ALMA Memo 599 The ALMA Calibrator Database I: Measurements taken during the commissioning phase of ALMA T.A.vanKempen1,2,R.Kneissl1,N.Marcelino3,E.B.Fomalont3,1,D.Barkats1,S.A. Corder1,3,R.Lucas1,4,A.B.Peck1,3,andR.Hills1,5 1JointALMAObservatory(JAO),AlonsodeCo´rdova3107,Vitacura,Santiago,Chile 2AllegroARCnode,LeidenObservatory,NielsBohrweg2,Leiden,theNetherlands∗ 3NationalRadioAstronomyObservatory,(NRAO),520EdgemontRoad, Charlottesville,VA,USA 4InstitutdePlane´tologieetd’AstrophysiquedeGrenoble(IPAG),414RuedelaPiscine, St-Martind’He´res,France 5CavendishLaboratory,19JJThomsonAve,CambridgeCB30HE,UnitedKingdom Submitted: December2,2014 Abstract InterferometricALMAobservationsrelyuponaccurateandtrustworthycalibrationoftheinstrumen- talandatmosphericphaseandamplitudevariationsinordertoproducehighqualityscientificdata.Short observations(scans)ofradiosourcesattheappropriatecadencesareincludedinobservingprogramsin ordertodeterminethesevariationswithtimeandfrequency. Quasarsareamongthebestradiosource candidatesbecausetheiremissionisbright,lackspectralfeaturesandarepoint-like;typicallytheymea- sure less than 0.01” in size. Since quasar flux densities vary both in time and frequency, specialized ALMAobservationsareneededtofindthequasarsthataresufficientlybrightbetween90and900GHz, theALMAfrequencyrange. Oncefound,theappropriatequasarfluxdensitiesmustbemonitoredwith periodic ALMA observations to track. Hence, in 2011, the ALMA calibrator database was initiated duringthecommissioningandscienceverificationstageofALMA.Thismemodescribesthedatabase structureandcontents,theobservationalanddatareductionstrategy,anddiscussesmanypropertiesof theradiosourcesobservedtothisdate.Duringthecommissioningphase,twomajorobservingprograms were started: First, a wide survey of about 600 quasars was carried out with the aim to obtain a suf- ficiently large sample of sources to be used as phase reference sources. Second, a selection of about 45brightandrelativelystablequasars,calledgridsources,weremonitoredregularly. Thesamplewas distributedrelativelyequallyoverthesky.Thesearecandidatesforbandpasscalibrations,buthavesince also evolved into secondary amplitude calibrator candidates. Both programs have continued into the operationalphaseofALMA. ∗contacte-mail:kempenatstrw.leidenuniv.nl 1 Introduction A radio wave in its path through to atmosphere above each antenna to the correlator varies in gain and phase (delay) over a variety of time-scales. The changes must be calibrated before ALMA can produce scientifically useful data, often in the form of images or spectral cubes. Gain and phase changes in the electronicsgenerallyhaveatimescaleofmanyminutestohours. Shorter-termvariationsaredominated bytheatmosphericdelayandabsorptionoftheradiosignalaboveeachantenna,introducedbyatmospheric turbulence. Phasefluctuationsattheshortesttime-scales(<2min)canbecorrectedusing181-185GHzmeasure- mentsfromradiometersoneachoftheantennas. Thefluctuationsinthemeasuredwatervaporemission alongthelineofsightcanbeconvertedintodelaychangesoftheincomingsignal(SeeALMAMemo593, Nikolicetal.,2012). The variations on timescales between these two cases (∼4 min to half an hour) can be calibrated by switchingobservationsbetweenanearbycalibratorandthesciencetarget. Quasi-stellar objects (often referred to as QSO’s or quasars) provide the majority of calibrator radio sourcesbecausetheyarebrightatradiowavelengthsandhavepoint-likeemission. Quasarsarefoundinall regionsintheskyandareobservablethroughtheGalacticdisk. Essentially,quasarsprovide”test”signals relativelycloseintheskytothescientifictarget. Whenobservationsarealternatedbetweenthetwo, any changes in the gain and phase