Astronomy&Astrophysicsmanuscriptno.shell cESO2013 (cid:13) January18,2013 HI shells in the Leiden/Argentina/Bonn HI survey S.EhlerováandJ.Palouš AstronomicalInstitute,AcademyofSciences,BocˇníII1401,Prague e-mail:[email protected] ReceivedSeptember6,2012;acceptedNovember9,2012 ABSTRACT 3 1 Aims.Weanalysetheall-skyLeiden/Argentina/BonnHIsurvey,whereweidentifyshellsbelongingtotheMilkyWay. 0 Methods. We used an identification method based on the search of continuous regions of a low brightness temperature that are 2 compatiblewithgivenpropertiesofHIshells. n Results.Wefound333shellsinthewholeGalaxy.ThesizedistributionofshellsintheouterGalaxyisfittedbyapowerlawwiththe a coefficientof2.6correspondingtotheindex1.8inthedistributionofenergysources. Theirsurfacedensitydecreasesexponentially J withascalelengthof2.8kpc. Thesurfacedensityofshellswithradii 100 pcinthesolarneighbourhoodis 4kpc−2andthe2D porosityis 0.7. ≥ ∼ 7 ∼ 1 Keywords. ISM:bubbles–Galaxy:structure–radiolines:ISM ] A 1. Introduction nalgalaxies,wherewehavetoconsidertheorientationrelativeto G thelineofsight. IntheMilkyWay,therearemanyobservations h.Theneutralinterstellarmedium(ISM)ingalaxiesisfullofstruc- of individual shells and several papers dealing with the global ptures. Some of them are turbulent, wildly changing structures, picture(Heiles1979;McClure-Griffithsetal.2002;Daigleetal. -othersaremorestable,whileHIinparticularshowsverypromi- 2007;Ehlerová&Palouš2005),buteventhesedonotcoverthe onentstructures.Weseeroughlyellipticalobjectswithsizesfrom wholeGalaxy. r tafewpctomorethan1kpc,whichmayexpandintotheambient smedium.IntheMilkyWaytheywereidentifiedforthefirsttime a by Heiles (1979) and since then they have been discovered in [ many externalgalaxies, both spiral and irregular, including the Since2005theall-skyLeiden/Argentina/BonnHIsurveyhas 1LMCandSMC(foramorecompletereviewsee,e.g.,Bagetakos beenavailable(LAB;Kalberlaetal. 2005). Surveysof thesky vetal.2011). availablefromtheParkestelescope(GASS;Kalberlaetal.2010) 3 andtheEffelsbergtelescope(EBHIS;Kerpetal.2011)existor 0 ShellsmaybetheresultsofdifferentprocessesintheISM.It will exist in the near future and have a higher resolution than 0isachallengetodistinguishbetweenshellscreatedbytheenergy LAB. There are other surveys with a higher resolution (The 4insertedbystellarwinds,stellarradiation,andsupernovaeexplo- Southern Galactic Plane Survey, the Canadian Galactic Plane .sions;andstructurescreatedbyotherdrivingmechanisms,such 1 Survey, the VLA Galactic Plane Survey), but they cover only as the infalling high-velocity clouds, gamma ray bursts, or the 0 theregionalongtheGalacticequator. 3turbulencedrivenbygalacticdifferentialrotation.Wallsofshells 1maydevelopRayleigh-Taylor,Vishniac,Jeans,orotherinstabil- :ities,therebycreatinghigh-densityandlow-temperatureregions, v whichbecomesitesofsecondarytriggeredstarformation.Dense i Identificationofshellsisusuallydonebyeye. Humaneyes Xcondensationsof cold gas along the edgesof shells, secondary areverygoodatconnectingpartiallydisconnectedpartsofshells star formation, and compact HII regions of second-generation r and using some software tools (e.g.some routines in KARMA astars have been observed by Zavagno et al. (2006), Deharveng Gooch 1997) may help. Still, an objective method is prefer- et al. (2010), and others (see also review by Elmegreen 2012). able. There are a few attempts at an automatic identification However,manydetailsonthestarsvs. shellsconnectionremain processforshells:Thilkeretal.(1998)comparedahydrodynam- unknown,e.g.,thenumberofshellsperassociation,theenerget- ical model with observations; Daigle et al. (2007) based their icsinvolved,andthefractionofshellswithnonstellarorigins. methodonfindingsimple dynamicalcharacteristicsin HI data; Due to our position inside the Milky Way, observations of andEhlerová&Palouš(2005)(Paper1)locatedcontinuousre- shells in the Galaxy present special problems: unlike observa- gionsof low brightnesstemperature. The first attempt was not tionsofexternalgalaxies,anobserverhastoprocessasubstan- followedanyfurther,probablybecauseitisverydifficulttocon- tial part, ideallyall, of the skyto coverthe wholeGalaxy. The structa satisfactory and universalhydrodynamicalmodelcom- structures do not lie at almost the same distance as in external parabletotherealdata. Thesecondapproachisaimedatsmall galaxies, therefore results are more dependent on the distance bubbles (typically 10 pc) and was used for a portion of the ∼ ofobjectsfromthe Sun. Anotherdifferenceis theshapeof the CanadianGalacticPlaneSurveydata,whichfound 7000bub- ∼ shells perpendicular to the galactic symmetry plane. It is di- bleswithradiir < 40pc. Thethirdapproachisdevelopedin sh rectlyobservableintheMilkyWay,whilenotsoclearinexter- thiscommunication. Articlenumber,page1of19 2. Dataandmethods p,wemayartificiallyconnecttwoormorephysicallyunrelated, butneighbouring,structures,whileforahighvalueof pwemay 2.1.Leiden-Argentine-Bonnsurvey divideonephysicalstructurewithanirregularshapeintoseveral The Leiden-Argentine-Bonn HI survey (LAB, Kalberla et al. smallerpieces. 