Mon.Not.R.Astron.Soc.000,000–000 (2008) Printed7May2008 (MNLATEXstylefilev2.2) α GHASP : An H kinematic survey of spiral and irregular α galaxies - VI. New H data cubes for 108 galaxies. Epinat, B.1, Amram, P.1, Marcelin M.1, Balkowski C.2, Daigle, O.1,3, 8 0 Hernandez, O.3,1, Chemin, L.2, Carignan, C.3, Gach, J.-L.1, Balard, P.1 0 1Laboratoire d’Astrophysique de Marseille, OAMP, Universit´e de Provence & CNRS, 2 Place Le Verrier, 13248 Marseille Cedex 04 France 2 2GEPI, Observatoire de Paris-Meudon, Universit´e Paris VII, 5 Place Jules Janssen, 92195 Meudon, France. y 3LAEet Observatoire du mont M´egantic, Universit´e de Montr´eal, C. P. 6128 succ. centre ville, Montr´eal, Qu´ebec, Canada H3C3J7 a M 7 Accepted. Received; inoriginalform ] h ABSTRACT p We present the Fabry-Perot observations obtained for a new set of 108 galaxies in - o the frame of the GHASP survey (Gassendi HAlpha survey of SPirals). The GHASP r survey consists of 3D Hα data cubes for 203 spiral and irregular galaxies, covering t s a large range in morphological types and absolute magnitudes, for kinematics anal- a ysis. The new set of data presented here completes the survey. The GHASP sample [ is by now the largest sample of Fabry-Perot data ever published. The analysis of 1 the whole GHASP sample will be done in forthcoming papers. Using adaptive bin- v ning techniques based on Vorono¨ı tessellations, we have derived Hα data cubes from 6 which are computed Hα maps, radial velocity fields as well as residual velocity fields, 7 position-velocitydiagrams,rotationcurvesandthekinematicalparametersforalmost 9 all galaxies. Original improvements in the determination of the kinematical parame- 0 ters, rotation curves and their uncertainties have been implemented in the reduction . 5 procedure.Thisnew methodis basedonthe whole2Dvelocityfieldandonthe power 0 spectrum of the residual velocity field rather than the classical method using succes- 8 sive crowns in the velocity field. Among the results, we point out that morphological 0 position angles have systematically higher uncertainties than kinematical ones, espe- : v cially for galaxies with low inclination. Morphological inclination of galaxies having i no robust determination of their morphological position angle cannot be constrained X correctly.Galaxies with high inclination show a better agreementbetween their kine- r maticalinclinationandtheirmorphologicalinclinationcomputedassumingathindisk. a The consistency of the velocity amplitude of our rotation curves have been checked usingtheTully-Fisherrelationship.Ourdataareingoodagreementwithpreviousde- terminations found in the literature. Nevertheless, galaxies with low inclination have statistically higher velocities than expected and fast rotators are less luminous than expected. Key words: Galaxies: spiral; irregular;dwarf; Galaxies: kinematics and dynamics; 1 INTRODUCTION Fabry-Perot for studying their kinematical and dynamical properties through the ionized hydrogen component. As it is nowadays largely admitted, 2D velocity fields al- Studying the links between parameters reflecting the lowthecomputationof1Drotationcurvesinamorerobust dynamical state of a galaxy will help us to have a better way than long slit spectrography. Indeed, first of all, the understandingof theevolution of galaxies. This sample has spatialcoverageislargerandmoreover,thekinematicalpa- been constituted in order: rameters are determined a posteriori instead of a priori in longslitspectrography.Inthatcontext,wehaveundertaken (i) to computethe local Tully-Fisher relation; the kinematical 3D GHASP survey (acronym for Gassendi (ii) tocomparethekinematicsofgalaxies indifferenten- HαsurveyofSPirals).TheGHASPsurveyconsistsofasam- vironments(field,pairs,compactgroups,galaxiesincluster) ple of 203 spiral and irregular galaxies, mostly located in for discriminating secular evolution from an external origin nearby low density environments,observed with a scanning (e.g. Garrido et al. 2005); 2 B. Epinat et al. (iii) to study the distribution of luminous and dark halo B,thecommentsforeachindividualgalaxyaregiven.InAp- components along the Hubble sequence for high and low pendix C, the different tables are given while in Appendix surface brightness galaxies, for a wide range of luminosi- D the individual maps and position-velocity diagrams are ties in combining the optical data with the radio ones (e.g. shown.TherotationcurvesarefinallydisplayedinAppendix Spanoet al. 2007; Barnes et al. 2004); E while the numerical tables corresponding to the rotation (iv) to model the effect of non axisymmetric structures curvesare given in AppendixF. like bars, spiral arms, oval distortions, lopsidedness in When the distances of the galaxies are not known, a the mass distribution using both N-body + hydrodynamic Hubbleconstant H0=75 km s−1 Mpc−1 is used throughout numerical simulations (e.g. Hernandez et al. in prepara- thispaper. tion), kinemetric analysis (e.g. Krajnovi´c et al. 2006) and Tremaine-Weinberg method to measure the bar, spiral and innerstructure pattern speeds (Hernandezet al. 2005a); (v) toanalyzethegaseous velocitydispersion andtolink 2 OBSERVATIONS AND DATA REDUCTION it with thestellar one; 2.1 The GHASP sample (vi) to create templates rotation curves (e.g. Catinella et al. 2006; Persic & Salucci 1991; Persic et al. The GHASP survey was originally selected to be a sub- 1996) and templates velocity fields; sample complementing the radio survey WHISP (West- (vii) to map the 2D mass distribution using 2D veloc- erbork survey of HI in SPirals galaxies) providing HI ity field, broad band imagery and spectrophotometric evo- distribution and velocity maps for about 400 galax- lutionary models; ies (http://www.astro.rug.nl/∼whisp). The first set of (viii) tosearchforlinksbetweenthekinematics(shapeof GHASPgalaxieswasselectedfromthefirstWHISPwebsite rotationcurves,angularmomentum,...)andtheotherphys- list butsomeofthemhaveneverbeenobservedbyWHISP. icalpropertiesofgalaxieslikestarformation rate(e.g.com- Thus only 130 galaxies have finally been observed in both parison with star forming galaxies like blue compact galax- surveys.Thecomparison between thekinematics of neutral ies, ...); and ionized gas coming from the GHASP and the WHISP (ix) to produce a reference sample of nearby galaxies datasets is possible for a sub-sample of 31 dwarf galaxies to compare to the kinematics of high redshift galaxies studiedbySwaters et al.