of the quasar signals are transferable to the target data, largely removing variationsintroducedbytheatmosphere. Assuch,itisessentialtoobtaintheparametersofquasarsacross theskybeforecarryingoutthescientificobservations. At radio frequencies below 20 GHz the spectral index, α where flux ∝ να, lies between +0.5 to - 0.7. At the typical ALMA frequency of 300 GHz, the spectral index is more constrained, ranging from -0.2to-1.0. Theemissionmechanismissynchrotronandthespectralindexisassociatedwiththepower distribution of the radiating electrons and with the optical depth in regions of the radio source. Planck results revealed another clear difference between the spectral slope from 300 GHz and that above 1000 GHz (Planck Collaboration et al., 2011a), likely caused by the increase in thermal emission at higher frequencies. Foragooddiscussiononthenatureofspectralslopesathighradiofrequenciesaswellasthe mostrecentobservationalevidence,seePlanckCollaborationetal.(2011b,a,2013). Duetointrinsicphysicalchangesofquasar,observedfluxdensitiesvaryintimeandfrequency.Typical flux density changes are about 10% over a month. However, some quasars exhibit much larger changes, perhaps by a factor of two or three over a month or two. In a handful of quasars changes of an order of magnitude have been observed. Changes of up to 5% on a daily basis are also known. Quasar emission does not show periodic changes, and the most extremely variable quasars tend to stay very variable for manyyears. AttheALMAresolutionof 0.1”,thevariablequasarfluxdensitiesremainpoint-like,because theemissionchangesoccurwithinaregionofabout0.005”ofthenucleus. The use of quasars as calibrator sources has motivated other radio observations over the decades to carry out monitoring programs which are contained in catalogs, many of which are publicly available. ExamplesaretheOVRO(Richardsetal.,2011),AT20(Massardietal.,2011),VLA(Patnaiketal.,1992), VLBA(Beasleyetal.,2002;Fomalontetal.,2003),SMA(Gurwelletal.,2007)andrecentPlanckcatalogs (Planck Collaboration et al., 2011b,a, 2013). One of the oldest quasar catalogs is the Third Cambridge Catalog (Edge et al., 1959). Due to the available southern declinations at ALMA, the AT20G (Massardi etal.,2011)isparticularlyinteresting,andcontainsmanypotentialALMAcalibrators. SinceALMAobservesatamuchhigherfrequencythanisavailablefrommostoftheabovecatalogs, excepting the extensive catalogs of the SMA and PdB 1 (Gurwell et al., 2007), but both located on the northern hemisphere, there was strong motivation to start an ALMA monitoring program and database construction suitable for (sub)millimeter calibration. It was also important to measure the spectral index of these sources in order to estimate their flux densities at frequencies over the entire ALMA frequency range,inparticulartofrequenciesabove300GHzwherelittletonoinformationexists. The first part of this memo describes the ALMA’s calibrator database. It then outlines the two main observing programs: 1) the database filling of quasars over as much of the observable sky as possible, 2) the observations and monitoring of about 45 bright and/or stable quasars at 95 and 350 GHz. Fainter quasars can be used as phase references of science targets, while the brighter sample are used for more stablebandpasscalibrations,especiallythosewherestrongsignalsareneededinnarrow-bandchannels. 