2005) is an all-sky survey, a combination of the Leiden- Connecting2Dstructuresinto3Donesmayincreasethean- Dwingeloosurvey(Hartmann&Burton1997)andtheInstituto gularsizesofstructures,sotheycanexceedthevaluedpixmax. ArgentinodeRadioastronomíasurvey(Bajajaetal.2005),with angularresolution36arcmin(thepixelsize30arcmin)andve- 2.2.3. PropertiesofHIshells locity resolution1 kms 1 (the channelwidth is 1.0306kms 1). − − Weusedthedatacubelimitedinv to( 250,+250)kms 1. WedonotprescribeanydesiredshapeforanHIshell, weonly lsr − − requestthat structures, which we call shells, span a certain ve- locity extent and that their shapes do not change significantly 2.2.Identificationprocess betweenconsecutivevelocitychannels. The first condition is simple. We demand that the velocity Theidentificationprocessisdividedintothreesteps:1)wetrace extentofthestructure∆visgreaterthanagivenvalue. Theve- continuousregionsaroundlocalminimain2Dchannelmaps;2) locityextentis we construct3Dstructuresfrom2Didentificationsfromstep 1 byconnectingconsecutiveoverlappingregions;3)weeliminate ∆v=v v , (1) max min structuresthatarenotcompatiblewiththegivenpropertiesofHI − shells. where vmin and vmax delimit the interval in which the structure exists. To estimate how much the shape of the structure changes 2.2.1. Definitionofa2Dstructure betweenthevelocitychannelswecalculatetheparameterP 2 Inchannelmapswe searchforregionswith locallylow bright- N (iv,iv+1) nesstemperatures. Theregionhastobecontinuous,i.e. allpix- P2(iv)= max(Nbot(hiv),N (iv+1)), (2) elsthatbelongtoitmustbeconnectedbytheiredges,notonlyby pix pix touchingtheircorners,andtheirbrightnesstemperaturesmustbe whereN (iv)isthenumberofpixelsbelongingtothestructure pix lowerthanorequalto a certain levelof brightnesstemperature in the channeliv, N (iv,iv+1) is the numberof overlapping both T . All neighbouringpixels, whichare notmembersof the pixels,i.e. thosepixelswithpositionsil,ib,whichbelongtothe Blevel identified region, have to have brightness temperatures higher structureinbothchannelsivandiv+1. than T . These conditions allow embedded clouds with a If P has a low value, it means that the lb-size or the lb- Blevel 2 highertemperature. position in subsequent velocity channels are very different. It Foreachchannelmap,therangeofT isexamined,start- mayalso mean thatthe 3D shape ofthe structure changessub- Blevel ing ata levelofnoise and growingto the maximumbrightness stantiallybetweenthetwoneighbouringchannels. Itservesasa temperature in the map. For the LAB processing we used the warning that the structures might not be compact or uniform. squarerootscaleforthebrightnesstemperature(i.e. weworked Since an ideal expanding shell, viewed in velocity channels, with √T ,notdirectlywithT ),thelowestT weprocessed startssmall, reachesits maximumsize, andthen shrinksagain, B B Blevel was0.64K, √T wasincreasedbystepsof0.2. we determinethe average P andthe minimumvalue P , Blevel 2,ave 2,min The identified regionhas sizes in l and b directions. These notinthe wholevelocityrangeofthestructure, butonlyinthe sizeshavetobewithinlimits(d ,d ). Foreachregion limited range of velocity channels between the two local max- pixmin pixmax we find the highest T for which it is fulfilled. We used imain P ,thefirstafterthelowestvelocitychannel,thesecond Blevel 2 d =3,whicheliminatedsinglepixelsandverysmallstruc- thelastbeforethehighestvelocitychannel. pixmin turesandsetstheminimumsizeofastructureto1.5◦. Thevalue ThecalculationofP2 issimilar,butnotidentical,tothecal- omfedapnixinmgax,,bsuettwtoe9fi0n(de.ogu.t4t5h◦a)t,ihtagsivneoscgloeoardprhesyuslitcsa;lworenwuemreeraibclael cduifflaetrieonncoefitshtehoatv1er)lahpepreinwgepcaorammpeatreertpheinntuhmebperervoifouovsesrtelapp.pTinhge to identify a whole range of structures under many conditions pixelswiththelargerofthesetwochannelstructures(itwasthe occurring in the data. A lower value of d would prevent smalleronepreviously),and2)inthisstepwealreadyevaluatea pixmax identification of angularlylarge shells. Very few 2D structures structureconstructedfromseveraloriginallyindividual2Dparts. have