(2002)andforanothersub-sample (Puech et al. 2006; Epinat et al. 2007). Indeed, it is nec- of 19 early-type galaxies analyzed by Noordermeer et al. essary to disentangle the effects of galaxy evolution from (2005), the remaining part of the WHISPsample being yet spatial (beam smearing) and spectral resolution effects. unpublishedbutmostoftheHImapsareneverthelessavail- able on theWHISPwebsite. ThispaperisthesixthofaseriescalledhereafterPaper MostoftheGHASPtargetswerechoseninnearbylow- I to V (Garrido et al. 2002, 2003, 2004, 2005; Spano et al. densityenvironments.Nevertheless,somegalaxiesinpairsor 2007) presenting the data obtained in the frame of the inlowdensitygroupshavebeenobservedmainlywhenthey GHASP survey. The data gathered with the seven first ob- were selected by WHISP (see individual comments in Ap- serving runs havebeen published from Paper I to IV. Dark pendixB).SevengalaxiesoftheGHASPsamplearelocated matterdistributioninasub-sampleof36spiralgalaxieshave innearbyclusters(UGC1437intheclusterAbell262,UGC beenpresentedinPaperV.Thispaperpresentsthelastun- 4422and4456intheCancercluster,UGC6702inthecluster published 101 Hα data cubes of the GHASP survey. It in- Abell1367,UGC7021,7901and7985intheVirgocluster). cludes108galaxies(sevendatacubescontaintwogalaxies), More Virgo cluster galaxies observed with the same instru- providing106velocityfieldsand93rotationcurvesresulting ment havebeen published elsewhere (Chemin et al. 2006). from observational runs eight to fourteen. This represents Figure1displaysthedistributionofthewholeGHASP the largest set of galaxies observed with Fabry-Perot tech- surveyinthemagnitude-morphologicaltypeplane.Wehave niques ever published in the same paper (Schommer et al. foundintheliteraturemeasurementsforbothM andHub- B 1993 sample consists of 75 cluster galaxies in the southern bletypefor198galaxies(overthe203galaxies).Amongthe hemisphere observed with Fabry-Perot techniques and was sample of 203 galaxies, 83 are strongly barred galaxies (SB thelargestsamplepublishedtodate).Includingtheprevious orIB)and53aremoderatelybarredgalaxies(SABorIAB). papers(PaperItoIV),theGHASPsurveytotalizes Fabry- The GHASPsample provides a wide coverage in galaxy lu- Perotdatafor203galaxiesobservedfrom1998to2004.The minosities(-16≤M ≤-22),thusingalaxymasses(109M⊙–5 b GHASPsampleisbynowthelargestsampleofFabry-Perot 1011M⊙)andinmorphologicaltypes(fromSatoIrregular). data everpublished. Thewell-knownrelationforspiralandirregulargalaxiesbe- In section 2, the selection criteria of the GHASP sam- tween the morphological type and the absolute magnitude ple,theinstrumentalset-upof theinstrumentand thedata isobservedthroughtheGHASPsample.Withinthedisper- reductionprocedurearedescribed.Insection3,differentmo- sionofthisrelation,the203GHASPgalaxiesarereasonably menta of the data cubes are presented as well as the new distributed through the plane down to low magnitudes (≥- method to build the rotation curves and to determine the 16) and to early-types spiral (≤0, Sa). The whole GHASP uncertainties.Ananalysis oftheresidual velocityfieldsand datasetofgalaxiesisreasonablyrepresentativeoftheLEDA ofthekinematicalparametersisthusgiven.Insection4,the sample (see Paper IV). Tully-FisherrelationisplottedfortheGHASPgalaxiespre- Thejournaloftheobservationsforthe107newgalaxies sentedinthispaper.Insection5,wegivethesummaryand isgiveninTableC1.The108thgalaxy,UGC11300hasbeen conclusions. InAppendixA,wepresentsomedetails onthe observedforthethirdtimeinordertochecktheconsistency method used to compute the rotation curves. In Appendix of the new data reduction method (the second observation GHASP : An Hα kinematic survey of spiral and irregular galaxies - VI. 3 publishedinPaperIVwasalreadydoneinordertocompare thenew GaAs camera with the ”S20” photocathode), lead- ing to 107 new galaxies. Note that the right ascension and thedeclination given in table C1 are thecoordinates of the kinematicalcenter(andnotthemorphologicalonesgivenin HyperLedaor in LEDA, except if stated otherwise). 2.2 The instrumental setup In order to map the flux distribution and the velocity fields of the sample of galaxies, high spectral resolution 3D data cubes in the Hα line have been obtained. This has beenachievedusingafocal reducercontainingthescanning Fabry-Perot interferometer attached at the Cassegrain fo- cus of the 1.93 m OHP telescope (Observatoire de Haute Provence).Theinstrumentprinciplesandcharacteristicsare the same as for papers I, II, III, IV and V. The detector, a newgenerationimagephotoncountingsystem(IPCS)isthe sameastheoneusedforPaperIV(withaGaAsphotocath- ode).Thepixelsize is0.68′′(howevertheangularresolution of our data is limited by the seeing, about ∼3′′, e.g. Table C1), the field of view is 5.8 square arcmin and the velocity sampling is ∼5km s−1 (for a resolution of ∼10 km s−1). 2.3 The data reduction The Fabry-Perot technique provides an Hα profile inside eachpixel,sothatatypicalvelocityfieldofaGHASPgalaxy containsthousandsofvelocitypoints.Formostofthegalax- ies observed with GHASP, the velocity field is not limited totheHIIregionsbutcoversmostofthediffuseemissionof the disk, as can be seen on the figures. The detection limit of our device is about 10−18 erg cm−2 s−1 arcsec−2, with a S/N ratio between 1 and 2 for a typical 2 hours exposure timeaccordingtofigure2ofGach et al.(2002).Thisinsures a good detection of the Hα diffuse emission of the disk for most of the galaxies since most of the Hα emission found below 1.6 10−16 erg cm−2 s−1 arcsec−2 may be considered as filamentary and/or diffuse according to Ferguson et al. (1996).Thewaytoderivethedifferentmomentmapsofthe 3Ddatacube(HαlinemapsandHαvelocityfields)areex- plained in D2006. The Hα image is a pure monochromatic image of thegalaxy (continuumand [NII] free). In a few cases, when the velocity amplitude is compa- rable to or higher than the width of the interference filter, its transmission is not necessarily centered on the systemic Figure 1. Top: distribution of morphological type for almost velocityand onesideofthegalaxy maybebettertransmit- all of the GHASP sample (201/203 galaxies). Middle: distri- ted than the other side, leading to an artificial asymmetry bution of the absolute B-band magnitude for almost all of the GHASPsample(198/203galaxies).Forbothtopandmiddle,the in the intensity of the Hα emission (see Paper IV for ad- bluehash,redhashandresidualwhiterepresentrespectivelythe ditional explanations as well as in individual comments on stronglybarred,themoderatelybarredandthenon-barredgalax- each galaxy in AppendixB). ies. Bottom: distribution for almost all of the GHASP sample Thedataprocessingandthemeasurementsofthekine- (198/203 galaxies)inthe”magnitude-morphological type” plane matical parameters aredifferentfrom thoseused inPaper I distinguishingstronglybarred(bluesquares),moderatelybarred toV.Thedataprocessingusedinthispaperisbasicallythe (redtriangles)andunbarredgalaxies(blackcircles). sameastheonedescribedbyDaigle et al.