1seee.g.http://sma1.sma.hawaii.edu/callist/callist.html 2 Figure1: Screenshotoftheweb-basedinterfacetothecalibratordatabase. Thefirsttabcontainsthequery parameters. The ALMA calibrator database Inordertocompileandhaveavailableinformationaboutquasarcalibrators,adedicateddatabasesystem wasdesigned. TheALMAcalibratordatabaseisaread/writedatabase. Ithasthecapabilitytobothaccept newentriesandmodifyexistingones. Entriesaredenotedasmeasurements. OnlyALMAstaffmembers areabletoenterandmodifymeasurements. Asareference,alistofover5,000calibratorcandidateswasmade,filledmainlyfromlowfrequency catalogs (VLA, SMA, CRATES, AT20), with the CRATES compilation being the main contributor. The inclusionoftheVLAcatalogandtheSMAcatalogcontributionsarespecificallyaimedathigherfrequen- cies. Sincemanyofthesourceswereprobablytooweaktobeusefulevenat90GHz,anALMAobserving databasefillingprogrambeganin2011toidentifythesuitablystrongcalibratorsofthislist. Theseobser- vationsaredescribedinthenextsection. Anotherproblemwastheuseofinaccuratepositionsforsomeof the sources in the catalog. To remedy this, VLBI-quality positions, often accurate to 0.001”, from ICRF andrelatedcatalogsweresubstituted,whereavailable. Finallymanysourcesinthe20GHzAT20gcatalog were included to fill the southern sky (dec < -40◦) with candidate sources. Note that few VLBI-quality positionswereavailableinthesouthernsky. Most catalog updates are now done using flux densities measured regularly with ALMA. Measure- ments of the degree and angle of the linear polarization have also started, and occasionally an improved sourcepositionisadded. NearlyallofthequasarsthatareusedarepointsourceswithALMAresolution, so that detailedstructural information is notneeded. Afew sources with faint large-scalestructure (e.g., 3c273)arenotedinthecatalog.Suchsourcesarestillgoodcalibratorsexceptattheverylowestresolutions. WhenALMAbeginshigherresolutionobservations,thecontributionofthermalemission,expectedtobe somewhatextended,morestructureinformationandimageswillbeprovided. Thedatabaseisaccesiblethrough: http://almascience.eso.org/sourcecatweb/search The query form is on the first tab, as shown in Fig. 1. By position, the database can be queried on Source name or coordinates. A radius in degrees can be given. In addition, constraints can be put in by energy,bothinfrequencyandinfluxdensity,aswellastime,expressedbyastartandenddate.Afterquery, theresultsarefoundinthesecondandthirdtabs,providingthenumbersandagraphicalrepresentation(See Fig. 2and3 IfacalibratorisdesignatedintheSourcename,theninformationaboutthissourceisgivenintheResult TableandtheResultPlot. Ifanon-calibratorsourceisspecified,thenallALMAcatalogcalibratorswithin theSearchRadiuswillbelistedintheTableandPlot. Mostcalibratorsdohaveseveralnames. TheofficialALMAnamedesignationhastheformJHHMM- DDMM for all objects using the J2000 coordinate system. However, the catalog contains major aliases 3 Figure 2: Screenshot of the web-based interface to the calibrator database. The second tab provides the resultingsourcesproducedbythequery. Figure3: Screenshotoftheweb-basedinterfacetothecalibratordatabase. Thethirdtabshowsagraphical interface. Totheleft,resultsfoundwithinthedefinedradiusareshown. Ifanindividualsourceisselected, thefluxdensityhistoryisplottedtotheright. 4 forallsourcessoselectionbyothercommonnamesisavailable. Forexample, therecognizednamesfor J1229+0203are: J1229+020,3C273,3c273,B1226+023. ForJ1325-4301: B1322-427,J1325-430,cena. Itisadvisedtocheckalternativenames. Somesources(e.g. CentaurusAascena)areonlyincludedtobe usedascalibratorsifcertaincontrolledconditionsaremet. After querying the database, results are provided in a table in the second tab. The table is produced with the source names, right ascension, declination, separation from the queried coordinates, observed frequency and its associated ALMA Band, flux density, flux density error uvmax, uvmin and the date at whichthiswasobtained. ThistabisavailablefordownloadinCVSandJSONformats. Thelasttabshows thespatialdistributionofcalibratorsaroundtheprovidedlocationuptothechosenradius. Thepointsize of the calibratos is scaled to the brightness, while the color shows the highest ALMA Band within the querydefinitionswhereameasurementexists. SeeFig. 