sizes larger than d = 90, only a few identifications Thespectrumthroughthecentreoftheshellshouldbeseen in the solar neighbourhopioxdm,axwhere such angularly large shells asanoticeable,smoothdepressioninthebrightnesstemperature mightoccur. profile compared to the profile outside the shell. We construct therelativeprofile 2.2.2. Creatinga3Dstructure ∆TB(iv)= Tin(iv) Tout(iv), (3) | − | Foreach2Dregionidentifiedinthepreviousstepwecheckthat whereTinistheaveragebrightnesstemperatureinpixelsbelong- ithasacounterpartinsubsequentchannelmaps.Wemeasurethe ing to the structure in a given velocity channel and Tout is the overlapbetweentheneighbouring2Dstructures,andifitislarge averagevalue of TB in pixelsjust outside the structure. Which enough, we connect them in one 3D structure. This continues pixelsareoutsideisdeterminedforeachvelocitychannelsepa- forallvelocitychannelsandforallstructures,untilallpossible ratelytoaccountforthechanginglb-sizeofthestructure. Then connectionsaremade. wedefine Theoverlapconditionisthatatleastafraction pofthearea P =max (∆T ), (4) ofthesmalleroneofthesetwostructuresisoverlapped.Weused 3 v B avalue p = 0.5,butresultswerenotverysensitivetothisvalue i.e., the maximum difference in average values of brightness (we also tested values 0.1 and 0.9). If we use a low value of temperature inside and outside the shell in all channels where Articlenumber,page2of19 S.EhlerováandJ.Palouš:HIshellsintheLeiden/Argentina/BonnHIsurvey Fig.1. RelativenumberofstructuresidentifiedinLABasafunctionoftheirvelocityextent(leftpanel);therelativenumberofstructuresthat fulfilltheP criterion(hatched),P criterion(cross-hatched),P criterion(empty),andallthree(solid)asafunctionoftheextentofvelocity 2,min 2,ave 3 (rightpanel). the given structure exists. Clearly visible shells have high P . d = 90 as a minimum and a maximum size of the struc- 3 pixmax Low P is connected with poorly visible shells, but also with turesandtheoverlappingparameterp=0.5,weidentified17535 3 shellsinaregionwithhighT gradients. 3Dstructures. Themajorityofthesestructuresspanonlyafew B The3DstructureisclaimedtobeanHIshellifitfulfillsthe velocity channels (see Fig. 1, left panel): 2152 structures (12 followingthreeconditions: %)haveavelocityextentgreaterthanorequaltofourchannels (4.1kms 1), 775structures(4.4%) have an extentgreaterthan ∆v ∆v ; P ; P ; (5) − ≥ min 2,min ≥P2,min 2,ave ≥P2,ave orequaltoeightchannels(8.2kms−1). Weeliminatedirregular withapossibleauxiliary“visibility”parameter, structuresbyadoptingparameters = 0.2and = 0.5 2,min 2,ave P P (seeFig.1,rightpanel,showingthefractionofstructures,which P , (6) 3 3 ≥P fulfill the requiredcriteria). Geometrical conditions and 2,min wstrhiecrtneetshseovfatlhueesseoafrc∆hv.min, P2,min, P2,ave, and P3 determine the vPe2l,oavceiteylimexitneantte).feAwsesrhtohwann∼in4F0i%g.o1f,tthheesmheolsltsr(edsetrpiecntiPdvseopnarthameir- eteristhevisibilityparameter :onlyabout40 50%ofshells 3 P − haveparameterP greaterthan4K.Thevalueof =4K will 2.3.Anewversionoftheidentifyingalgorithm 3 3 P bejustifiedinthenextsection. Thenumericalcodedescribedinthispaperisasuccessortothe Themostimportantcriterionineliminatingshellsisthemin- codedescribedandusedinEhlerová&Palouš,2005. Themain imum required velocity extent ∆v . The lowest acceptable min differencesfollow. valueisprobably∆v = 4kms 1,andthesafeoptioniscloser min − 1. The currentversion deals with the whole datacube at once, to∆vmin = 8kms−1. Thisparameterisrelatedtothedispersion whiletheoldversionhadtocutthebigdatacubeintosmaller velocityintheISM.Avalueof∆vmin lowerthanthetypicalve- subcubesandthenconnectthembacktogether. locitydispersionleadstoidentificationofsomefalsestructures, Cutting the big datacube into smaller ones led to artificial whilethehighervaluemayeliminatesomerealstructures. The edgeproblems(mostimportanttodistortionsintheshapeof dispersionvelocityinISMvariesgreatly,dependingonthecon- identifiedstructuresandthefailureofsomeidentifications). ditionsinISM,butitstypicalvaluefortheHIemissionisaround 2. Thelaststepintheidentification,“PropertiesofHIshells”, 7kms 1. − wascompletelychanged. Theoldversionusedtheanalysis oftheT (v)spectrumthroughthecentralpixelofthestruc- B tures, the current version used the geometrical parameters 3.2.ComparisonstopreviouslyknownHIshells P and P , and an auxiliary visibility parameter P . 