(2006b)(hereafter D2006).Oneofthemainimprovementsimplementedinthis dataprocessingistheuseofadaptivespatialbinning,based on the 2D-Vorono¨ı tessellations method applied to the 3D datacubes,allowingtooptimizethespatialresolutiontothe signal-to-noise ratio. Letusalso mention thatwhenenough starsorbrightHIIregionswereavailableinthefieldofview, 4 B. Epinat et al. wecorrectedtheobservationfromtelescopedrift(orinstru- 3 DATA ANALYSIS mental flexures) when necessary. Hereafter, we just point 3.1 Different maps from the 3D data cube out the main difference between the method described in D2006 and themethod used in this paper. For each galaxy, in Appendix D, from Figure D1 to D106, we present up to five frames per figure: the XDSS blue The main difference is the criterion used to fix the size (or red) image (top/left), the Hα velocity field (top/right), of the bins. Indeed, with the spatial adaptive binning tech- the Hα monochromatic image (middle/left, eventually the nique, a bin is accreting new pixels until it has reached a Hα residual velocity field (middle/right) and finally the given criterion given a priori. In D2006, the criterion is the position-velocity diagram along the major axis (bottom) signal-to-noise ratio of theemission linewithin thebin.For when it can be computed. The white and black cross in- each bin, thenoise isdetermined from ther.m.s of thecon- dicates the center used for the kinematic analysis (given tinuum (the line free region of the whole spectrum). The in Table C1, e.g. Appendix A for determination) while the signal-to-noiseratioisthustheratiobetweenthefluxinthe black line traces the kinematical major axis deduced from lineandther.m.s.inthecontinuum.Whileforthedatade- the velocity field analysis (e.g. section 3.2) or the morpho- scribed in D2006, the number of channels scanned is large logical one (taken from HyperLeda) when no position an- enough(36channels)todetermineproperlythenoiseinthe gle of the kinematical major axis could be derived using continuum,thisisnotanymorethecaseherewherethenum- the kinematic (e.g. Table C2). This line ends at the radius ber of channels scanned is smaller (24 channels). For the D25/2 correspondingtotheisophotallevel25mag arcsec−2 same givenspectral resolution (fixedbytheinterferenceor- in the B-band (given in Table C3) in order to compare the derandtheFinesseoftheFabry-Perotinterferometerused), velocity field extent with the optical disk of the galaxies. the ratio between the number of channels containing the Position-velocity diagrams are computed along the axis de- continuum and the number of channels containing the line fined by this black line, using a virtual slit width of seven isthuslarger(byafactor3/2)inD2006thanhere.Further- pixels, and the red line on the position-velocity diagram is more,thecriterioninD2006isnotrelevantanymoreforthe therotationcurvededucedfromthemodelvelocityfield(see GHASPdata.Instead ofthesignal-to-noise ratio, thecrite- nextsection)alongthisvirtualslit.Whennofitissatisfying rion used here is simply the square root of the flux in the (generallybecauseofpoorsignal-to-noiseratio),weusedthe line,thatisanestimateofthePoisson noiseinthelineflux. real velocity field instead of the model (see individual cap- tionsinFiguresD1toD106).Therotationcurvesarefound in Appendix E (figures) and F (tables). Colour version of A second major improvement is the suppression of the therotationcurvesinAppendixEareonlyavailableonline. ghosts duetoreflection at theinterfaces air/glass of thein- Rotation curves are computed and displayed following the terferometer (Georgelin 1970). The reduction routine takes methoddescribedinsection3.2.Thesefiguresarealsoavail- into account the front reflection (between the interference able on theWeb site of GHASP: filter and the interferometer) and the back reflection (be- http://FabryPerot.oamp.fr. tween theinterferometerand thecamera detectorwindow). Inordertoillustratetheprintedversionofthepaperwehave The ghosts are calibrated and then subtracted thanks to chosen to display the diversity through four galaxies hav- bright stars. ing different morphological types,see AppendixD19 (UGC 3740, SAB(r)c pec), D31 (UGC 4820, S(r)ab), D45 (UGC Another major improvement is an automatic cleaning 5786, SAB(r)b), D56 (UGC 7154, SBcd). The Appendix D of thevelocity fields.Theoutskirtsof agalaxy,where there maps of theother galaxies are only available on line. is no more diffuse Hα emission, has to be delimited. Due Only the first page of Appendix F that contains the to residual night sky lines and background emissions (af- tables corresponding to the rotation curves of the two first ter subtraction), adaptive binning produces large sky bins galaxiesisdisplayedontheprintedversionofthepaper,the with a given signal-to-noise ratio or given flux. These bins remaining part of AppendixF being available on line. containing only sky emission are separated from the bins of the galaxy thanks to a velocity continuity process. The velocity field is divided in several regions where the veloc- ity differencebetween contiguous binsis lower than a given 3.2 Construction of the Rotation Curves and cutoff value. The regions with too low monochromatic flux Determination of the Uncertainties and too large bins are erased. The given cutoff fixed a pri- A new automatic fitting method has been developed to ori isaboutonetenthofthetotalamplitudeofthevelocity derive automatically a rotation curve from the 2D veloc- field (let’s say 30 km s−1 for a velocity field with an overall ity field. This method makes the synthesis between (i) the amplitude of ∼300 km s−1). method used in Paper I to IV, (ii) the method