2foranexample. Ifacalibratorisselected, the fluxdensityhistoryofthatsourceinallALMABandsdefinedduringthequery. Ifnoconstraintswereput in,thefluxdensityhistoryofallALMABandsisplotted. 5 Database Filling ForadequatecalibrationsofALMAobservations,atleasttwotypesofcalibratorsareneeded. First,strong quasars are used to determine the bandpass (receiver response over the range of observed frequencies) of each antenna system. These does not need to be close to the science target, but are required to be as bright as possible at the targeted bandwidth and frequency. In practice, a sufficient number of these bright calibrators are needed so that at least one of them has elevation above about 40◦ at any sidereal time. Asecondsampleofcalibratorscanbeweakerbutmustbemuchmorenumerousinordertohavea highprobabilityoffindingasuitablecandidateclosetoanypositioninthesky,thuscoveringallpotential science targets. Their strength needs to be sufficiently large to determine the amplitude and gain at each antennatoabout3%ingainor5◦inphasewithaboutaoneminutescan. Notethattheserequirementsare slightlylessstrictatobservationsabove400GHz. The initial observations in determining these calibrators was started during the Commissioning and ScienceVerification(CSV)phaseofALMA.Thisworkhascontinuedduringthescienceoperationsphase andwillcontinueformanymoreyears. Othercalibratortypes, forexamplepolarizationcalibrators, will bedefinedandobservedwhenneeded. Grid Quasarsample Acandidatebrightquasarsamplewascollectedfrompreviouscatalogs(existingcatalogsfromtheSMA, ATCA, VLA; Beasley et al. (2002); Gurwell et al. (2007); Massardi et al. (2011)), searching for those estimated to be stronger than 1 Jy at 100 GHz and a declination below +45◦. An assessment of their variability was also made. Note that some of these quasars are located at northern latitudes (δ > 20◦) andarethusonlyavailableforabrieftimeeachday. Thefinallistof45quasarswaschosentocoverthe sky relatively uniformly and includes those both with the most stable emission, as well as the strongest semi-regular flaring quasars. This set of sources was named the Grid sample, and is primarily aimed at bandpasscalibrationswheregoodsignal-to-noiseisneededforaccuratenarrow-channelreceiverresponse calibration. TheGridsourceswerealsoextensivelyusedforalargenumberofothercommissioningtasks (e.g.,receiverstabilitytesting,pointingmodels,bandpassstability,etc). TheuseoftheGridsampleassecondaryabsolutefluxdensitycalibratorsevolvedoverthe2011-2013 period because of the difficulty in using solar system objects (SSO). Although SSOs possess relatively accurate flux density models, a suitable SSO is often not available during an observations, or at a high frequency and resolution, all available SSOs are too resolved to calibrate longer baseline data. Thus, monitoringthegridsampleonasemi-regularbasiswasinitiatedinordertoestimatetheirfluxdensityat anytimeandfrequencytoa10%level. The main goals of the monitoring were: (1) to quantify variability of the brightest quasars at various ALMAfrequencies;(2)toidentifythebestcandidatesforbandpasscalibrations;(3)toprovideuptodate flux densities necessary for commissioning and other start-up calibrations and (4) investigate their use as secondary flux calibrators. The sample was monitored as often as technical, weather and other time constraintsallowedduringnight-time(Localtime16:00-08:00),complementedbyday-timeobservations, if