesmoredifficult.Thetimebetweenbotheventsis12s– temporarilysaturatesthereceiverandcausesintermodulationproducts (e.g.,at1425MHz).Bottompanel:twoshortburstsofRFI,likelyradar whichisatypicaldelayforRadarpulses.Sometimesuptothree pulses,withawidthofafew100kHz. consecutiveburstswereobserved,butnevermore.Wespeculate that the 100-m dish may have received reflected radar pulses, suchasfromairplanes.However,theoriginalemitterwasnever extending the fitting procedure to the time-frequency domain. identified. Therevisedalgorithmmadeuseofthefactthatthesystemtem- peratureinthebaselinechangesonlyslowlywithtime. 3.2. New2Dbaselinefitting Further tests revealed, however, that faint sources were sometimes surrounded by negative baseline artifacts. The Winkel et al. (2010) proposed adaptive Gaussian smoothing Gaussian-smoothing baseline algorithm relies on proper flag- to calculate baseline solutions for each individual calibrated gingormaskingofsources.Otherwise,thefluxcontributionof spectrum. Even though this approach performed well in most sourceswillbesmearedintothebaselinesolutionitself.Inthis cases, we improved the robustness of the baseline solution by context, a source is thought to be anything that superposes the A41,page5of22 A&A585,A41(2016) baseline, such as H clouds in the Milky Way and its halo or 3.4. T -basedweightingscheme sys other galaxies (both in continuum and spectral line emission). ForEBHISdata,thesystemtemperaturesvarybetweenthedif- The baseline algorithm itself tries to mask sources, for exam- ferent feeds and naturally also between the phases where the ple,byiterativeflaggingof3σoutliers.However,sincebaseline- noise diode is switched on and off. It is possible to achieve a fittingisdoneontherawspectraldatathathaveabout500mK slightly lower final noise level of the data by weighting each rms,faintsourcesarenotalwaysreliablydetected. spectrum with its associated system temperature (see Winkel It turned out that in these cases, polynomials are better et al. 2012b). Since we do not use frequency switching to re- suited for baseline estimation. They can describe underlying move the bandpass from our data, we can reconstruct the full fluctuationswell,butproduceshallowertroughswhennomask frequency-dependentsystemtemperatureduringcalibrationand is provided. Because the Gaussian-smoothing baseline method baselinecalculations. benefitedfromextensiontothetime-frequencyplane(t, f),we One difficulty remains: the H emission line itself con- implemented 2D polynomial fitting, where the baseline, y , is b tributes to the system temperature at the appropriate frequen- describedby cies. In theory, one had to add each individual H profile to (cid:88) the underlying continuum T spectrum to obtain the proper yb = αi,jfitj. (2) weights.Thiswouldbedisadsvyasntageousinpracticebecausethe i,j≥0 individual 500 ms-line profiles are very noisy; σ ∼ 0.5 K. rms Suchweightingwouldintroduceanadditionalnoisecomponent. Baselines in EBHIS have a complicated structure, especially Therefore, we decided to compute the average H line profile along the frequency axis. To avoid unreasonably high polyno- persubscanandperbeamanduseitasthesystemtemperature mial orders, we fit the data on tiles of 1024 spectral channels inputfortheweighting.Strictlyspeaking,thisattemptisanover- timesthenumberofdumpspersubscan(about40).Tosuppress simplificationformapswheretheT levelvariessignificantly sys sharp gaps in the baseline solution at the tile edges, we inter- during the subscan, for instance, owing to large changes in el- leavetiles(overlap:512channels)andinterpolatebetweeneach evation. However, since all feeds experience the same gradient oftwoadjacentsolutionsusingsigmoidthresholds.Goodresults in system temperature, only a higher order effect is introduced are achieved with polynomial orders of ten in spectral and two into the weighting, and that can be neglected for our purpose. intemporaldirection,whileallowingonlyonecrossterm,α1,1, IntroducingtheTsys-basedweightingschemeimprovesthefinal todescribemildtiltsinthe2Dbaseline. rmslevelbyabout1to2%. To