based on tilted-ringmodels found for instance in the ROTCURrou- Karma (Gooch 1996) and its routine Koords have tineof Gipsy(Begeman1987)and(iii)themethodusedby beenusedtocomputetheastrometry.XDSSBluebandim- Barnes & Sellwood (2003). ages or XDSS Red band images when blue image was not Warps are mainly seen in galactic disks at radii R > availablearedisplayed(seeindividualcaptionsinFiguresD1 R . Tilted-ring models have been developed to model the opt toD106).Systematiccomparisonbetweenthesebroad-band distribution of neutral hydrogen for which warps of the HI images and thefield stars in high resolution continuum im- diskmay bemoreor less severe.Incase of awarp, amono- ages (with no adaptivebinning) were made in order to find tonic change of the major axis position angle (PA) and of thecorrect World Coordinate System for each image. theinclination (i) is observed. GHASP : An Hα kinematic survey of spiral and irregular galaxies - VI. 5 On the other hand, within the optical disk, the kine- than the rotation curve and allows to follow the peak-to- matic parameters PA and i do not vary significantly and peakorpeak-to-valleyvelocitydistributionalongthemajor change monotonically with the radius (Paper I to IV and axis. Hernandezet al.2005b).ThevariationofPAandiwiththe radiusismorelikelyduetononcircularmotionsintheplane 3.3 Residual velocity fields ofthedisk(e.g.bars)thantomotionsoutoftheplane(like warps) and looks like oscillations around a median value. As detailed in the previous paragraph and in Appendix Thus, we do not allow PA nor i to vary with theradius. A, the main assumption necessary to derive a rotation Using tilted-ring models, the errors on the parameters curve from the observed velocity field is that rotation is are the dispersion of the kinematical parameters over the dominant and that all non circular motions are not part of rings.Themethoddevelopedhereusesthewholeresidualve- alarge-scale pattern.Thefivekinematical parameterscom- locity fieldtoestimate thedispersion inducedbynon circu- putedfrom thevelocity field todrawtherotation curveare lar motions and not only thesegmented information within determinedfromdifferentsymmetrypropertiesoftheradial each ring as it is thecase in tilted-ringmodels. velocity field. The influence of small errors in these param- Our fitting method is similar to Barnes & Sellwood eters is to produce patterns with characteristic symmetries (2003)method.Twodifferencesmayneverthelessbepointed in the residual velocity field. This was first illustrated by out. A minor difference is that they use a non parametric Warneret al. (1973) and by van derKruit & Allen (1978). profile while we fit an analytic function (more details on Intheir schematic representation of theresidual motions in thebuildingoftherotationcurvearegiveninAppendixA). diskgalaxies(themodelledvelocityfieldcomputedfromthe Themajorimprovementisthecomputationofthekinematic rotation curvehasbeensubtracted totheobservedvelocity uncertainties. Indeed, the statistical uncertainty on the fit field),abaddeterminationofoneorseveralkinematicalpa- is unrealistically small (Barnes & Sellwood 2003), because rameters leads to typical signatures in the residual velocity thenoiseonthedataisconsideredasablankrandomnoise. field(e.g.velocityasymmetryaroundthemajoraxisincase Thatisnotthecasebecausethenoiseintheresidualvelocity of a bad position angle determination, ...). The residual ve- fieldismainlyduetononcircularmotions(bar,ovaldistor- locity fields plotted for each galaxy in Appendix D clearly tions,spiralarms, local inflowsandoutflows,...) andtothe show that these typical signatures are not seen, this means intrinsicturbulenceofthegasthathavecharacteristiccorre- that the best determination of the kinematical parameters lation lengths. Inorder totakeit intoaccount, we compute has been achieved. the errors with the power spectrum of the residual velocity Thedeviationfrompurelycircularvelocitycanbelarge. field, applyinga Monte-Carlo method (see AppendixA). In a forthcoming paperthese residual velocity fields will be Rotation curves for the barred galaxies of our sample analyzed in terms of bars and oval distortions, warps, spi- have been plotted without correction for non-circular mo- ral arms (streaming motions), outflows and inflows, ... (e.g. tions along thebar. Fathiet al. 2007). Therotationcurvesaresampledwithrings.Withinthe The mean velocity dispersion on each residual velocity transition radius (definedin AppendixA), thewidth of the fieldhasbeencomputedforeachgalaxyandtabulatedinTa- rings is set to match half the seeing. Beyond that radius, ble C2, they range from 4 to 54 km s−1 with a mean value each ring contains from 16 to 25 velocity bins. around13kms−1.Figure2showsthattheresidualvelocity Thecurvesareplottedwithbothsidessuperimposedin dispersioniscorrelatedwiththemaximumamplitudeofthe thesame quadrant,usingdifferent symbolsforthereceding velocity field (shown by the dashed linear regression), this (crosses) and approaching (dots) side (with respect to the trend remains if we display the residual velocity dispersion center). The black vertical arrow on the x-axis represents versusthemaximumcircularvelocity(notplotted).Surpris- theradiusD25/2whilethesmallergreyarrowonthex-axis ingly,barredgalaxiesdonothave,inaverage,ahighermean represents thetransition radius, always smaller than D25/2 residualvelocitydispersionthanunbarredgalaxies(notplot- by definition. ted).Thismay beexplained bythefact that thenumberof For galaxies seen almost edge-on (inclination higher binscontaminated by thebar is usually rather low with re- than 75◦) our model does not describe accurately the ro- spect to the total bins of the disk. Indeed, this is not the tation of a galaxy since the thickness and the opacity of casefordisksdominatedbyabar.Comparedtothegeneral the disk cannot be neglected anymore. Indeed, on the one trend, we observe a set of about a dozen of galaxies with hand it is well known that, due to inner galactic absorp- ahigh residual velocity dispersion (pointsabove thedotted tion, edge-on galaxies tend to display smoother inner ve- lineinFigure2).Thesepointscorrespondtogalaxieshaving locity field and rotation curve gradients than galaxies with strong bar or spiral structure and to data of lower quality: loworintermediateinclinationsand,ontheotherhand,due (i) galaxies dominated by strong