possible given technical constraints. During the more recent science operation phase, the first night in eachblock(bothscienceandCSV)hasbeenreservedtoperformmonitoringaspartoftheregressiontests. Givennight-timeavailabilityofasource,theaimwastoobtainaminimumofonemeasurementatBand 3 every two weeks and one measurement at Bands 6 and/or 7 every two weeks. Observations at Band 7werepreferred. However,Band6measurementswereobtainedifweatherconstraintsonsubmillimeter observationswereinplace(seethenextsectionormemo2foramoredetaileddescriptionofthemonitoring strategy). The monitoring program was initiated already in June 2010, but regular observations started in June 2011.Thirty-oneofthe45sourcesweresuccessfullymonitoredatbi-weeklyintervals,whiletheremaining were monitored at longer intervals up to 5 months. Most sources north of 20◦ were among those not monitoredregularly.Thenumberofantennas,thearrayconfigurationandweatherconditionsvariedgreatly over the course of the monitoring period. However, with as few as five antennas, accurate flux density measurementscouldbeobtained. 6 Table1: SpectralWindowFrequenciesforcalibratormeasurementsinBands3,6and7. Receiver Frequency DatabaseFrequency Sensitivity1 LSB1 LSB2 USB1 USB2 LSB USB [GHz] [GHz] [GHz] [GHz] [GHz] [GHz] [mJy/min] 3 97.5 99.0 109.0 110.5 98.25 109.25 2.0 6 221.5 220.32 234.5 236 228.0752 3.3 7 336.5 338.0 348.5 350.0 343.252 5.2 1Undertypicalweatherconditionsforthatband. 2DatabasefrequenciesareaveragedforBands6and7. WideQuasarsample A large number of calibrators are needed in order to calibrate the shorter term gain and phase changes, whichrelyonaseparationtothescientifictargetaslittleaspossible.Inordertofindoneofthesecalibrators withinafewdegreesofanyplaceintheALMA-visiblesky,morethan1000sources,evenlydistributed,are required. Extensivelistsofcalibratorsourceswereavailable,mainlyfromlowfrequencysurveyslessthan 43 GHz (VLA, at20g), with some high frequency data from the SMA and Planck (Planck Collaboration etal.,2011b). Thecandidatelistwasconstructedusingtheabovedatabasestofindthosethatmighthaveafluxdensity greaterthan1.0Jyat95GHz. Afirstsampleofabout600candidatesourceswereselectedbasedontheir strengthandpositioninthesky. Additionalcandidateswereaddedwithin10◦ofknowntargetsthatwould beobservedintheALMACycle0andCycle1projects. Thisentiresampleofsourcesisdenotedasthe wide quasar sample. It is not meant to be complete in any respect, and it is continually being enlarged sinceitisimportanttouseaquasarcalibratorthatisasclosetothetargetaspossible. Themaincriterion isthatthecalibratorbedetectableatthe10-sigmalevelperantennagaindeterminationforaone-minute integration. Between August 2011 and August 2013, the flux densities of these quasars were observed and the resultsplacedintotheALMAcatalog.Observationstoenlargethewidequasarsamplehavecontinuedinto theoperationalphaseofALMAandwillbeprogressingforthenextseveralyears. Inaddition,monitoring thesesourcesisbeingperformedtoupdatefluxdensitiesolderthan1year. At95GHz,usinganarraywith 32 12-m antennas and a 2-GHz bandwidth, both polarizations, an acceptable flux density level is 0.1 Jy. Thereareanestimated7,000sourcesintheskythatmeetthiscriterion. Thegoalistoprovidesufficiently strongcalibratorsevery3-4degreesforanypointinthesky. ALMAobservationsoveralargeareaofsky, andtargetedsearchesnearpotentialALMAtargetswillbecontinuouslyperformedduringtheoperational phase. 