2,min 2,ave 3 We compared our LAB identifications to two previous papers Bothversionsusedtheconditionwiththe∆v ,whichwas min on HI shells in the Milky Way: Heiles (1979) and McClure- setinthepreviousversionto∆v =4kms 1. min − Griffithset al. (2002),whichcontainthe majorityof knownHI The spectrum analysis worked well for small structures in supershells. Heiles’s list is divided into two parts: stationary smooth surroundings. For larger structures and a more vi- shells, which do not change their size with velocity; and ex- olent medium, it was sensitive to the exact position of the pandingshells, which show differentsizes in differentvelocity studiedspectrumandtothedefinitionofthebackgrounden- channels. hancingthepossibilityof false identifications. Geometrical Table1showsacomparisonof82supershellsfromthetwo parametersfromthecurrentversionaremorerobust. paperswithourLABidentificationsforthewholedatacube(the middle section of the table) and separately for the second and 3. Results thirdquadrants(the rightsection of the table). We chose these two quadrants, because they contained only the outer Galaxy, 3.1.LAB where structuresare less crowded, which makes identifications We used the LAB survey with v 250 kms 1 as an input more accurate. We distinguish whether the shells have similar lsr − | | ≤ datacube for our identifying algorithm. Using d = 3 and sizes(“volumerestricted”)ornot(“volumeunrestricted”). pixmin Articlenumber,page3of19 Fig.2. RegionofGS242-02+37,oneoftheexpandingshellsinHeiles(1979),asseenintheLABsurvey.Theoriginalidentificationisshownby theblackcross(lengthofitsarmscorrespondstopublishedsizes),ournewidentification(thestructureGSH243.5-02.5+043.3inappendedtables) isshownbythebox.Leftpanelisthelb map,rightpanelisthelv diagram. − − Fig.3. RegionofGS228-05+47,oneofthestationaryshellsinHeiles(1979),asseenintheLABsurvey.Theoriginalidentificationisshownby theblackcross(lengthofitsarmscorrespondstopublishedsizes),ournewidentification(thestructureGSH226.5-09.5+031.9inappendedtables) isshownbythebox.Leftpanelisthelb map,rightpanelisthelv diagram. − − Table1.ComparisonofidentificationswithHIsupershells:thewholeGalaxy(themiddlesection),the2ndand3rdquadrants(therightsection). num vol. restr. vol. unrestr. num vol. restr. vol. unrestr. Q2+3 Heiles1979,stationary 46 14(30%) 33(72%) 18 6(33%) 15(83%) Heiles1979,expanding 17 7(41%) 9(53%) 7 4(57%) 4(57%) McClure-Griffithsetal2002 19 10(53%) 17(89%) 4 3(75%) 4(100%) Notes.Theleftsectiondenotessources,fromwhichpreviouslyknownHIshellsweretaken. Themiddleandrightsectionsgivethetotalnumber ofshells(inthewholeGalaxyorinthe2ndand3rdquadrant,respectively)andthenumberofshells,whichwereidentifiedagainbyourmethod (percentageofthetotalisinparentheses). Wedistinguishtwocases: thefirst,wherewedemandthatsizesofthecomparedstructuresaresimilar (vol.restr.),andthesecond,wherewedonotputanyrestrictionsonsizesofthestructures(vol.unrestr.). Articlenumber,page4of19 S.EhlerováandJ.Palouš:HIshellsintheLeiden/Argentina/BonnHIsurvey 1. ∆v = 4 kms 1 (designated as delv4 in Fig. 4, an unde- min − mandingcondition); 2. ∆v =4kms 1,P 0.2,P 0.5(designatedC2_4 min − 2,min 2,ave ≥ ≥ inFig. 4); 3. ∆v = 8 kms 1, P 0.2, P 0.5 (designatedas min − 2,min 2,ave ≥ ≥ C2_8inFig. 4); 4. ∆v = 8 kms 1, P 0.2, P 0.5, P = 4 K min − 2,min 2,ave 3 ≥ ≥ (designatedasC3_8inFig. 4,thestrictestsetofconditions). Furthermore,wedemandedthatsizesofshellsweresimilar (“volumerestricted”). Correspondingfractionswere 61 %(the 1stset ofconditions),54%(2nd),33%(3rd),and26%(4th). Eachstructureinthepreviousversionhadavisualquality,which describes how good it looked (a purely subjective criterion: a shell with q = 1 is clearly visible in velocity channels and is probablysphericalorelliptical;ashellwith q = 4isinconspic- Fig. 4. Comparisons between previous and current versions of the searchalgorithms. Differentsymbolsdenotedifferent setsofparame- uousinvelocitychannels,withalowcontrasttoitssurrounding tersputonnewidentifications(seethetextfortheexactexplanationof and probably fragmented and irregular). The comparison was delv4,C2_4,C2_8,andC3_8).They-valueshowswhatfractionofold performedseparatelyforshellswithdifferentqualities(Fig. 4), identificationsisalsoobservedwiththenewalgorithm. andthebest-lookingshells(visualquality1)hadcorresponding fractionsrangingfrom80 % to 55 %. Fractionsof reidentified shells for volume-unrestrictedcomparisonswere 88 % (the 1st ThebestagreementwasfortheMcClure-Griffithsshells(in setofconditions),81%(2nd),76%(3rd),and72%(4th). allcasesmorethan50%ofsupershellswerere-identified),the Wealsocomparedwhatfractionofthoseshellsthatfulfilled worstagreementforHeiles’stationaryshells. Thisisduetothe the given set of conditionswere also classified as shells by the quality of data used in the different searches. Stationary