improve the baseline solution even further, we ini- tially used the LAB survey to provide a spectral mask around 3.5. Stray-radiationcorrection MWemission.Thisissupplementedwithinformationfromthe HyperLEDA (Makarov et al. 2014) database2 containing ex- Galacticemissionat21-cmisseeninalldirections,butpredom- tragalactic H objects and NVSS (Condon et al. 1998) to flag inantlyfromtheGalacticplaneextendingacrossthewholesky. strongcontinuumsources(≥1.5Jy).Weakercontinuumsources Antennasidelobes,pointingtosuchregionsofhighfluxdensity, arehandledwellbysubtractingtheaverageT levelfromeach sys will eventually produce artificial signals unrelated to the inten- spectral dump prior to computing the baseline. Finally, a full- sityreceivedbytheprimarybeam.Thisso-calledstrayradiation resolutiondatacubeofEBHISitselfisusedtomaskH emission varieswithtimeandseason(vanWoerden1962)andismostcrit- fromtheMWanditshaloinaniterativescheme. ical for observations of high-galactic latitude objects with faint H emission. In extreme cases, stray radiation may provide a largercontributiontoameasuredspectrumthanthetruebright- 3.3. FrequencydependencyofT cal nesstemperatureoftheobservedskyposition. Winkel et al. (2010) explain the intensity calibration of EBHIS The side-lobe structure of a telescope depends on antenna dataincorporatingthefrequency-dependenceofthesystemtem- type and design. For a paraboloidal reflector, the extended side perature, T . There, the temperature, T , of the noise diode, lobes, called stray cones, are mainly caused by the support sys cal which is fed into the receiver, was assumed to be constant legs that carry the prime focus cabin and/or secondary mirror over frequency. In the meantime, based on the methods pro- (Kalberla et al. 1980a, their Fig. 6). Radiation is also received posed in Winkel et al. (2012b), we could derive the frequency fromregionsoutsidetherimofthereflector,thespilloverregion. dependence of T using continuum calibration sources (e.g., Reflectingsurfaceswithintheaperturecanalsocauseadditional cal 3C48,NGC7027).Unfortunately,suchameasurementistime- sidelobes(seeSect.3.5.4). consuming, since each beam must be positioned individually Theimpactofstrayradiationcanbeminimizedbyreducing ontothecontinuumsource. thenumberofscatteringsurfaceswithinthetelescopeaperture. Asafasteralternative,weusedanabsorber–placedaround PrimeexamplesaretheBellLabshornreflectorantenna(Stark thesevenfeeds–toobtainasimultaneousandalmostRFI-free etal.1992)andtheGreenBankTelescope(GBT;Prestageetal. measurement of all noise diodes. Using the hot absorber alone 2009). For these telescopes the stray radiation is reduced sig- cannotprovideabsolutecalibrationoftheT spectra.However, nificantlyincomparisontoastandardparaboloid;i.e.,themain cal it is well suited to assessing the frequency-dependent behavior. beamefficiencyisincreasedfromabout70%to90%.However, The absolute value of T at 1.42 GHz can then be measured the remaining 10% of side-lobe efficiency can still be serious cal using observations of the IAU standard position/source S7 at enough that a numerical correction of the observations is un- v ≈ 0 kms−1, as explained in Winkel et al. (2010). This pro- avoidable(Boothroydetal.2011). lsr cedure was repeated several times during the whole observing FirstattemptsbyvanWoerden(1962)tocorrectobservations campaign and yielded consistent results for the Tcal spectra of of the Dwingeloo telescope for stray radiation were limited by all14channels(7feedswith2polarizationseach). a lack of sufficient computing power. Kalberla et al. (1980a,b) providedthefirstcorrectionsfortheEffelsberg100-mtelescope. 2 http://leda.univ-lyon1.fr/ Lockmanetal.(1986)obtainedsimilarresultsbybootstrapping A41,page6of22 B.Winkeletal.:TheEffelsberg-BonnHSurvey:MilkyWaygas from the Bell Labs Survey (Stark et al. 1992), which is af- withinthefirstminimumoftheantennadiagram.Theexistence fected only a little by stray radiation. Subsequently, Hartmann ofasolutionofEq.