7 Observational Strategy and Data Reduction Methods ObservingFrequencies Thefluxdensitymeasurementsforthegridsourcesweremadeatatleasttwofrequenciesinordertodeter- minethespectralslopeofeachsourcesothatitsfluxdensitycouldbeestimatedatallALMAfrequencies. ThelowestfrequencyBand3(95GHz)isanobviouschoicewhere: (1)quasarswithanegativespectral index are strongest; (2) The ALMA sensitivity is greatest; (3) the weather effects are smallest; (4) most instrumentaleffectsarestable.Thesecondfrequencychosenwasband7(350GHz)whichprovidesalarge frequencyleverarmfromband3. ALMAsensitivityisstillgood, andweatherconditionsaresuitableat theALMAsitemorethan40%ofthetime. Whentimepermittedoriftheweatherwasnotoptimum,Band 6(230GHz)observationswerecarriedout. Theobservingset-upsandotherobservingparametersforthe threefrequenciesaregiveninTable1.Alsoincludedisthermssensitivityfora1minutescanateachofthe threebandsundertypicalconditionswith12antennas. Allfour2-GHzbasebands(twoineachsideband) were used during all observations. Observations were done in full polarization (XX, YY, XY and YX) modeinpreparationforthepolarizationcapabilitiesofALMA.Thisconfigurationgives64channelsover eachofthefour2GHzbandwidthandtwopolarizations. The number of available antennas and their locations were secondary for both types of observations, with five antennas the minimum number. Typically, 10 to 16 antennas were included. Because virtually all of the quasars are point sources, any configuration was satisfactory from the 20-m ACA baselines to the 1000-m ALMA baselines. Only when solar system objects were observed (see below), did the configuration play an important role. Some quasars do show structure. This will be discussed in more detailinthedatareductionsection. ObservationStrategy For the Grid sample, the sources were separated into six observing groups spread over about four hours in right ascension. Each group contained from 8 to 10 quasars. Consecutive groups had one quasar in common for redundancy. The six groups were made at Band 3, 6 and 7 and contained the same source selection. Eachgroup/frequencyobservationiscalledasessionwhichlastsfrom45minto60min. Each sourcewasobservedonce. Eachsessioncontainedthefollowingscans: 1. Focusandpointingoffsetsweredeterminedatthebeginningofeachsessionweremadeatallbands. Onlyforbands6and7,pointingoffsetsweredeterminedforeachsourcescan. Atband3thiswas notnecessary. 2. Integration times were longer at the higher frequencies because of the relative weakness of most quasarsandthelesssensitivesystem. 3. Thesystemtemperaturewasdeterminedatthebeginningofeachscan. Thecalibrationisimportant inremovingthereceiverandskycontributionstothesystemsensitivityandtheamplitudelevels. 4. No elevation correction was made for any dish efficiency change. Even at band 7, the estimated efficiencyis<3%at30◦elevation. Inordertodeterminetheabsolutefluxdensityofthegridquasars,asolarsystemobjectwithaknown fluxdensitymodelwasincludedasthefirstscan,whenpossible. ThelistofSSO’sthathavemodelsgood to10%andarenottoolargeinangularsizeare: Neptune, Uranus, Titan, Callisto, Ganymede, Marsand Pallas. Pointing, focus and system temperature measurements were made at the beginning of each SSO scan. Aboutone-thirdofthetime,noSSOwasavailable(inmostcases,SSOsweretooresolvedtobeused or too close to the planet in case of the Jovian moons or Titan) and grid observation would not be run. However, asmoreconfidencewasobtainedinthefluxdensityvaluesmeasuredinpreviousobservations, thegridobservationswererunwithnoSSO,andafluxdensityforoneofthequasarswasassumedfromthe previousrun(s). Theuncertaintiesintheabsolutefluxdensityscalefromthishybridapproacharediscuss inthismemoandinmemo2. 8 Thewidequasarobservationshadsimilarobservingstructureandparameterstothegridobservations. Themajorityofthemwererunonlyatband3. ThesessionsbeganwithanSSOobjectandagridquasar thathadawell-determinedfluxdensity. IfaSSOobjectwasnotavailable,thegridcalibratorprovidedthe absoluteamplitudescale. Theremainderofthesessionconsistedof4to12objectsthatwerewithinabout 30◦ of the grid calibrator. The sources observed were either data filling of potential calibrators over this partofthesky,orthecheckingofparticularcalibratorcandidatesthatwereclosetotargetsthatweregoing tobeobservedbyALMA. DataReduction/Pipeline Each session was calibrated, edited, imaged and analyzed independently, using an up-to-date version of CASA (Jaeger, 2008). The results were generally available (inserted into the ALMA catalog) from one day to one week after the session. This fast turn around was important in order for the reduction of the earlysciencedatatohavethefluxdensityofthegridsourceifitwasneededasthefluxcalibratorforthe observation. The initial reduction script developed into a semi-automatic pipeline with python scripts connecting theCASAtasks. Manydisplaysandtablesweregeneratedinordertoassesstheproblemsandqualityof thedata. AdescriptionofthispipelineisontheTwikioftheALMADepartmentofScienceOperations: https://wikis.alma.cl/bin/view/DSO/HowToReduceCalSurveyData. Themainstepsinthereductionsareas follows: 1. Convertthearchivedataintoameasurementset. 2. Applyaprioricalibrationsandflagging. 3. Chooseonequasarasthemaincalibratoranddeterminebandpassandamplitudescale,assumingthe quasaris1.0Jy 4. Averagethe64frequencychannelsineachspw 5. phaseself-calibrateeachscantoremovetroposphericphasefluctuations 6. Averagethescanforeachbaseline 7. IfaSSOwasobserved,theratioofitscalibratedvisibility(basedonthe1-Jycalibrator)toitsmodel visibilitygivesthescalingfactorneededforeachspw/pol.OnlySSOvisibilitydatabeforethemodel visibilitydroppedbelow20%duetoresolutioneffectswereused. ForveryresolvedSSO,perhaps only3to5baselineswereavailable. 8. IfnoSSOobjectwasobserved,thenthefluxdensityscaleisdeterminedbythefluxdensityassumed forthemaincalibratorfromadifferentorigin. 9. The flux densities of all of the sources were then determined from the average of the corrected baselinevisibilityamplitudes. 10. The measurement error was obtained as follows :First, the rms flux error per baseline was deter- mined. Thiserrorwasthendividedbythesquare-rootofthenumberofantennasusedforthesource. Regardlessoftheaboveresults,alowerlimitfluxdensityerrorof3%atband3and5%atbands6 and7ofthetotalfluxwasapplied. Adetaileddiscussionabouttheerrorestimatesisgivenbelow. 11. Anestimateofthepercentagepolarizationwasdetermineforeachsourcebythecombinationofthe √ fourvisibilityamplitudecorrelations: ((XX−YY)2+(XY+YX)2).Theonlypolarizationcalibration wastoremovethedelaybetweentheXXandYYcorrelationssothattheXYandYXvalueswere coherentovereachspw. 12. ForBand3,afluxanderrorestimatewasobtainedforthelowersidebandanduppersidebandsepa- ratelysincethefrequencydifferenceis10%. Allfourspwwereaverageforband6andband7. Both linearpolarizationswerecombined. 9 Thetypicaloutputfromthepipelineisshownbelow.ItcanbeingesteddirectlyintotheALMAcatalog using XMLRPC scripts. It has the following entries: Source name, RA, RA error, Dec, Dec error, Cen- ter frequency, flux density, statistical flux density error (see below on assessment of flux density errors), percentageofpolarization,erroronpolarizationpercentage,angles,structureintheformofUVMINand UVMAX, and date of observation. ’NE’ means No Entry. For some sources, more commonly accepted namesareusedmainlyforhistoricalreasons.TheJ2000namesandaccuratepositionsarecorrectlyassoci- atedwiththecommonnamesinthedatabase. UVMIN=0.0meansthatnolarge-scaleemissionhasbeen detectedyet. UVMAX=-XXmeansthatatabaselinelengthofXXwavelengths, noresolution. Notice the 3c273 has large-scale structure which is seen in the visibility data when less than 20 kλ is observed. PositionsarecurrentlylabelledasNE.Itispossibletoderivepositionsofquasarsinthefuture,ifrequired. ************************************************************************************************* SOURCE RA RAE DEC DECE FREQ FLUX FLUX_E %POL %PERR PANG UVMIN UVMAX DATE OBS J1037-295, NE, NE, NE, NE, 343.48E+09, 0.56, 0.04, 0.06, 0.03, NE, NE, 0.0, -1470.1, 2013-12-11 J1058+015, NE, NE, NE, NE, 343.48E+09, 1.42, 0.10, 0.07, 0.04, NE, NE, 0.0, -1367.3, 2013-12-11 J1107-448, NE, NE, NE, NE, 343.48E+09, 0.63, 0.05, 0.04, 0.03, NE, NE, 0.0, -1466.8, 2013-12-11 J1146+399, NE, NE, NE, NE, 343.48E+09, 0.80, 0.07, 0.06, 0.05, NE, NE, 0.0, -1320.7, 2013-12-11 3c273, NE, NE, NE, NE, 343.48E+09, 1.91, 0.14, 0.02, 0.02, NE, NE, 20.0, -1196.6, 2013-12-11 3c279, NE, NE, NE, NE, 343.48E+09, 7.77, 0.35, 0.09, 0.03, NE, NE, 0.0, -1160.7, 2013-12-11 J0854+201, NE, NE, NE, NE, 343.48E+09, 1.51, 0.08, 0.05, 0.03, NE, NE, 0.0, -1470.4, 2013-12-11 J0635-751, NE, NE, NE, NE, 343.48E+09, 0.36, 0.05, 0.08, 0.08, NE, NE, 0.0, -1327.4, 2013-12-11 ************************************************************************************************* BecausemostGridandWidecalibrationsessionshave15ormoreantennasandpointsourcesarebeing observed, very strict standards in consistency can be applied in order to remove data that is marginally suspect. Afterthefirstpassthroughthepipeline,therearemanyplotsandtablesfromwhichoutlierscan bedetected.Thestrongestcomparisonisthederivedfluxdensityforeachsourceforeachantenna/spw/pol. For a point source, the flux density should be the same and any errant antenna can be easily spotted and removed. Examples: Ifthepointingisnotoptimumforanantenna,thentherelativegainsoftheantenna foreachsourcewillvarymorethantheotherantennas;Ifthetsysofanantennaisanomaloushighormuch morevariablewithelevationthanotherantennas. Becauseoftheredundancyofthedatainobtainingflux densitiesofpointsources,flaggingsuchmarginaldatawillproducemoreaccurateresults. Sourcestructure Most sessions have sufficient u-v coverage, even with one scan, to determine any large or small scale structure associated with the quasar. Because even a small amount of contaminating data or large phase fluctuationscansuggestfalsesourcestructure,severalmethodsareusedtodeterminenon-pointlikeemis- sion of the quasar. These are judged by the reducer and resolution effects are manually inserted into the ALMA data base. At the present time only about 10 sources out of about 500 well-studied sources are sufficientlynon-pointlike. Eventheseaffectthecalibrationsatthe3%and5◦level. Thethreemethodstodetermineifthereissourcestructureare: 1. Thepipelinedeterminesthefluxdensityandestimatederrorofeachsourceinthreeu-vranges:flux- short,flux-medandflux-long.Ifflux-longislessthan85%oftheaverageoftheothertwo,awarning statementisgivenintheformabovethatthesourcemayberesolvedatlongbaselines 2. Iftheflux-shortrmserrorissignificantlylargerthantheothertwoerrors,thenawarningstatement isgivenintheformabovethatthesourcemayhavelarge-scalestructure. Sincelarge-scalestructure tends to produce fluctuating correlated flux densities at the short baselines (rather than a simple increaseincorrelatedfluxdensity),thiscriterionisused. 3. Images are routinely made of all observed sources. Large-scale structure is often obvious in the imagedisplay. 4. For a source that is suspected to be resolved at long baselines, the plot of correlated flux density versusuv-distanceshouldconfirmtheresolutionsuggestedbythepipeline. Often,thelossoflong baselinefluxisacoherenceproblemwhenthephasefluctuationsarelarge. 10