shells previousmethod (Fig. 4). The fractions are 25 % for the first were re-identifiedwith differentsizes more frequentlythan ex- setofconditions,29%(2nd),48%(3rd),and74%(4th). That pandingshells:thetraceofastationaryshellinadatacubeisless means that three fourths of shells fulfilling the strictest set of sharplydefined,lesspronounced,andnotasspecificasthetrace conditionswere also defined as shells by the previousmethod. ofanexpandingshell,andthereforeitiseasilymistakenbyboth Onefourtharenew: theseshellswererejectedbecauseofedge humaneyesandnumericalmethods. Figures2and3showone effectsinherenttothepreviousversionofthecode(describedin expandingandonestationaryshellfromthelistbyHeiles(1979) theSect. 2.3). andthecorrespondingLABidentifications. Comparisonsinthe The summary of these results follows. The less strict the outerGalaxyweremorefavourable.Thisis,webelieve,because setofconditionsusedforthenewidentifications,thebetterthe theISMintheinnerGalaxyisveryviolent,andthereisanover- agreementwegetwiththepreviousversionofthesearchingal- lapfromnearandfarregionsalongthelineofsight, leading to gorithm. Visuallyniceshellsaremuchlesssensitivetothepre- problemsin artificialconnectionsof unrelatedstructures, all of cisechoiceofconditionsandmethods. Differencesariseforthe whichmakeidentificationsintheinnerGalaxylessreliable. lesspronouncedandnotveryregularstructures. We concludethat our methodis reliable in the outerMilky By using the fourth set of conditions (∆v = 8 kms 1, min − Way,becauseweareabletoidentifythemajorityofknownex- P 0.2, P 0.5, P = 4 K), we chosestructuresthat 2,min 2,ave 3 pandingHIsupershellsandapartofknownstationaryHIsuper- havere≥gular,notwil≥dlychangingshapesthatarenoticeableand shellsagain.However,evenintheouterMilkyWaytheexistence have a large velocity extent. All previously known supershells ofsomeofthesepublishedshellsisdoubtful.Ourmethodisless belongtothisgroup,andthemajorityofshellsfromthisgroup reliableintheinnerMilkyWay. werealsoidentifiedbythepreviousmethod. Thisisthedefault All LAB identificationsconnectedto the previouslyknown settingforanyfurtherdiscussion. HIsupershellshad∆v 9kms 1,P 0.2,P 0.7,and − 2,min 2,ave ≥ ≥ ≥ P 4.0K. 3 ≥ 4. Discussion 3.3.Comparisontothepreviousversionofthesearch 4.1.Skydistributionofidentifiedshells algorithm We identify 333 shells in the LAB data, which fulfill the fol- Thepreviousversionofthesearchalgorithmwasappliedtothe lowing criteria : ∆v = 8 kms 1 (the minimum velocity min − Leiden-Dwingeloo HI survey (only the northern sky, Ehlerová span); P 0.2, P 0.5(geometricalparameters); and 2,min 2,ave &Palouš2005).WecomparedcurrentLABidentificationswith P = 4K (the≥visibilitypa≥rameter). Theirpositionsonthesky 3 ourpreviousidentifications. We restrictedourselvestothesec- andinvelocityspaceareshowninFigs. 5and6. EllipsesinFig. ondquadrant,whichwasfullyobservedinthepreviousversion 6correspondtosizesofshells. Angularsizesofshellsareshown andwhichdidnotsufferfromshellcrowdingintheinnerGalaxy. inFig. 7. Themajorityofstructuresareelongatedintheldirec- Wecomparedwhatfractionofshellsidentifiedbytheprevi- tionratherthanin b,whichmaybeexplainedasaconsequence ousmethod(277in totalin the 2ndquadrant)wasre-identified ofthelowershellcolumndensitiesinhighbdirections. by the new method with various values of search parameters OnlineTableA.1,availableattheCDS, containsthenames ∆v ,P (theaverageoverlap),P (theminimumoverlap), andpositionsofallidentifiedstructures: thenamewithcoordi- min 2,ave 2,min andP (visibilityparameter).InFig. 4weshowthecomparison natesofthecentre;andtheminimumandmaximuml,bandv , 3 lsr for wherethestructureisvisible. Articlenumber,page5of19 Fig.5. Distributionof333identifiedshellsonthesky.Leftpanel:lb-map;rightpanel:lv-map. Fig.6. IdentifiedHIshells,Hammer-Aitoffprojection,centeredonl=0◦,b=0◦. 4.2.Galacticdistributionofshells of sizes in l and b directions), the expansionvelocity v (cal- exp culated as half of the measured velocity extent), the estimated The kinematic distances of identified HI shells are derived us- energyinput L neededto create the shell based on the solution ingtherotationcurveofBrand& Blitz (1993),assuming R = ofWeaver etal. (1977),assumingthe densityfromKalberla & 8.5 kpc, V = 220 kms 1. We consider only shells in⊙the − Dedes (2008), and the approximateage of the shell t (based outer Galax⊙y. We exclude shells found at very low velocities exp onthesolutionofWeaveretal.1977). v < 10 kms 1, since identifications at these low velocities lsr − | | Figures 8 and 9 show galactic coordinates of shells and are heavily influenced by the local emission. We also disre- theirpositionsinthegalacticplane. Atgalactocentricdistances gardshellsclosetothecentreandanticentredirections(l<20◦, greaterthan19kpc,thereareshellsonlyinthesecondquadrant, l > 340◦, 170◦ < l < 190◦)andshellswithveryhighbcoordi- othersarenearlyempty. Thisisespeciallystrikingforthe third nates(b > 50◦),sincenoncircularmotionsmaybecomparable | | quadrant,whichis, fromthegeometricalpointofview,equiva- toprojectedrotationalmotions. Theseselectioncriteriaconcern lenttothesecondone. thefollowinganalysesandfiguresfromFig. 