(7)isthereforewarranted.Astandardproce- et al. (1996) and Kalberla et al. (2005) developed a correction dureforfindingthesolutionistousesuccessiveapproximations, for the resurfaced Dwingeloo dish, Higgs & Tapping (2000) asproposedbyBracewell&Roberts(1954). for the 26-m Telescope at the Dominion Radio Astrophysical Knowing these basic limitations, we can develop a strategy Observatory,Bajajaetal.(2005)forthe30-mtelescopeatVilla for deriving correct brightness temperatures. Most important is Elisa, Kalberla et al. (2010) for the Parkes 64-m dish, and tofindwaystodetectstrayradiationeffectsintheobservations Boothroydetal.(2011)forthe100-mGBT. and, likewise, to separate regions in the antenna diagram that cause contributions. As far as possible, we measure parame- ters of the antenna diagram P, next we model near-side-lobe 3.5.1. Basics structures (Sect. 3.5.3). For side lobes far from the main beam, The antenna temperature T observed by a radio telescope is we study details of the telescope construction and employ ray- A givenforeachoftheindividualreceiversbyaconvolutionofthe tracingtofindthecriticalregionsinthepattern(Sect.3.5.4). true brightness temperature distribution T on the sky with the Inmostcasesitisimpossibletodetermineaccurateabsolute beampatternPoftheantenna: side-lobelevelsfromraytracingalone,butbyvaryingindividual freeparameters(intensitiesfor69side-lobestructuresinthefinal (cid:90) T (x,y)= P(x−x(cid:48),y−y(cid:48))T(x(cid:48),y(cid:48))dx(cid:48)dy(cid:48). (3) versionofthefarside-lobemodel),wegetsatisfactorysolutions. A Asimplebutpowerfultestforidentifyingacriticalregioninthe antenna diagram is to correct the observations with an unrea- HereweuseanapproximationinCartesiancoordinatestosim- sonablyhighside-lobelevel,whichthencausesovercorrections. plifytheexpressionoftheconvolutionintegrals.Ingeneral,T is A Theresultingspuriousnegativefeaturesincorrectedprofilesand time-andfrequency-dependent,sphericalcoordinateshavetobe datacubescanbeeasilyspottedbyeye. used,andtheintegrationneedstotaketheobservablepartofthe Characteristic of stray radiation errors is the variability of sky (i.e., the horizon) into account, as well as ground reflectiv- spurious line components that also led to the detection of stray ity.Forthemainbeamandallsidelobes,atmosphericattenuation radiation in the first place (van Woerden 1962; Kalberla et al. andrefractionneedtobeconsidered,too. 1980b). We make use of this effect. An iterative determination ForthepatternPoftheantennaweusethenormalization of side-lobe levels requires a sufficiently large sample. We use (cid:90) EBHIS for declinations 0◦ < δ < 60◦ and velocities −100 < P(x,y)dxdy=1 (4) v <+100kms−1. lsr Coherentregionsin(l, b, v )areselectedthatcontainlittle ornoHemissionbutaresensitlisvretostrayradiationeffects.First andsplittheantennadiagramintothemainbeamarea(MB)and we use LAB data to verify the absence of significant emission thestraypattern(SP): features.Nextwedeterminepeakdeviationsandindependently (cid:90) thermsscatteroverlargeremission-free(l, b, v )regionswith T (x,y) = P(x−x(cid:48),y−y(cid:48))T(x(cid:48),y(cid:48))dx(cid:48)dy(cid:48) lsr A the aim of minimizing these errors. Our analysis is hampered MB (cid:90) bythefactthatweonlyhaveasingleskycoverage,soitisnot + P(x−x(cid:48),y−y(cid:48))T(x(cid:48),y(cid:48))dx(cid:48)dy(cid:48). (5) possible to compare different EBHIS observations at the same SP position.Strayradiationproblemsare,however,recognizableas Bydefiningthemainbeamefficiency, discontinuities between individual fields that cause a patchy or blockystructure(compareFig.11,toppanel).Ourstrategyisto (cid:90) improvetheSRcorrectionsbysuccessiveapproximations,mod- η ≡ P(x,y)dxdy, (6) MB ifyingantennaparametersbutalsothecorrectionalgorithmitself MB (Sect.3.6),aswellastheestimateforthebrightnesstemperature ofthetelescope,Eq.(5)canberewrittenas distributionT inEq.(7). (cid:90) T (x,y) 1 T (x,y)= A − P(x−x(cid:48),y−y(cid:48))T(x(cid:48),y(cid:48))dx(cid:48)dy(cid:48). 3.5.3. Themultibeamantennapattern:nearsidelobes B η η MB MB SP (7) TosolveEq.