bars (UGC 89 and UGC to the actual thickness of the disk, using a simple rotation 11407), or strong spiral structures (UGC 5786 and UGC modelintheplaneofthegalaxydisk,motionoutofthedisk 3334) are not correctly described by our model which does are wrongly interpreted as circular motions in the disk. As not take into account non axisymmetric motions; (ii) the a consequence, for most of highly inclined galaxies, the fit velocity field of the lower quality data (UGC 1655, UGC convergestowardsunrealisticlowinclinationvalues,leading 3528, IC 476, UGC 4256 , UGC 4456, IC 2542, UGC 6277, to modelled velocity fields and rotation curves having too UGC9406andUGC11269) presentameansizeofthebins high velocity amplitudes. Thus, for NGC 542, UGC 5279, greaterthan25pixelsandanintegratedtotalHαfluxlower UGC 5351, UGC 7699, UGC 9219, UGC 10713 and UGC than4.5Wm−2 (aroughcalibration ofthetotalHαfluxof 11332, no rotation curve has been plotted. For them, the GHASP galaxies using the 26 galaxies we have in common position-velocity diagram givesa moresuitable information with James et al. 2004 has been made, assuming a spectral 6 B. Epinat et al. dueto thedifference in wavelength between the calibration and theredshifted Hα lines, thecoating of the Fabry-Perot interferometer induces a small systematic bias (phase shift effect) to the absolute systemic velocities. We tabulate the systemicvelocitieswithout correctingthemfrom thisphase shift because the dispersion by the phase effect is typically of the same order of magnitude than the dispersion of the systemicvelocitiesfoundinHyperLeda(PaperIV)andalso because the forthcoming analysis and in particular the ro- tation curvesdo not depend on thiseffect. In Figure 3, the kinematical position angles obtained byGHASParecomparedwiththephotometricpositionan- gles (found in HyperLeda). The error bar on the morpho- logical position angle, which is generally not homogenously given in the literature (or not given at all), has been esti- mated using theaxis ratio and optical radius uncertainties. Thegalaxydiskintheskyplaneismodelledbyanellipseof axisratiob/awhereaisequaltoD25.Giventheuncertainty on D25, ∆D25, a circle of diameter D25-∆D25/2 having the same center than the ellipse is considered. A line passing Figure 2. Dispersion in residual velocity field versus maxi- through the intersection between the ellipse and the circle mumvelocity,sortedbyHubblemorphologicaltype:blackcircles and their common center is thusdefined.The angle formed 0≤t<2, red triangles 2≤t<4, blue squares 4≤t<6, green rhom- between the major axis of the ellipse and the previously buses6≤t<8andpinkstars8≤t<10.Thedashedlinerepresents defined line represents the 1-σ uncertainty on the position the linear regression on the data. The points above the dotted angle. line are discussed in section 3.3. UGC 3334 labelled with an ar- rowhasactuallyaresidualvelocitydispersionof54kms−1 (see Forall galaxies, HyperLedareferences alist ofposition TableC2). angles from which they often computed one position an- gle value. HyperLeda does not compute a value when the dispersion or the uncertainty is too large. Indeed, the posi- ratio Hα over [NII] of 3:1). Figure 2 also shows that, for a tion angle may be quite different from a study to another, given velocity amplitude, this correlation does not clearly depending on (i) the method, (ii) the size of the disk and dependon themorphological type.Wenotethewell known (iii) the broad band colors considered by the different au- fact that late-typegalaxies havein average a lower velocity thors(nonhomogeneityinradiusandcolorsmeasurements). amplitude than early-typeones. WhennovalueiscomputedinHyperLeda,weputthewhole For most of the galaxies seen almost edge-on (i higher list in TableC2.Moreover, tomakeit readable andtomin- than 75◦), due to the thickness of the disk, no model has imize the dispersion on Figure 3, we only plot the morpho- beenfitted(seeprevioussubsection) thusnoresidualveloc- logicalvaluefoundclosestfromthekinematicalpositionan- ity fields can beplotted. gle. In Figure 3, we have distinguished the bulk of galaxies (blackcircles)forwhichtheagreementisrathergood(lower than 20◦ see 3, bottom) from (i) the galaxies for which no 3.4 Kinematical parameters accurate morphological position angle has been computed TableC2givestheinput(morphological) parameters ofthe (redopencircles)and(ii)thegalaxieshavinganinclination fits and the results of the fits (output parameters, χ2, and lowerthan25◦ (bluesquares).Indeed,somegalaxiespresent parameters oftheresidual maps).TableC3gives somefun- a disagreement between kinematical and photometric posi- damental parameters of the galaxies compiled in the litera- tion angle larger than 20◦. Most of these galaxies have (i) ture(morphologicalandHubbletype,distance,MB,D25/2, a bad morphological determination of the position angle or axis ratio, HI maps available in literature), together with (ii) have kinematical inclinations lower than 25◦or (iii) are maximum velocity parameters computed from the rotation specificcases (namely UGC3740, IC476, UGC4256, UGC curves(V ,qualityflagonV ).Thefourgalaxieslarger 4422) andarediscussedinAppendixB.Ontheotherhand, max max thanourfieldofviewareflagged intheTableC3.Forsome Figure3showsthatmorphological positionangleshavesys- galaxies for which the signal-to-noise ratio or the spatial tematically higheruncertainties thankinematical ones, this coverageistoolow, thefitcouldnotconvergecorrectly and is specially true for galaxies with low inclination. Quanti- one or two parameters (i and PA) were usually fixed to tatively, for kinematical inclinations greater than 25◦, the the morphological values to achieve the fit. These galaxies mean error on morphological position angles is ∼13◦ while are flagged with an asterisk (∗) in the Table C2. When it the mean error on kinematical position angles is ∼2◦. For is not the case, parameter determinations are discussed in inclinations lower than 25◦ the difference in the methods is Appendix B. For some extreme cases, even when i and PA even larger: the mean error on morphological position an- werefixed,thefitdoesnotconverge.Inparticularforgalax- gles is ∼27◦ while the mean error on kinematical position ies having high inclinations, then no model was computed angles is ∼3◦. For comparison, Barnes & Sellwood (2003), (see section 