8onwardsonly. OnlineTableA.2,availableattheCDS,containsthefollow- Figure10 showsthe rv diagram,i.e. the dependenceofthe ing information about these shells: the galactocentric distance expansionvelocityoftheshellonitsradius. Theexpansionve- R, the radius of the shell r (calculated as a geometricalmean locityoftheshelliscalculatedas0.5∆v,andtheradiusisthege- sh Articlenumber,page6of19 S.EhlerováandJ.Palouš:HIshellsintheLeiden/Argentina/BonnHIsurvey Fig.7. Angularsizesofidentifiedshells. Fig. 10. rv diagram of supershells. Evolutionary lines (Weaver et al. 1977) for three luminosities are overlaid (dotted lines), as well as three lines of different times (dashed lines), for the density n = 0 0.3cm 3. Differentsymbolsdenoteestimatedenergyinputstoshells, − whenrealdensitiesaretakenintoaccount(seetextfortheexplanation): crosses for L < 0.1 SNMyr−1, open circles for 0.1 SNMyr−1 L 1.0SNMyr−1,trianglesforL>1.0SNMyr−1. ≤ ≤ Calculatedenergyinputsaregivenbydifferentsymbolsin Fig. 10. The analytical solution (Weaver et al. 1977, lines in Fig. 10) is overlaid for three different luminosities (0.1SNMyr 1,1.0SNMyr 1,10.0SNMyr 1),assumingtheISM − − − densityn = 0.3cm 3 (whichisanoverestimateforthemajor- 0 − ity of shells) and forthree differentevolutionarytimes (1 Myr, 10Myr, and100Myr). Thissolutionservesasa guideline,es- pecially for ages, since the evolutionary age in Weaver et al. (1977)doesnotdependondensityandluminosity,whilethecal- Fig.8. Positionsofidentifiedsupershells: thegalactocentricdistance RandthedistancefromtheGalacticplanez. culatedluminositydependslinearlyonthedensity. Wedetecta fewshellswithagesbelow5Myr,whichisgivenbytheresolu- tionofthesurveyandconstraintsonthesizes. Thereisonlyone shellwithagegreaterthan100Myr. Thislackofoldshellsmay be explained by dissolution due to random motions and turbu- lence ordue to destructionby spiralarms, differentialrotation, andothergalacticforces. 4.2.1. Sizedistributionofshells Weassumeapower-lawdistributionforsizesofshells dN(r ) sh = r α. (7) sh− dr A sh Wecanthencalculatetheaveragesizeofshellsr (d )atagiven sh hc heliocentric distance d using observed minimum and maxi- hc mumsizesr (d )andr (d ): min hc max hc oFfigth.e9.SunDiisstsrihbouwtinonanodftswupoecrsirhcellelsswinitthhegaGlaacltaocctiecntprliacndei.stTahnecepsoosifti8o.n5 r (d ) = RrrmminaxN(rsh)rshdrsh kpcand15kpcareoverlaid. sh hc RrrmminaxN(rsh)drsh = α−1r (d )1−(rrmmainx((ddhhcc)))α−2. (8) ometricmeanofsizesinlandbdirections. Usingtheanalytical α 2 min hc 1 (rmin(dhc))α 1 solution forthe shell evolution(Weaveret al. 1977), we derive − − rmax(dhc) − the required energy inputs for individual shells and their ages, Since we restrict the minimum and maximum angular size usingthedensitybasedonthedistributionofKalberla&Dedes on identified shells, r and r must depend on the distance min max (2008)andthewarpparametersofLevineetal.(2006).Wefind (seeFig. 11,leftpanel). With knownorassumed r , r , and sh min that 11 shells require energyinputs higherthan 1 SNMyr 1. r (assumptionsareconnectedtorestrictionsonangularsizes − max ∼ Articlenumber,page7of19 1 Q2D 0.1 0.01 8 10 12 14 16 18 20 22 R (kpc) Fig.12. 2DporosityoftheouterMilkyWay. Fig.11. Dependenceofminimum(dash-dottedline),average(solid line),andmaximum(dashedline)radiusoftheshellr asafunctionof sh theheliocentricdistanced . hc calculateσ . WeapplythesamemethodasinPaper1,briefly gsh described in the Sect. 4.2.1. It is based on selecting subsets of detectedstructures),we cancalculate α. Solutionslie in the thatcontaininformationforachosenintervalofheliocentricdis- interval(2.2,3.1),thebestsolutionisfor2.6.Thisvaluemightbe tancesandsizes. Theradialdistributionoftheshellsurfaceden- artificiallyflatterthantherealvalue,sinceouridentifyingalgo- sityΣ(R)isbasedonthenumberofshellswiththegalactocentric rithmformsonebigshellfromtwosmallshellsmorefrequently distanceRcorrectedforthelimitationscausedbyafiniterange thanviceversa,producingaflattersizespectrum. ofheliocentricdistances.Resultsdependontheselectedinterval