(7),itisnecessarytoworkoutthecompleteantenna response, P,foreachofthesevenbeamsoftheEffelsbergmul- To determine the main beam-averaged brightness temperature, tifeedsystem.Forpracticalreasonswedistinguishbetweennear TB,weneedtoknowtheantennapattern, P,withsufficientac- sidelobesaroundthemainbeamwithinadistanceof4◦ andfar curacy, but also the true brightness temperature T on the sky, sidelobesfartherout.Bothregionscoverapproximatelyhalfof which a priori is unknown. Apparently we need an approxima- theantenna’ssolidangleoutsidethemainbeam. tionforT. Using strong radio continuum sources such as CasA, it is hardly possible to measure accurate side-lobe structures below 3.5.2. HowtosolveEq.(7) −40 dB. Such observations are typically restricted to distances (cid:46)1 deg from the main beam but cover only (cid:46)50% of the near There are two unknowns in Eq. (7), T and T. Furthermore, side-lobe beam’s solid angle required in Eq. (7). We therefore B the solution depends on the choice of the main beam area MB. model the side lobes out to radial distance of 4 deg from the We first consider conditions for a solution of this equation. main beam in a similar way to what was previously done for Replacing T on the lefthand side with T leads to a Fredholm GASS. B equationofthesecondkind,whichcanbesolvedunambiguously The far-field pattern of an antenna is approximated by the forη >0.5(Kalberlaetal.1980a).State-of-the-arttelescopes auto-correlation of the aperture plane distribution. We used MB have main beam efficiencies η (cid:38) 0.7, defined for the region themeasuredfeed-hornresponsepattern(Kelleretal.2006)and MB A41,page7of22 A&A585,A41(2016) the telescope geometry, including shadowing caused by the fo- dα[deg] cuscabinandthefeedsupportlegstoderivethecomplexaper- 4 2 0 2 4 − − ture distribution function for each feed. The pattern was then 4 central calculated by Fourier transformation in a similar way to Baars feed (2007,theirSect.2.2). For the seven beams, we distinguish between the central feed and the six surrounding feeds in a hexagonal layout. The offset beams have a radial offset of 17.3 cm from the optical 2 axis. As mentioned in Sect. 2, the seven-beam receiver is ro- tatedduringobservationstoaccountforthechangingparallactic angle that would otherwise distort the resulting scan pattern in the equatorial coordinate frame. Unfortunately, this means that eg] d 0 the feed horns rotate relative to the focus cabin’s support legs. [ δ This causes the already complex aperture distribution function d to change with time. For the sake of processing speed, we ac- countedforthisrotationonlyinstepsof5◦.Wealsotestedaless accurate stepping of 10◦, which was previously used to correct 2 datafromGASS(Kalberlaetal.2010),butdifferencesarebarely − noticeable.Weconcludethattheapproximationofthebeamro- tationwithin±2◦.5(EBHIS)or±5◦ (GASS)isaccurateenough tosolveEq.(7)forthenearsidelobeswithoutnoticeableuncer- tainties. Utilizing the four-fold symmetry of the antenna aper- 4 ture, in total 18 antenna diagrams for the offset feeds need to − 4 be provided during the solution of Eq. (7). As an example, we offset plotinFig.5thediagramforthecentralfeedandanoffsetfeed feed showingthecomalobeatapositionangleof60◦. For the correction algorithm, we averaged the previ- ously modeled side-lobe intensities on a fixed grid containing 2 2160 cells in cylindrical coordinates for the inner 4◦ of the an- tenna diagrams.The cellshave an azimuthalextent of5◦ and a radial extent of 0◦.125, covering the radial range of 0◦.25 (first minimum)upto4◦. g] e d 0 [ δ 3.5.4. Thefarsidelobes d The far side lobes are determined by details of the telescope structure. It is difficult to measure these side lobes (e.g., Higgs 2 1967;Hartsuijkeretal.1972),soweusedray-tracingtomodel − them. While in the near side-lobe range, details of each of the individualantennadiagramsneedtobeconsidered,anditissuf- ficienttouseasinglecommonfarside-lobediagramforallofthe receivers.Forthesubsequentdiscussionoffar-side-lobeeffects, 4 weconsiderthetelescopeasatransmittingsystem. − The most important side lobes that are far away from the -60 -40 -20 0 mainbeamofthe100-mtelescope,weredeterminedbyKalberla Pattern[dB] etal.(1980a,seetheirFig.3).Thesearethestraycones,caused byreflectionofplanewavesfromtheprimarymirroratthefeed Fig.5.Syntheticantennapatternswithinaradiusof4◦ forthecentral support legs. The second most important structure is the spill- beam(top)andanoffsetbeam(bottom)withacomalobeataposition over at the edge of the main reflector. Its irradiation level de- angleof60◦. pendsontheedgetaperoftheprimaryfeeds. In addition, we considered spherical waves originating in the feeds and determined their reflections at the support legs. direction makes the situation very uncomfortable since it can Figure 6 shows these structures. The eightfold symmetry is produce multiple reflections. A part of the rays being reflected causedbythefourlegs.Eachlegconsistsoffourmaintubes;two offtheapexroofcanhitfeedsupportlegslocatedeastorwestof of them, separated by 1.98 m, are visible from the feed. These the roof. This causes secondary stray cones. Only a fraction of sidelobeshaveaninterestingstructure,butitturnedoutthatthey thesesecondarystrayconesarereflectedontothesky.Theother areratherunimportant. part is reflected to the main mirror, causing caustics. A minor The Effelsberg 100-m telescope has a Gregorian secondary fractionofthesecondarystrayconescanhitthefenceattherim focus with several feed systems in an apex cabin. The seven- ofthereflectorbeforeescaping.Thesefeaturesweredetermined feedsystemissituatedinprimaryfocus,andtheroofoftheapex by ray-tracing. Figure 7 shows the resulting pattern. We disre- cabin is closed during EBHIS observations. This roof causes garded further reflections caused by support legs to the north reflections that are offset from the main axis by 40◦. The side andsouthofthefocuscabin. lobesareeasilydeterminedfromthegeometryoftheapexroof. Theapexlobesdiscussedherediffersignificantlyfromside- However, that these reflections are oriented in the east-west lobepropertiesoftheoriginaltelescopedesign.Initiallytheroof A41,page8of22 B.Winkeletal.:TheEffelsberg-BonnHSurvey:MilkyWaygas dα[deg] dα[deg] 20 10 0 10 20 50 0 50 − − − 20 50 10 g] g] e e d 0 d 0 [ [ δ δ d d 10 − 50 − 20 − -90 -85 -80 -75 -70 -73 -71 -69 -67 -65 Pattern[dB] Pattern[dB] Fig.6.Sidelobescausedbyreflectionsofsphericalwavesfromthefeed Fig.7.Sidelobescausedbyreflectionsfromtheroofofthesecondary atthesupportlegsofthe100-m(intensitiesrelativetomainbeam). focus cabin onto the feed support legs (intensities relative to main beam).Thelargeverticalfeatureisasecondarystraycone.Theinner lobes are caustics caused after reflections from two tubes of the feed of the apex cabin was designed so that reflections occurred in supportlegsbacktotheprimarymirror.Thefourspot-likesidelobes arecausedbythefencearoundtherimoftheparaboloid. four triangular lobes in between the support legs (see Fig. 3 of Kalberlaetal.1980a).Thenewconstructionallowsfastswitch- ingbetweenprimaryandsecondaryfocusbutcausesextremely complicatedside-lobestructuresfaroffthemainbeamthatcan of data on the sphere. Most important for our needs is that this onlypartlybemodeled.Wealsodonotaccountforminorcon- schemeallowsthespatialresolutionoftheT spectratobeeasily B struction details like stairs, cable ducts, and cross ties of the matched to the needs in accuracy during the correction. A sin- prime-focussupportlegs. gle parameter, N , is sufficient to define the resolution of T side B Theside-lobelevelsfromthespilloverlobeswereestimated (Górskietal.2005,Table1).HereN = 256correspondstoa side fromtheedgetaperofthereceiverfeed.Forthestrayconeswe sizeof13.(cid:48)7perpixel,whichinmostcasesisadequatefornear used previous results from Kalberla et al. (1980a), but in the side-lobecorrection.ForN =64,apixelsizeof55.(cid:48)0isgiven, side new software the radial side-lobe levels of the cones are mod- whichwouldbemoresuitableforfarside-lobecorrections. eled by a Gaussian (2◦ FWHM). All far side-lobe components Side-lobecorrectioncomputingspeeddependscriticallyon wereadjustedindividuallyinawaysimilartowhatisdescribed thenumberoftheprocessedT spectra.High-resolutionsingle- B byKalberlaetal.(2005),whenweweresearchingforaconsis- dishdataforthecorrectionofthefarsidelobesareonlyneeded tentsolutiontoEq.(7). inregionswithsignificantfluctuationinbrightnesstemperature. The HEALPix scheme allows a flexible scaling, and we have 3.6. Thecorrectionalgorithm chosen Nside = 128 for regions with column densities NH > 5×1021 cm−2,Nside = 64for1021 < NH < 5×1021 cm−2,and OurcorrectionalgorithmisbasedonKalberlaetal.(1980a)and Nside =32forNH <1021cm−2(seeFig.8). wasextendedlaterformultibeamsystems(Kalberlaetal.2010). A near side-lobe correction demands a high-resolution ap- Somefurtherimprovementsweremade,themostimportantones proximation of the true-sky brightness temperature T accord- B concerning the all-sky brightness temperature distribution, T , ing to Eq. (7). We initially used the LAB survey, resampled to B that is needed for the deconvolution according to Eq. (7). As N = 256andlaterto N = 512.AfterthispreliminarySR side side previouslymentioned,thebasicstrategywastoutilizetheLAB correction based on low-resolution data, we employed the SR- datasettocalculateaninitialguessoftheSRcorrection,i.e.to corrected EBHIS dataset itself (N = 1024, i.e., full-angular side solve the side-lobe-pattern integral in Eq. (7). After optimizing resolution, θ = 3.(cid:48)4). GASS data were supplemented on the pix free parameters in the antenna pattern for the LAB-based SR southern sky on the same HEALPix grid. The last step was re- correction,weswitchedovertouseEBHISitselfandGASSdata peatedseveraltimesinconjunctionwitharecomputationofthe fortheinputsky.Again,severaliterationswereprocessedtofind 2D baseline models (Sect. 3.2). Thus we increased the accu- theoptimalsolution. racy step-by-step for the near side-lobe correction. In addition Foradeconvolutiononthesphere,anequalpixel-arearepre- weimprovedtheaccuracyofthedeconvolvingkernelfunction: sentationofT isdesirable.HereweusedtheHierarchicalEqual foreachbeamfrominitially396tofinally2160samplesofthe B Area isoLatitude Pixelisation (HEALPix Górski et al. 2005) side-lobestructure.Thebeamrotationwastracedinitiallywithin scheme. HEALPix is a versatile structure for the pixelization ±5deg,finallyaccurateto±2.5deg. A41,page9of22 A&A585,A41(2016) InFig.9(leftpanel)weshowthecomparisonoftotalcolumn 60◦ densities between both surveys. The relation between the two 135◦ 30◦90◦ 45◦ ◦0 315◦ 270◦ 225◦ imsoadlmelofistttpoetrhfeecdtalytaa: one-to-one correlation, based on a linear 0◦ NHEBHIS =1.0023(4)·NHLAB−0.3(3)×1018cm−2. (8) To account for both errors in x and y, orthogonal distance re- −30◦ gression(Boggs&Rogers1990)isused(providedbytheSciPy −60◦ 122290122log(N[cm])−HI merroodrusl,euosdedr;aJsonweesigethatsl.f2o0r0t1h)e.rPergorpeesrsisotna,tiastriecafoluconldumusnindgenersriotyr propagationofthefrequency-dependentrmsaccumulatedinthe Fig.8.ColumndensitydistributionfromtheLABsurvey.Theisophotes columndensityintegral. indicateNH=1021cm−2andNH=5×1021cm−2,limitingrangeswith Likewise,thecenterpanelofFig.9displaystherelationship different HEALPix resolutions that were used for far-side-lobe stray of brightness temperatures between EBHIS and LAB. Again, a radiationcorrection. one-to-onecorrelationisfound: TEBHIS =1.0000(6)·TLAB+12.2(5)mK. (9) Alltheseimprovementsinthecorrectionprocedureledtoa B B more precise and efficient algorithm but eventually did not im- By computing the intensity-weighted velocity (Moment-1) for provethequalityofthesolution,asdesired.Becausewecannot each spectrum, it is possible to test the correctness of the fre- modelallparametersoftheantennadiagramwithsufficientde- quency/velocity scale (see Fig. 9, right panel). The median ve- tail,residualSRfeaturesappearinthedata.Unaccountedforare locitymismatchbetweenEBHISandLABis0.07kms−1.This substructuresofthefeedsupportlegssuchasstairwaysandca- indicatesawell-behavedrelationship.Foraninsignificantnum- blechannels.Incalculatingcaustics(Fig.7),weonlytookthree