3.2). usingthedifferencebetweenmorphological andkinematical AsunderlinedinPaperI,Garrido(2003)andPaperIV, parameters,estimatedthatnonaxisymmetricfeaturesintro- GHASP : An Hα kinematic survey of spiral and irregular galaxies - VI. 7 duce inclination and position angle uncertainties of 5◦ on average. Thehistogramofthevariationbetweenkinematicaland morphological position angles given in Figure 3 (bottom) indicates that (i) for more than 60% of these galaxies, the agreement is better than 10◦; (ii) for more than 83%, the agreementisbetterthan20◦;(iii)thedisagreementislarger than 30◦ for 15% of these galaxies. Inotherwords,theposition oftheslitinlongslitspec- troscopy (which is usually based on the major axis deter- mined from broad-band imagery) with respect to the ac- tual position angle may be not negligible, highlighting the strengthoftheintegralfieldspectroscopymethodstodeter- minethe position angles (see also illustrations in Paper IV, Chemin et al. 2006 and Daigle et al. 2006a). In Figure 4, the inclinations obtained by GHASP are compared with the photometric inclinations. On the top panel the photometric inclination is the one computed us- ingacorrectionfactordependingonthemorphologicaltype (Hubble1926): 1−10−2logr25 sin2i= 1−10−2logr0 where r25 is the apparent flattening, and logr0 = 0.43+ 0.053 t for the de Vaucouleurs type t ranging from -5 to 7, and logr0 =0.38 for t higher than 7 (Paturel et al. 1997). On the middle panel the photometric inclination i is derived from the axis ratio b/a without any correction (cosi = b/a). As for the position angles, red open circles are the galaxies for which the morphological position an- gle could not be determined accurately. Blue squares are thegalaxies forwhich thedifferencebetween morphological and kinematical position angle exceeds 20◦. Thedispersionaroundthey=xline(equalitybetween Figure 3. Top: kinematical versus morphological (HyperLeda) the morphological and kinematical inclinations) decreases positionanglesofthemajoraxis.Galaxiesforwhichnoaccurate with the inclination. The main discrepancy is found for morphologicalpositionanglehasbeencomputedareshownbyred low inclinations. The corresponding galaxies are discussed opencircles;galaxieshavinganinclinationlowerthan25◦aredis- in Appendix B (notes on individual galaxies). On the one playedbybluesquares;theothergalaxiesarerepresentedbyblack hand, the morphological inclination of the galaxies having circles.Bottom:histogramofthevariationbetweenkinematical norobustdeterminationoftheirmorphological position an- and morphological positionangles. The redhash, blue hash and gle cannot be constrained correctly. On the other hand, residualwhiterepresentrespectivelythegalaxiesforwhichnoac- galaxies for which the position angle disagreement is rel- curatepositionanglehasbeenmeasured,forwhichinclinationis lowerthan25◦ andtheothergalaxiesofthesample. atively high havea high dispersion and theirmorphological inclination is statistically overestimated. Excluding these galaxies which have a bad position cal inclination may mean that the morphological thickness angle estimation, for low inclination systems, kinematical corrections are overestimated. methodsmayunderestimatetheinclinationoralternatively, Thehistogram of thedifferencebetween morphological morphological estimations may be overestimated. In aver- and kinematical inclinations (Figure 4, bottom) shows that age, the errors on morphological inclinations (∼6◦) and on a difference of inclination larger than 10◦ is found for 40% kinematical inclinations (∼8◦) are comparable. Whatever of thesample. the method used, the determination of the inclination of galaxies havingalowinclination remainslessaccuratethan for more inclined galaxies. 4 THE TULLY-FISHER RELATION The comparison of the two plots in Figure 4 (top and middle) shows that galaxies with high inclination have a Among the present sample of 108 galaxies, we have plot- betteragreement between their kinematical inclination and ted the Tully-Fisher relation (Tully & Fisher 1977, M as B theirmorphologicalinclinationcomputedconsideringathin a function of log2V ) for a sub-sample of 94 galaxies in max disk. The actual thickness of the disk may not be repro- Figure 5. The 14 other galaxies are not considered in the duced by our simple thin disk velocity field modelling. If present discussion because (i) for five galaxies the rotation it is the case, the kinematical inclination may be system- curvedoes not reach themaximum rotation velocity (UGC atically underestimated. Alternatively, the good agreement 1655, UGC 4393, UGC 6523, UGC 8898 and UGC 9406); between thin disk morphological inclination and kinemati- (ii) noBmagnitudeisavailable for onegalaxy (UGC3685) 8 B. Epinat et al. and (iii) no velocity measurement neither on the rotation curve nor on the position-velocity diagram is possible for eight othergalaxies (see Table C3). The maximum velocity V has been obtained from max the fit to the velocity field. The error on V is the max quadraticcombinationoftheerrorduetotheuncertaintyon theinclination(theproductV ×siniisconstant)andthe max median dispersion in therings oftherotation curvebeyond D25/10. In the cases where the rotation curve has no point beyond that radius, we replace this term by the intrinsic uncertainty on the velocity determination due to the spec- tral resolution (8 km s−1). For the highly inclined galaxies for which no correct fit was possible with our method (be- causeitdoesnottakeintoaccountthethicknessofthedisk, see section 3.2), we computed V from the Hα position- max velocitydiagramcorrectedfromthephotometricinclination. For them, the error on V is simply the intrinsic uncer- max taintyonthevelocitydetermination.Fortheparticularcase ofUGC5786,thefitisnotgoodenoughtouseittocompute V because of thelong bluenorthern tail and because of max the strong bar. We estimated V to 80 km s−1 by eye max inspection of the rotation curve. These galaxies are flagged in Table C3. The solid line in Figure 5 is the relation found by Tully & Pierce (2000): M =−7.3[log2V −2.5]−20.1 B max In Figure 5 (Top), the error bars on the velocity are displayed and galaxies with inclination lower than 25◦ are distinguished (blue open squares). We clearly notice that these galaxies have statistically higher velocities than ex- pectedfromtheTully & Pierce(2000)relation.Thiseffectis dueto thelink