Analternativemethodtocalculatethecoefficientα,usedin ofdistances, butin a particularway: results, whichare consis- Paper 1, is to take such a subset of identified shells, which in tent with the assumed exponential decrease (i.e. show a high a certain range of distances contains shells in a given range of correlation between observed numbers of shells with numbers sizes. Resultsofthismethodcoveranintervalofαandinclude predictedfrom Eq. 11) form a distinct group, while the others thevalue2.6calculatedhere. Uncertaintiesinthedetermination form another one with a wide spread in values of σ . Good gsh ofαarefairlylarge,probablybecauseouridentifyingalgorithm solutionsforσ havetheaveragevalueof(2.8 0.5)kpc. gsh ± isbetteratlocalizingstructuresthanatdeterminingtheirprecise After extrapolating solutions for the size distribution of sizes. shellsandtheradialdependenceoftheshellsurfacedensity,we Theaveragevalueofαinexternalgalaxies(Bagetakosetal. canestimate thetotalnumberofshells andalsothe porosity of 2011)is 2.9, which is a slightly steepervalue than ours. Their theGalaxy. Theseextrapolationsarenottooprecise,since they valueisanaveragethroughseveraltypesofgalaxiesandthereis combineallinaccuraciescomingfromerrorsinthederivationof adependenceofthecoefficientαonthegalaxytype(earlytype αandσ , andtheyalsoincludetheeffectofselectioncriteria gsh spiralstendtohavesteeperslopes). AccordingtotheirFig. 28, described in the Sect. 3. Taking the minimum size of a shell ourvalueof2.6correspondstoScgalaxies. to be 100 pc and the maximum size to be 1 kpc, we come to The coefficient α is connected to the power-law coefficient theconclusionthatthereshouldbesomethinglike400to1000 βoftheluminosityfunctionofsourcescreatingshells, e.g. OB shellsintheouterGalaxy,andabout2000to5000suchshells associations inthewholeGalaxy. The number surface density of HI shells with radii r Φ(L) L−β. (9) 100 pcinthesolarneighbourhoodisabout4+1 shellsperkshpc2≥. ∝ 2 Apartfromuncertaintiesconnectedwithextra−polating,thereare AccordingtoOey&Clarke(1997)αandβareconnectedas intrinsicvariationsinthesurfacedensity(seee.g. Fig. 9),most α=2β 1. (10) probablyconnectedtospiralarms. Thederivedvalueofsurface − densityishigherthan—thoughnotincomparableto—theaver- Therefore,β 1.8,whichissomewhatflatterthantheexpected agevalueofthesamequantityforanumberofexternalgalaxies luminosity fu∼nctionof OB associations in the Milky Way (2.0; (THINGSgalaxies,Bagetakosetal.2011). Thetypicalvalueis seee.g.McKee&Williams1997). (0.1-1)shellsperkpc2,thoughtherearesomegalaxiesinthatpa- perthatreachhighervalues. Thisdiscrepancymaybecausedby eitherthelessstrictconditionsputontheidentificationmethod 4.2.2. Radialdistributionofshells forHIshellsonoursideandstricterconditionsontheirs(among others,Bagetakosetal.(2011)demandedquiteahightempera- Weassumethattheshellsurfacedensityfollowsanexponential turecontrastfortheirshellsandregularellipticalshapes),orby distribution adifferentpointofview:frominsideintheMilkyWayandfrom Σ(R)=Σ0e−(R−R⊙)/σgsh (11) outsideforexternalgalaxies.Theinsideviewismuchmoresen- sitivetoshellsthatdidnotyetbreakthroughtheHIdisc. Onthe whereΣ(R)isthesurfacedensityofHIshells. other hand, the surface density of OB associations in the solar Typicalsizesofshells,hencetherelativeproportionofiden- neighbourhoodis about7 perkpc2 (McKee & Williams 1997), tified to total number of shells, change with heliocentric dis- which is comparable to our value: we would instead expect a tance, and therefore to get intrinsic values of Σ(R) we have to higher number of shells than OB associations, since the typi- correct the observed values. With the corrected value we can callifetime of a shellis longerthan foran OB association, but Articlenumber,page8of19 S.EhlerováandJ.Palouš:HIshellsintheLeiden/Argentina/BonnHIsurvey clustering of OB associations may decrease the numberof OB Ehlerová,S.&Palouš,J.2005,A&A,437,101,paper1 associationspershell. Elmegreen,B.G.2012,inIAUSymposium,Vol.284,IAUSymposium,317– Theporosity Q, whichis a ratioof thesurfaceoccupiedby 329 Gooch,R.E.1997,PASA,14,106 shellstothetotalsurface(the2Dporosity)orthevolumeoccu- Hartmann,D.&Burton,W.B.1997,AtlasofGalacticNeutralHydrogen piedbyshellstothetotalvolume(the3Dporosity),isderived.In Heiles,C.1979,ApJ,229,533 theinnerGalaxytheporosityascalculatedfromderiveddistribu- Kalberla,P.M.W.,Burton,W.B.,Hartmann,D.,etal.2005,A&A,440,775 tions,islarge,muchlargerthan1,whichpredictsthesignificant Kalberla,P.M.W.&Dedes,L.2008,A&A,487,951 Kalberla,P.M.W.,McClure-Griffiths, N.M.,Pisano,D.J.,etal.2010,A&A, overlappingofshells.TheporosityoftheouterGalaxyissmaller 521,A17 (see Fig. 12), where values lower than 0.01 are reached at Kerp, J., Winkel, B., Ben Bekhti, N., Flöer, L., & Kalberla, P. M. W. 2011, ∼ galactocentricdistances of 20 kpc. Local valuesat the solar AstronomischeNachrichten,332,637 ∼ neighbourhoodareQ =0.7andQ =0.4(assumingacylin- Levine,E.S.,Blitz,L.,&Heiles,C.2006,ApJ,643,881 2D 3D derwith height1 kpc)forshells with radii r (0.1,1.0)kpc. McClure-Griffiths,N.M.,Dickey,J.M.,Gaensler,B.M.,&Green,A.J.2002, sh ∈ ApJ,578,176 Error boxes are similar to those in the case of number surface McKee,C.F.&Williams,J.P.1997,ApJ,476,144 density. Porosity (especially Q3D) is mostly dependent on the Oey,M.S.&Clarke,C.J.1997,MNRAS,289,570 maximumsize of the shell, which we assume to be 1 kpc, and Thilker,D.A.,Braun,R.,&Walterbos,R.M.1998,A&A,332,429 thereforevaluesquotedhereapproachtherealvalues,sincethe Weaver,R.,McCray,R.,Castor,J.,Shapiro,P.,&Moore,R.1977,ApJ,218, 377 contributionofsmallbubblesisnegligible. Zavagno,A.,Deharveng,L.,Comerón,F.,etal.2006,A&A,446,171 Daigle et al. (2007) calculate the partial porosity of shells with radii r (5,40) pc in the region of the Perseus arm sh Q =0.007+0.0∈25. We can do the same thing by extrapolating 3D 0.003 our solutions−to small shells and get the result Q = 0.003. 3D Thisissurprisinglygoodagreement.Ifthisisnotacoincidence, itmeansthatwebothdetectthesametypeofobjects-indifferent evolutionaryphases-eventhoughweusesubstantiallydifferent approaches. Our somewhat lower value could mean that some ofthesmallobjectswillnotevolveintolargestructures(suchas bubblescreatedbysinglesupernovaexplosions). 5. Conclusions We haveidentified333HIshellsintheLeiden/Argentina/Bonn HIsurveyoftheMilkyWayusinganautomaticsearchingalgo- rithm. Theadvantageofthisapproachistheuniformityofcon- ditionsthroughouttheMilkyWay. Ourmethodisquitesensitive forlocating“somethinginteresting”,while itis notaseffective formeasuringtheprecisesizesoftheshells. Wediscoveredtheasymmetrybetweenthesecondandthird Galactic quadrants in the distribution of shells at large galac- tocentric distances. At distances greater than 19 kpc there are shells only in the second quadrant. The third quadrant, which isanequivalenttothesecondonefromthegeometricalpointof view,containsnoshellsatthesegalactocentricdistances. Because of crowding and distance ambiguity in the Milky Way, we base the following conclusions on shells in the outer Galaxy. We foundthattheirradialdistributiondecreasesexpo- nentially on a scale length of 2.8 kpc, and their size distribu- tion is a power law with the coefficient of 2.6 (correspond- ∼ ingtoapower-lawcoefficientoftheluminosityfunctionofOB associations 1.8). The surface density of shells with radii <0.1,1.0>k∼pcinthesolarneighbourhoodis 4kpc 2. − ∼ Acknowledgements. ThisstudyhasbeensupportedbytheCzechScienceFoun- dation grant 209/12/1795 and by the project RVO: 67985815. This research madeuseofNASA’sAstrophysicsDataSystem.Theauthorswouldliketothank theanonymousrefereeforhelpfulproposalsandJimDaleandtheA&Alanguage editorJoliAdamsforthehelpwiththemanuscript(anyremainingineleganceis purelyourown). References Bagetakos,I.,Brinks,E.,Walter,F.,etal.2011,AJ,141,23 Bajaja,E.,Arnal,E.M.,Larrarte,J.J.,etal.2005,A&A,440,767 Brand,J.&Blitz,L.1993,A&A,275,67 Daigle,A.,Joncas,G.,&Parizeau,M.2007,ApJ,661,285 Deharveng,L.,Schuller,F.,Anderson,L.D.,etal.2010,A&A,523,A6 Articlenumber,page9of19 A&A–shell,OnlineMaterialp10 AppendixA: Onlinetables TableA.1containsobservedpropertiesofHIshells.Column1is arunningnumberofthestructure,Col.2isthenameofstructure (GSH +galacticcoordinatesofthecentreandtheradialveloc- ity). Theremainingcolumnsdescribetherangesincoordinates, wherethestructureisvisible: Col. 3 andCol. 4the longitude, Col. 5 and Col. 6 the latitude, Col. 7 and Col. 8 the radial velocity. TableA.2containsderivedpropertiesofHIshells. Column 1isarunningnumberofthestructure,Col. 2isthenameofthe structure(GSH+galacticcoordinatesofthecentreandtheradial velocity). Col. 3 is the galactocentricdistance of the structure (usingtherotationcurveofBrand&Blitz1993),Col.4isthera- diusoftheshell(calculatedasthegeometricmeanofdimensions inlandb),Col. 5istheexpansionvelocity(calculatedasahalf ofthevelocityextentofthestructure).Columns6and7areesti- matesoftheneededenergyinputintotheshellandtheevolution- ary time based on the modelof Weaveret al. (1977). Shellsin theinnerGalaxyareexcludedfromthistable,asareshellsfound atverylowvelocities v <10kms 1,shellsclosetothecentre lsr − | | andanticentredirections(l < 20◦, l > 340◦, 170◦ < l < 190◦), andshellswithveryhighbcoordinates(b >50◦). | |