berofpositions,weobserveoutliers.Weattributethistoresidual scatteringsurfacesintoaccountanddisregardedfurtherscatter- RFIandotherartifactsineitherofthedatasets,whichcanhave ingatthefeedsupportlegs.Alsosomesubstructuresoftheroof astrongimpactontheMoment-1calculation.Weminimizethe of the apex cabin were ignored. We did not attempt to model potentialimpactofartifactsinthedataontheMoment-1calcu- reflectionsfromthesubreflectormount. lationbyapplyingamaskwithathresholdof2K. In AppendixA.1 we show maps of the near-side-lobe, far- While the two surveys agree extremely well on average, it side-lobe, and total SR correction (in H column densities, hastobenotedthatthebrightness-temperaturescatteraboutthe Fig.A.2),whichwasappliedtotheEBHISdata.Toallowcom- linearrelationisrelativelyhigh.WeattributethistotheLABsur- parison,thethreefiguressharethesamecolorbarscaleasFig.1. vey because a comparison of EBHIS with GASS reveals much Itisremarkablethatthefar-side-lobecorrectioninseverallow- lowerscatter;seeAppendixA.2,whereweprovidecomparison column density regions at higher Galactic latitudes is greater plots for EBHIS, GASS, and LAB computed for the common than the reconstructed column densities in these fields, which data slice (−4.5◦ ≤ δ ≤ −0.2◦). Unfortunately, we could not underlinestheimportanceofSRcorrection. find the cause of this enhanced scatter. One well-known prob- lemoftheLABsurveyisthatitisnotfullysampledinthespatial domain. But since we compare spectra on the original pointing 4. Dataquality positions,theangularsamplingisirrelevanthere.Forcomplete- In this section we assess the data quality of EBHIS in detail. ness,wealsoshowsimilarcomparisonplotsforGASSvs.LAB As with many other large-scale single dish surveys, there are onthesouthernhemisphereinAppendixA.2. stillresidualartifactsvisibleinthedata.Oneimportantreasonis Anotherimportantfinding,whichwemadeduringourtests, the sheer amount of data that makes automatic processing nec- isthatEBHISandGASS(seconddatarelease),aswellasLDS essary. In Winkel et al. (2010) and Sect. 3, we presented how (thenorthernpartofLAB)andGASS,showedadiscrepancyin many problems with the raw data had to be dealt with. In gen- calibrationbyabout4%.Ontheotherhand,theIARcontribution eral the developed data reduction software proved to be very toLAB(δ≤−30◦)matchestheGASSflux-densityscale,which robust. In the following we comprehensively discuss the cases means that there is an inherent inconsistency in the LAB data where our data processing pipeline reaches its limits. This is betweenthenorthernandsouthernparts.Thisfindingledtoare- notbecauseweconsiderthefinaldataproductslackinginqual- visedintensitycalibrationscaleforthethirdGASSdatarelease ity compared to, for instance, GASS and LAB, but instead we (Kalberla&Haud2015).TherescaledGASSisveryconsistent wantthepotentialusersofEBHIStobeabletoworkefficiently with EBHIS and LDS in terms of column densities and bright- andbewell-informedwiththedata.Furthermore,wethoroughly ness temperatures (see Figs. A.3 and A.4). As a consequence, cross-checkedtheEBHISintensitycalibrationagainstLABand however,theIARpartofLABbecameinconsistentwithGASS GASS. (Fig.A.5). 4.1. Calibration:consistencycheckvs.LABandGASS 4.2. RFI Tohaveanindependentconsistencycheckoftheoverallbright- AccordingtoSect.3.1.1,theautomaticRFIflaggerisabletofind nesstemperaturecalibration,baselinesolution,andresidualSR, evenfaintnarrow-bandinterference.Asdescribedearlier,weat- wecomparedtheEBHISdatawiththeLABsurvey.Toavoiddis- tempt to subtract near-constant-amplitude narrow-band events tortionoftheresultsbygriddingeffectsandangularundersam- fromthedatatominimizelossofsamplescausedbytheflagging. plingoftheskyinLAB,wetooktheoriginalpointingpositions Thissubtractionisdonesubscan-wiseperreceiverfeedandpo- ofLABandcomputedaweightedaverageofEBHISdumpssur- larizationchannel.Foreachflaggedchannelintheaveragespec- roundingtheassociatedpositions. trum,wecalculatedthedifferencetothemeanvalueofadjacent A41,page10of22
Description:The list of books you might like