between inclination and velocity determina- tion.Indeed,onthevelocityfields,weobservetheprojected velocity on the line of sight: V ×sini. A given underesti- rot mate of the inclination thus leads to a higher overestimate on maximum velocity for low inclination galaxies than for highinclinationgalaxies.Thisalsoexplainsthestrongtrend forlow inclination galaxies toexhibitlarge errorbars.Con- sideringthiseffect,wechoosetoexcludethe15galaxieswith inclinations lower than 25◦ from the Tully-Fisheranalysis. Among the 79 remaining galaxies, the maximum ve- locity V is reached for 48 of them (black dots, large max size),probablyreachedfor17ofthem(bluesquares,medium size)andprobablynotreachedfor14ofthem(redtriangles, smallsize).TheyaredistinguishedinFigure5(Middle)and flagged in Table C3. The quality flag on the maximum ve- locity is deduced from (i) the inspection of the shape of the Hα rotation curves and position-velocity diagrams; (ii) from the comparison with HI velocity fields and rotation curveswhenavailable(seeTableC3);(iii)fromthecompar- isonoftheHαvelocityfieldsamplitudeswithHIlinewidths (see individual comments in Appendix B). It appears from Figure 4.Top: kinematical versus thick disk morphological in- this last point that the HI line width at 20% has most of- clinations. Middle: kinematical versus thin disk morphological tenthebestagreementwiththeHαvelocityfieldamplitude inclinations. Top and Middle: Galaxies for which no accurate (betterthanthelinewidthat50%).Figure5(Middle)con- morphologicalpositionanglehasbeencomputedareshownbyred firms the two classifications ”V probably reached” and max open circles; galaxies with a difference between the kinematical ”V probablynotreached”sinceforthemajority ofeach and morphological position angles larger than 20◦ are displayed max classthepointsarerespectivelyinagreementandabovethe with blue squares; the other galaxies are represented by black Tully & Pierce(2000)relation.Fromthetwoclasses”V circles. Bottom: histogram of the variation between kinemati- max cal andmorphological inclinations.The redhash, bluehash and reached”and ”Vmax probably reached”, wefindthefollow- residual white represent respectively the galaxies for which no ing relation: accurate position angle has been measured, for which the differ- GHASP : An Hα kinematic survey of spiral and irregular galaxies - VI. 9 M =(−6.9±1.6)[log2V −2.5]−(19.8±0.1) (1) B max This relation is displayed as a dotted line in Figure 5, in whichmorphologicaltypesaredistinguishedforthetwobest classes (black circles from 0 to 2, red triangles from 2 to 4, blue squares from 4 to 6, green rhombuses from 6 to 8 and pink stars from 8 to 10). Coefficients have been com- puted using the mean of the coefficients obtained (i) using a fit on the absolute magnitudes (as dependant variables) and (ii) using a fit on the velocities (as dependant vari- ables). The difference in the slope determination by these twomethodsisquitelargeduetoastrongscatterinourdata (the error on the parameter in equation 1 is half that dif- ference).Indeed,usuallyoneuseslocalcalibratorsforwhich distancemeasurementsareaccurate(basedonCepheids,red giants branch, members of a same cluster, ...) leading to a small scatter in the data. From our data, the main diffi- culty is that the distance determination is mostly based on the systemic velocity corrected from Virgo infall (see Ta- ble C3), and that no error bar on the magnitude can be easily estimated. Thus we haveno reason to be more confi- dent on the velocity measurements (mainly affected by in- clination determination) than on absolute magnitude mea- surements.However,despite thedispersion in ourdata, the resulting parameters using the mean of the two fits are in good agreement with Tully & Pierce (2000), even if our slope is a bit lower. A lower slope had already been ob- servedinHI(Yasudaet al.1997;Federspiel et al.1998)and more systematically in optical studies (e.g. Courteau 1997; Rubinet al. 1999; M´arquez et al. 2002; papers III and IV). For the Tully-Fisher relation we derived, on the one hand we observe that fast rotators (V > 300 km s−1: max UGC 89, UGC 4422, UGC 4820, UGC 5532, UGC 8900, UGC 8937 and UGC 11470) are less luminous than ex- pected, except maybe for UGC 3334 which is one of the fastest disk rotators (Rubinet al. 1979) (see discussion in AppendixB).Thistrendcanalsobeobservedinseveralop- ticalstudies(M´arquezet al.2002;PapersIIIandIV).Inter- estingly, these fast rotators are not observed in HI samples (Tully & Pierce 2000; Federspiel et al. 1998). This may be explained by the shape of the rotation curves of fast rota- Figure 5.TFrelationforour sampleof galaxies. Thesolidline tors:exceptforUGC8900,therotationcurvealwaysreaches represents the B magnitude Tully-Fisherrelation determined by the maximum velocity within the first five arcseconds, (i.e. Tully&Pierce(2000)fromnearbygalaxiesinclusters(UrsaMa- within our seeing). This inner maximum may be missed in jor,Pisces filament, Coma). Top:sorted byinclination- low in- HI because of beam smearing, averaging the maximum ve- clinationgalaxies(i<25◦):bluesquares;other galaxies(i≥25◦): locity reachedinthecenter.Notethattherotation curveof black circles. Middle: sorted by Vmax flags - Vmax reached: UGC5532isclearlydecreasingwhiletheotheronesareflat. black dots, large size; Vmax probably reached: blue squares, Ontheotherhandslowrotatorshaveasmallvelocitygradi- medium size; Vmax probably not reached: red triangles, small ent and within the optical regions the maximum could not size.Bottom:sorted bymorphological type - black circlesfrom be reached whereas HI observations would be able to mea- 0to 2;redtriangles from2to4;bluesquares from4to6; green sure it without any doubt. These two effects could explain rhombuses from 6 to 8; pink stars from 8 to 10; the dotted line representsthebestlinearfitonthedata. thetrend observed in optical Tully-Fisher relations. The result obtained for the Tully-Fisher relation is in agreement with the one obtained with the previous sam- understandingofthephysicsandevolution ofgalaxies. The ples(PapersIIIandIV).TheanalysisofthewholeGHASP GHASP sample, which consists of 203 spiral and irregular sample will be done in a forthcoming paper (Epinat et al. galaxies covering in a wide range of morphological types in preparation). and absolute magnitudes, has been constituted in