Multi-wavelength Data Analysis of ONCX-raySourcesSusanM.Hojnacki∗,Nicolas Grosso†,Giusi Micela∗∗,Donald Richards‡,NorbertSchulz§andDavid Huenemoerder§∗California Institute of Technology, 1200 East California Blvd,Pasadena, CA 91125†Observatoire Astronomique de Strasbourg, 11 ruedel’université,67000 Strasbourg, France∗∗INAF,Osservatorio AstronomicoG.S.Vaiana, Piazza delParlamento 1, 90134 Palermo, Italy‡PennsylvaniaState University,326 Thomas Building, University Park,PA16802§MITKavli Institute forAstrophysics and SpaceResearch, 70 VassarSt, Cambridge, MA 02139Abstract. To take advantageofthe increasingamount of availablemulti-wavelengthastronomicaldata,weare statistically mergingdatafromseveral wavelengthregimes to analyzeastronomicalobjects. This simultaneous analysis of emission acrossawide range of wavelengths will helpto provideacomposite understanding of young stellarobjects. Astatistical clustering techniquecoupled with fusedmulti-wavelengthdatafromthe optical,infrared, andX-raycan provideinsightinto thephysical mechanisms responsible forthe intenseemission fromyoung starsindifferentwavelengthregimes andcan be used to view trends andcorrelations between thoseregimes.Wepresentsome details of thedatafusionfollowed by our results thus faranalyzing spectraldatafrompre-main sequence(PMS) starsinthe OrionNebulaCluster (ONC).Keywords: <multi-wavelength; young stellarclusters; X-rayemission; statistical methods>PACS: 97.10.Bt,97.21.+a, 97.10.-q,95.75.PqINTRODUCTIONThe nature andoriginofthe oftenvariable-intensityemission fromyoung, low-massstars is an area of intense research in recentyears.The debate overthe physical mecha-nismssurrounding emissionishavingasignificantimpact on our understanding of thenature of interactions between PMSstars andtheir planet-forming disks,the timescaleoverwhich planetsformaround Sun-likestars,and thehigh-energy environmentoftheearlysolar system. Despite the vast amount of data nowavailablefromprevious andpresentgenerations of optical,infrared(IR), andX-ray facililities, statistically integrat-ingand combiningthisdatafor analysisisnot commoninthe astronomical communityandhas not yetbeen undertaken forspectral classificationofsources in starformationregions. We have statistically combined data fromthe optical,IR, andX-ray bands todeterminewhatinsightscan be gained from using suchamulti-wavelengthapproach.Ourresearch anddataanalysis of thefusedmulti-wavelengthdatahas thepotentialtoidentifyspectral outliers of interestamong thesources,classifysources faintinonewavelengthbandbycombinationofdatafromother wavelengthregimes,determineifspectral diagnosticscan be established that can distinguish between potentialemissionmechanisms, andsearch fortrends andcorrelations between wavelengthregimes.579Downloaded 15 Apr 2011 to 131.215.220.185. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions CREDIT LINE (BELOW) TO BE INSERTED ON THE FIRST PAGE OF EACH PAPER EXCEPT THE PAPER ON PP. 525-530 CP1248, X-ray Astronomy-2009: Present Status, Multi-wavelength Approach and Future Perspectives edited by A. Comastri, M. Cappi, and L. Angelini © 2010 American Institute of Physics 978-0-7354-0795-4/10/$30.00 CREDIT LINE (BELOW) TO BE INSERTED ON THE FIRST PAGE OF ONLY THE PAPER ON PP. 525-530 CP1248, X-ray Astronomy-2009: Present Status, Multi-wavelength Approach and Future Perspectives edited by A. Comastri, M. Cappi, and L. Angelini 2010 American Institute of Physics 978-0-7354-0795-4/10/$30.00DATATo obtainabaselinefor theanalysis,the initialrequirementwas that data exist inoptical,IR, andX-ray bands foreach source.Therefore acompletecaseanalysis wasperformed with asampleinitially limitedtoT-Tauri stars.The classificationalgorithmwasusedsolelyonthe spectral informationfromobservations of theONC.Datahasbeen collected from theHST TreasuryCycle program forthe ONC [2]for optical bandsB, V, H-alpha,I,and z-band. J, H, andKband data is fromCTIO[4].X-ray data isfromthe ChandraOrion Ultradeep Project (COUP,[5]). The resultisaconservative,“uncontaminated”, robust sample of 213 ONC PMSsources.PROCEDURE AND ANALYSISPrincipalComponent Analysis(PCA)was usedtoreducethe dimensionality of thedatasetand remove correlationbetween adjacentbands. PCA can uncovernew variables,calledprincipal components(PCs), whichaccount forthe majority of thevariance in thedata.There are no parameters or models to tweak andnocoefficients to adjust basedonuserexperience.The answer is basedonthe data itself andisindependent of theuser.PCA wasrun on thedataset of 5optical,3IR,and 35 X-ray spectral bands forthe213 sources in thesample. The first four PCs explain87.9% of thevariance andwereretained. Twounsupervisedclustering methods, agglomerativehierarchical clusteringandK-means clustering, were appliedsequentially to the4PCs to determinea“natural”partitioning of thesampleintoanumberofrelativelyhomogeneous groups or “clusters”of sources [1]. Thisyielded 18 clusters in thefinalpartitionofsources.The spectral cluster averages were computed foroptical andnear-IRpropertiesofthepopulationofthe ONC forsources where thesedataexist.The spectral clusters formsequences in logabsorbing column density (log NH)and logeffectivetemperature(logTef f). These propertieshavebeen determined independently of our research [5].Infraredexcess, an accretiondisk indicator,was examined using an indexbased onIR colors. △(H − K) wasusedratherthan △(I − K) to avoiderrors introduced byphotometricvariability since theoptical andIRdatawere not obtainedsimultaneously.The spectral clusters form asequencein△(H−K)(see Fig. 1).Fig. 2shows therelationshipbetween log NHandPC1 forthe sources in thesample.There are twoprominent outliers in Fig. 2: COUP 948 andCOUP 224. COUP 948 isan optical jetsource calledthe BeehiveProplyd. The spectruminthe X-ray hasaveryunusual, double-peaked spectral energy distributionwith prominentsoftcomponentsindicativeofstrong shocks in thejet collimationregion(Kastneretal. 2005).COUP224 is aheavily-veiled, hard X-ray source with strong Fe XXV emissionat6.7 keV.It is aprobableM-type source with equivalent widthoftheλ8498 Ca II lineof36Å,indicatingpossible vigorous accretion[6].Fig. 3shows aplotofmean log NHversus mean EM weighted-averageofthe thermalplasma temperaturefor each cluster with thesedata. Atrendcan be seen betweencluster positiononthe plot andthese parameters.The trends between cluster
View Full Document
Unlocking...