order to provide a kinematical reference sample of nearby galaxies. TheGHASPgalaxieshavebeenobservedintheHαlinewith 5 SUMMARY AND PERSPECTIVES a scanning Fabry-Perot,providing data cubes. The knowledge of the links between the kinematical and Wepresentinthispaperthelastsetof108galaxieslead- dynamical state of galaxies will help us to have a better ingto106velocityfieldsand93rotationcurves.Bynow,this 10 B. Epinat et al. workconsistsofthelargestsampleofgalaxiesobservedwith norobustdeterminationoftheirmorphological position an- Fabry-Perot techniqueseverpresented in thesame publica- gle cannot be constrained correctly. Galaxies for which the tion. Added to the four previous sets already obtained in position angle disagreement is relatively high have a high theframeofthissurvey(PaperItoIV),GHASPrepresents dispersionandtheirmorphologicalinclinationisstatistically thelargest sample of 2D velocity fieldsof galaxies observed overestimated. Galaxies with high inclination havea better at Hα wavelength. For each galaxy, we have presented the agreement between their kinematical inclination and their Hα velocity field, the Hα monochromatic image and even- morphological inclination computed assuming a thin disk. tually the Hα residual velocity field, the position-velocity Forgalaxieswithintermediatediskinclinations(higherthan diagram along the major axis and the rotation curve when 25◦ and lower than 75◦), to reduce the degrees of freedom available. in kinematical models, the inclination could be fixed to the Major improvements in the reduction and in the anal- morphological value. This is specially true when only low ysis havebeen developed and implemented: quality kinematical data are available as it is the case for high redshift galaxies. • in order to optimize the spatial resolution for a given • The Tully-Fisher relation found with this new set of signal-to-noise ratio, adaptative binning method, based on dataisingood agreement with Tully & Pierce (2000),even the 2D-Vorono¨ı tessellations, was used to derive the 3D if our slope is a bit lower. This trend for a lower slope has Hα data cubes and to extract from it the line maps and already been observed in HI by Yasudaet al. (1997) and theradial velocity fields; Federspiel et al.(1998).Galaxieswithinclinationlowerthan • the ghosts due to reflections at the interfaces air/glass 25◦, have statistically higher velocities than expected from oftheinterferometer,havebeenremovedinthedatacubes; theTFrelationderivedbyTully & Pierce(2000).Fastrota- • the analysis of the faint outskirts or diffuse regions is tors (V >300 km s−1) are less luminous than expected. max automatic; Thismaybeexplainedbytheshapeoftherotationcurvesof • thekinematicalparameters andtheirerrorbarsaredi- fast rotators. rectly derived from thevelocity field; • theuncertaintiesareestimatedfromtheanalysisofthe residual velocity field power spectrum; • the whole 2D velocity field has been used rather than ACKNOWLEDGEMENTS successive crowns in tilted-ring models to compute the ro- TheauthorswarmlythankDrO.Garridoforleadingorpar- tation curveand the error bars. ticipating to most of the observations. They also thank the Programme National Galaxies for supporting the GHASP The main results of this paper are summarized by the projectinallocatingcontinuouslyobservingtimeduringsev- following items. eral years, the Observatoire de Haute-Provence team for • Theabsenceoftypicalandwellknownbiasintheresid- its technical assistance during the observations, O. Boissin ualvelocityfields meansthatthebestdeterminationofthe for his technical help during the observing runs, and J. kinematical parameters has been achieved. Boulesteix for permanent support. They thank I. J´egouzo • The mean velocity dispersion on each residual veloc- and C. Surace for building the Fabry-Perot Database. This ity field ranges from 6 to 23 km s−1 with a mean value research has made use of the GOLD Mine Database and around 13 km s−1 and is strongly correlated with the max- of the NASA/IPAC Extragalactic Database (NED) which imum amplitude of the velocity field. For a given velocity is operated by the Jet Propulsion Laboratory, California amplitude, this correlation does not clearly depend on the Institute of Technology, under contract with the National morphological type. Only strongly barred galaxies have a Aeronautics and Space Administration. The authors have higher residual velocity dispersion than mild-barred or non also made an extensive use of the HyperLeda Database barred galaxies. Peculiar galaxies also show a high residual (http://leda.univ-lyon1.fr). The Digitized Sky Surveys velocity dispersion. were producedat theSpaceTelescope Science Instituteun- • The kinematical position angles obtained by GHASP der U.S. Government grant NAG W-2166. The images of arecomparedwiththephotometricpositionangles.Morpho- these surveys are based on photographic data obtained us- logicalpositionangleshavesystematicallyhigheruncertain- ingtheOschinSchmidtTelescopeonPalomarMountainand tiesthankinematical ones,thisisspecially trueforgalaxies the UK Schmidt Telescope. The plates were processed into withlowinclination.Whenusinglongslit spectroscopy,the the present compressed digital form with the permission of positionangleshouldbeknownapriori.Thisisusuallydone theseinstitutions. using morphological determinations based on broad band imagery. We have shown that in some cases the difference betweentheposition angledeterminedusing2Dkinematics REFERENCES and morphologies may be as large as 90◦ and that in any case the position angles are better determined by 2D kine- Amram P., Le Coarer E., Marcelin M., Balkowski C., Sul- matics. Thus, large differences between morphological and livan III W. T., Cayatte V., 1992, AAPS,94, 175 kinematical position angles may lead to incorrect rotation AmramP.,MarcelinM.,BalkowskiC.,CayatteV.,Sullivan curveandmaximumvelocitydeterminationwhenusinglong III W.T., Le Coarer E., 1994, AAPS,103, 5 slit spectroscopy. This may strongly bias mass distribution ArpH.,1966,Atlasofpeculiargalaxies.Pasadena:Califor- models and Tully-Fisher studies, highlighting the strength nia Inst.Technology, 1966 of integral field spectroscopy with a Fabry-Perot. Balick B., Heckman T., 1981, AAP,96, 271 • The morphological inclination of the galaxies having Barnes E. I., Sellwood J. A., 2003, AJ, 125, 1164
Description: