reprint from ApJ

15 Pages·2015·0.96 MB·English

TheAstrophysicalJournal,801:61(15pp),2015March1 doi:10.1088/0004-637X/801/1/61
(cid:2)C2015.TheAmericanAstronomicalSociety.Allrightsreserved.
CORRECTINGTHERECORDONTHEANALYSISOFIBEXANDSTEREODATA
REGARDINGVARIATIONSINTHENEUTRALINTERSTELLARWIND
P.C.Frisch1,M.Bzowski2,C.Drews3,T.Leonard4,G.Livadiotis5,D.J.McComas5,6,
E.Mo¨bius4,N.Schwadron4,andJ.M.Soko´łl2
1DepartmentAstronomyandAstrophysics,UniversityofChicago,Chicago,IL60637,USA
2SpaceResearchCentreofthePolishAcademyofSciences,Warsaw,Poland
3InstituteforExperimentalandAppliedPhysics,Christian-Albrechts-UniversityzuKiel,Germany
4SpaceScienceCenter,UniversityofNewHampshire,Durham,NH03824,USA
5SouthwestResearchInstitute,SanAntonio,TX78249,USA
Received2014June29;accepted2014December29;published2015March4
ABSTRACT
ThejourneyoftheSunthroughspacecarriesthesolarsystemthroughadynamicinterstellarenvironmentthatis
presentlycharacterizedbyaMach∼1motionbetweentheheliosphereandthesurroundingwarmpartiallyionized
interstellar cloud. The interaction between the heliosphere and interstellar medium is an evolving process due to
variable solar wind properties and the turbulent nature of the interstellar cloud that surrounds the heliosphere.
Frischetal.presentedameta-analysisofthehistoricaldataontheinterstellarwindflowingthroughtheheliosphere
and concluded that temporal changes in the ecliptic longitude of the flow direction with time were statistically
indicatedbythedataavailableintherefereedliteratureatthetimeofthatwriting.Lallement&Bertauxdisagree
with this result, and suggested, for instance, that a key instrumental response function ofIBEX-Lo was incorrect
and that the STEREO pickup ion data are unsuitable for diagnosing the flow of interstellar neutrals through the
heliosphere.Inthispaperwefirstshowthattemporalvariationsintheinterstellarwindthroughtheheliosphereare
consistent with our knowledge of the very local interstellar medium. The statistical analysis of the helium wind
dataisrevisited,andarecentcorrectionofatypographicalerrorintheliteratureisincorporatedintothenewfits.
Withthiscorrection,andincludingnonewerIBEXresults,thesecombineddatastillindicatethatachangeinthe
longitude of the interstellar neutral wind of λ = 5◦.6±2◦.4 over the past forty years remains statistically likely,
but an constant flow longitude is now statistically possible. Other scenarios for the selection of subsets of these
datausedinthefittingprocessproducesimilarconclusions.WeshowthatthespeculationsmadebyLallement&
BertinabouttheIBEXinstrumentalresponsefunctionareincorrect,andthattheirotherobjectionstothedataused
inthemeta-analysisareeitherincorrectorunproven.Furtherinvestigationsofthehistoricalinterstellarwinddata
andcontinuinganalysisofadditionalIBEXdatamayprovideamoredefinitiveresultonthestabilityoftheflowof
interstellargasthroughtheheliosphere.
Key words: ISM: structure – methods: data analysis – shock waves – solar wind –
spacevehicles:instruments–turbulence
historicaltimescales,ofabout40yearsforthespaceage,suggest
1.INTRODUCTION
thatinterstellarneutrals(ISNs),andthepickupions(PUIs)that
Thesolarsystemistravelingthroughadynamicallyevolving form from them after ionization, provide the best diagnostics
interstellar medium that leads to variations in the interstellar of the longitude of the interstellar wind flowing through the
wind through the solar system over geologically short time- heliosphere. The dynamic characteristics of the surrounding
scales (Frisch et al. 2011; Frisch & Mueller 2011). The Local LIC suggest that short-term variations in the interstellar wind
Interstellar Cloud (LIC) now around the heliosphere is a low- directionmayexist.
densityandpartiallyionizedcloud(Slavin&Frisch2008).The In the paper titled “Decades-Long Changes of the Inter-
relativemotionsoftheSunandLICcreatesawindofinterstellar stellar Wind Through Our Solar System,” Frisch et al. (2013,
gas and dust through the heliosphere whose velocity was first hereafter Paper I) have presented a meta-analysis of historical
defined by measurements of the fluorescence of solar 584Å observations of the interstellar Heo wind data and found that
photonsoffofinterstellarHeointhegravitationalfocusingcone the data indicated a statistically likely shift in the wind lon-
downwind of the Sun (Weller & Meier 1974). The absence gitude of ∼6◦.8 over the past forty years, and that a constant
of interstellar Ho at the LIC velocity toward the two nearby flow longitude is statistically unlikely. The historical data set
stars α Cen (1.3pc) and 36 Oph (6.0pc) indicate that the Sun included in situ measurements of the helium wind by IBEX
will emerge from the LIC in the upwind7 direction anytime (McComas et al. 2012; Mo¨bius et al. 2012; Bzowski et al.
withinthenext3500years(Lallementetal.1995;Woodetal. 2012) and Ulysses (Witte 2004), measurements of PUIs in
2000),andthatitenteredtheLICduringthepast10,000years thedownwindgravitationalfocusingconebytheACE-SWICS,
(Frisch1994).Variationsintheinterstellarwinddirectionover STEREO-PLASTIC, and MESSENGER-FIPS (Gloeckler et al.
2004; Drews et al. 2012; Gershman et al. 2013), the up-
6 AlsoatUniversityofTexasinSanAntonio,SanAntonio,TX78249,USA. windcrescentbySTEREO-PLASTIC(Drewsetal.2012),and
7 Theterms“upwind”and“downwind”refertothedirectionofthe measurements of the backscattered radiation from solar Heo
heliospherenose,whichinturnisdefinedbythedirectionoftheinterstellar 584Å emissions fluorescing from interstellar Heo in the inner
Heowindthatflowsintotheinnerheliosphere.
1
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
heliosphere (Weller & Meier 1974, 1981; Ajello 1978; Ajello 2.POSSIBLESPATIALINHOMOGENEITIESINLIC
et al. 1979; Dalaudier et al. 1984; Vallerga et al. 2004;
2.1.TurbulenceandEdgeEffects
Nakagawa et al. 2008). This was the first meta-analysis that
searchedforpossiblehistoricalvariationsinthedirectionofthe TheSunmovesthroughthesurroundinginterstellarcloudata
windofheavyISNatomsthroughtheheliosphere. relativevelocityof∼5AUyr−1,sothattheforty-yearhistorical
The original study in Paper I was restricted to heavy atoms recordoftheinterstellarwindvelocitysampledinterstellarscale
anddidnotincludelongitudeinformationfromthevelocityvec- lengths of ∼200AU. There is no ad hoc reason that the LIC
tors of either interstellar Ho or interstellar dust grains flowing should be homogeneous and isotropic over such small spatial
throughtheheliosphere.Bothsetsofparticlesarestronglyfil- scales.ThemeanfreepathofathermalpopulationofLICatoms
tered on a time-variable basis, because of the solar magnetic is∼330AU,9whichislargerthanthedistancetraveledthrough
activity cycle, throughout the heliosphere and heliosheath re- theLICduringthepast40years.Collisionalcouplingbetween
gions (Ripken & Fahr 1983; Rucinski & Bzowski 1996; Saul atoms will break down over scales smaller than the mean free
et al. 2013; Grogan et al. 1996; Landgraf 2000; Slavin et al. path, allowing the formation of eddies that perturb the gas
2012;Sterkenetal.2012;Strubetal.2014). velocity over heliosphere scale-lengths. Paper I suggested that
Inapaper“OntheDecades-LongStabilityoftheInterstellar ashiftintheHeoflowdirectioncouldbeduetothenonthermal
Wind through the Solar System,” Lallement & Bertaux (2014, turbulenceofeddymotions.
hereafter LB14) disagree with the conclusions in Paper I. The A second source of small-scale turbulence in the LIC is
different viewpoints presented in Paper I and LB14 fall into Alfvenicturbulence.TheLICinterstellarmagneticfield(ISMF;
twocategories,themostinterestingofwhichistheunderlying McComas et al. 2009; Funsten et al. 2013) affects clumping
science question that is encapsulated by the differences of the ofthechargedparticlesoverscalessmallerthanthecollisional
two titles On the “Decades-Long Changes of the Interstellar meanfreepath.TheLICmagneticfieldstrengthis∼3μG,based
WindThroughOurSolarSystem”comparedto“Decades-Long on the plasma pressure of energetic neutral atoms (ENAs) in
StabilityoftheInterstellarWindthroughtheSolarSystem.”The theupwinddirectionandthemagneticdistortionoftheheliotail
second area of disagreement centers on the selection of data (Schwadronetal.2011),andtheequipartitionofenergybetween
used in the analysis, and the IBEX instrumental parameters. magnetic and thermal energies in the LIC (Slavin & Frisch
LB14 favor the interstellar Heo flow direction determined in 2008).Forfieldstrengthsof3μGandlarger,theLorentzforce
Mo¨bius et al. (2004), where measurements of the interstellar will couple protons tightly to the ISMF, over ∼375 km scales
Heo windfoundbyUlysses,ACE,EUVE,andPrognoz6were for a gas temperature of 7000 K. This tight coupling acts as
combined to find the best-fitting direction for the interstellar aconduitformagneticturbulenceintotheLICthataffectsthe
wind.Thegroundworkforunderstandingtheinterstellarhelium neutralpopulationobservedbyIBEXthroughchargeexchange,
wind is presented in the set of papers that accompany Mo¨bius and would perturb the neutral particle velocities over spatial
etal.(2004).8 scalessmallerthanthecollisionalscale.
In this paper we first revisit the conclusions of Paper I, The location of the LIC in a decelerating group of clouds
with the proviso that new data on, and better modeling of, that appears to be a fragment of the Loop I superbubble shell
the interstellar wind are creating a rapidly changing scientific (Frisch et al. 2011) would promote turbulence in the local
data base upon which our conclusions rely. Data showing that ISMF. If the surrounding decelerating flow of ISM (Frisch
the interstellar wind could change over historical times are etal.2002,2011)consistsofcloudsincontactwitheachother,
discussed, and these data are consistent with inhomogeneities then macro-turbulence may form in the LIC at the acoustic
in the LIC over spatial scales comparable to the heliosphere velocity of about 8 kms−1. The angle between the LIC ISMF
dimensions(Section2).Newstatisticalanalysesareperformed andtheflowofLICgasthroughthelocalstandardrest(LSR)is
on the historical data (Section 3), utilizing a new published 86◦±14◦,indicatingthatthevelocityandmagneticfieldvectors
correction for the uncertainties on the NOZOMI interstellar are perpendicular (Frisch & Schwadron 2014). Depending
Heo wind direction (Section 4.3.3). A temporal change in the on the relative thermal and magnetic pressures, compressive
windlongitudeisfoundtobestatisticallylikely,butaconstant turbulencemayform.Magneticturbulencemaypropagatealong
wind direction also becomes statistically acceptable although thebridgeofpolarizeddustthatextendsfromneartheSunout
less likely than a variable direction. LB14 objected to the use to the North Polar Spur region (Frisch et al. 2014). Magnetic
oftheIBEX-LodatainrecoveringalongitudefortheHeo flow disturbanceswillpropagatealongtheISMFattheAlfvenspeed
that differed from the value found by Ulysses. Methods for of∼25kms−1 (usinganelectrondensityof0.07cm−3,Slavin
retrievingtheflowlongitudefromIBEX-Lodataarediscussed & Frisch 2008), organizing additional disruption of particle
inSection4.1,withfurtherdetailsontheinstrumentalresponse velocitiesoverscalesizessmallerthanthemeanfreepath.
functionandpropertiesinSection4.1.3andAppendixB.Cloud It is important to note that the thermal distribution of the
temperature plays an important role in the uncertainties of Heoatomssmoothsthetime-intervalsampledbythepopulation
the parameter space constrained by the IBEX-Lo data; the observed at 1AU because of the different propagation times
interstellar picture of the LIC temperature is discussed in offastandslowatomsthroughtheheliosphere.AMaxwellian
Appendix A. The LB14 comments on the STEREO data are gas at 7000 K has a Doppler width of ∼5 kms−1, which is
discussed in Section 4.2, where new data are mentioned that asignificantpercentageofthebulkLICmotion,causingfaster
contradict the hypothesis of significant azimuthal transport of particlestocatchuptoslowerparticlesintheincomingHeoflow.
PUIs inside of 1AU, and in Appendix C. Table 1 summarizes The Maxwellian distribution will not apply over microscales
thedataissues. lessthanthe330AUmeanfreepathofcharge-exchange,andin
principlethemeanfreepathsofthehigherenergyatomswillbe
largerthan330AUanyway.
8 Thewinddirectioninthisstudyneedstobecorrectedby0◦.7toaccountfor
anunrecognizeddifferenceincoordinateepochsofthecontributingdata 9 ThemeanfreepathintheLICisdominatedbycharge-exchangeand
(Frischetal.2009;Lallementetal.2010). induceddipolescatteringbetweenHoandH+atoms(Spangleretal.2011).
2
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
Table1
CommentsonLallement&Bertaux(2014)CriticismsofDataandUsageinPaperIa
Argument Comment Section
(Appendix)
CriticismsregardingIBEXanalysis
1 UlyssesandIBEXvelocity–longitude TrueforhotmodelsusingMaxwellian 4.1.14.1.2(B)
combinationsgivesameperihelion distributions,butirrelevanttoresults
2 IBEX-Loinstrumentalresponse Incorrect;theIRC,dead-timecorrection, 4.1.3(B)
correction(IRC)iswrong anddatathroughputarewellunderstood
3 CorrectIRCmakesIBEX Incorrect;theIBEXresultsrelyon 4.1.1
andUlysseslongitudesagree variationsofthepeakasafunctionof
latitudeandobserverlongitude
4 Warsawmodelingbasedon Incorrect;Warsawmodelscomparedto 4.1.3
wrongIRCcorrections databasedonacorrectandtestedIRC
CriticismsregardingSTEREOanalysis
5 Short-termelectron-impactionization Incorrect;short-termEIIisdistinguished 4.2,4.2.2
(EII)notincludedindataanalysis fromlong-termionizationsinmethodology
6 Orbitgroupingsarenot Incorrect;eachorbitpairingselects 4.2(C)
statisticallyindependent invariantdistributionsfromcomparisons
ofdataacquiredoverseparatetime-spans
7 Simulationsofuncertaintiesof Uncertaintysimulationsare 4.2(C)
STEREOdataareincorrect appropriateforvariablePUIdata
8 Secondaryoxygeninvalidatesusage Incorrect;secondaryoxygenpopulation 4.2.3
ofoxygenPUIs minorcomparedtoprimarypopulation,and
ifincludedpushlongitudesintothe
oppositedirectionthansuggested
9 LongitudesfromPUIfocusingcone Maybe,buttransporteffectsappeartobe 4.2.1
andupwindcrescentareupperlimits insignificant,basedoncomparisonsbetween
duetotransporteffectsinside1AU severaldifferentdatasets
CriticismsregardinguseofUltravioletData
10 Mariner10longitudeisbasedon Incorrect:Mariner10papersquote 4.3.1
bothHoandHeo longitudeofHeowindseparately
11 HeoflowdirectionfromPrognoz6 ThePrognoz6Heodirectionfromgeometric 4.3.2
geometricanalysisnotvalid argumentshasuncertaintiesandwasincluded
sinceitwasnotrepudiatedintheliterature
12 NOZOMIuncertaintiesof±3◦.4are True;theresultsofthe2014NOZOMI 4.3.3
toosmall erratumareusedinthispaper
Note.aNumbersinlastcolumnshowsectionorappendixwherethetopicisdiscussed.
Upper limits on an interstellar Lyα component at the LIC inthedownwinddirection,ortheG-Cloud(GC)intheupwind
velocitytoward36Oph,intheupwinddirection,showthatthe direction,10theshocksthatformatthesupersonicinterfaceswill
Sun is near the LIC edge (Wood et al. 2000). The high level drive compressional disturbances into the LICthat perturb the
ofLICionizationthatisinferredfromphotoionizationmodels densityandvelocityofthegasparticles.
requires that 22%, 39%, 0.80%, and 19% of the hydrogen, A supersonic interface would form between the LIC and
helium,neon,andoxygenatomsattheheliosphereboundaries GC if they are in contact. The angle between the LIC and
are ionized (Slavin & Frisch 2008, SF08). A thin conductive GC LSR velocity vectors is 17◦.2, and the relative velocity is
interface between the warm LIC and hot plasma of the Local 14.6kms−1.TheG-cloudhasatemperatureof5500±400K
Bubblemaintainstheseionizationlevels(SF08).Asharpdensity (Redfield&Linsky2008),correspondingtoanacousticvelocity
decreaseandvelocityincreaseisexpectedattheedgebetween of ∼7.7 kms−1, indicating a shock strength of Mach ∼1.9
theLICandconductiveinterface.Theunknowndistancetothe betweentheclouds.11
upwindedgeoftheLICallowsthatsuchaninterfaceiscloseto AsupersonicinterfacewouldalsoformbetweentheLICand
the heliosphere. However, no direct observational evidence of theBCiftheyareincontact.ThesecloudshaveLSRvelocities
thisinterfaceisyetfound. of 15.3±2.5 kms−1 and 7.1±2.3 kms−1 respectively, and
2.2.SupersonicCollisionsbetweenClouds
10 Anindependentsortingoflocalinterstellarvelocityabsorptioncomponents
intocloudsisprovidedbyRedfield&Linsky(2008)andweusethat
Disturbances due to the propagation of shocks through the
cloud-namingschemehere.
cloud are also possible sources of kinematical structure in the 11 UsinginsteadtheheliocentricvelocitiesfortheLICandGCintheupwind
LIC. The LIC belongs to a decelerating flow of ISM through direction,andnow-obsoletephysicalparametersforbothclouds,Grzedzielski
space that may create dynamical interfaces in the interaction &Lallement(1996)havepredictedthatasupersonicinterfaceispresent
betweentheLICandtheG-cloud,withtheLICcorrespondingtothe
regionsbetweencloudsinphysicalcontact(Frischetal.2011).
post-shockgasbecauseitiswarmerthantheG-cloud.Frisch&York(1991)
IftheLICisinphysicalcontactwitheithertheBlueCloud(BC) alsomodeledthelocalshockstructure.
3
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
Table2
FitstoHistoricalInterstellarWindData
Num. Basis Fita Reduced p-value
ofFit λ(deg.) χ2
A.PaperI:b
1 Linear 70.6(±1.6)+0.17(±0.06)∗t1970 0.97 0.49
2 Constantλ 75.1(±1.3) 1.71 0.031
B.CorrectedNOZOMIuncertainties:c
3 Linear 71.3(±1.6)+0.14(±0.06)∗t1970 0.7d 0.82
4 Constantλ 75.0(±0.3) 1.1 0.35
C.TwoIBEXdatapoints,modified584Åuncertainties:e
5 Linear 71.4(±1.6)+0.14(±0.06)∗t1970 0.8 0.3
6 Constantλ 75.0±0.3 1.3 0.2
7 Parabolicλfit 74.1(±1.4)−0.11(±0.11)∗t1970+5.0(±2.1)10−3∗t12970 0.6 0.14
Notes.
aλistheeclipticlongitudeoftheinterstellarwindandt1970istheelapsedtimesince1970.Ap-valuelargerthan
0.05passestheptestandthereforeisnotunlikely.
bThisfitistheoriginallinearfitinPaperIthatusedtheNOZOMIvaluefromNakagawaetal.(2008).
c ThesefitsusedthehistoricalwinddatainPaperI,exceptthattheuncertaintiesfortheNOZOMIvalueare
correctedtoincorporatetheerratumofNakagawaetal.(2014).Thislinearfitshouldreplacethefitlistedin
PaperI.SeeSection4.3.
dThisfitisacceptedandismorelikelythanfitnumber4.Thesmallvalueofχ2suggeststhattheuncertainties
maybeoverestimatedby∼20%.
eThesefitsincorporateuncertaintiesforthewindlongitudefortheSTP2-1,Mariner10,andSOLRAD11Bdata
thatincludeonlythelongitudeuncertainties(seeSection4.3).ThegeometricPrognoz6directionisomittedfrom
thefitaccordingtothediscussioninSection4.3.TheHeo winddirectionsfromtheindividualIBEX2009and
2010observingseasonsareincludedindependently(Sections4.1.2,3,andseeBzowskietal.2012).
ananglebetweenthetwovectorsof49◦ (cloudLSRvelocities required to pass the p test12 so that this result is highly likely.
are from Frisch & Schwadron 2014). The LIC sound velocity The small value for χ2 suggests that the uncertainties may be
is 8.6 kms−1, for a 7000 K perfect gas with mean molecular overestimated by about ∼20%. The corresponding fit for the
weight 1.29. The ∼11.9 kms−1 relative velocity indicates a assumptionofaconstantflowlongitudegivesλ = 75.0±0.3
supersonic interface of Mach number ∼1.4 between the two (fit 4), which has a p-value of 0.014 and is now a statistically
cloudsiftheyareindirectcontact.TheLICiswarmerthanthe acceptablefit,butwithalowerlikehoodthanafitwithavariable
BC,whichhasatemperature3000+−21000000K(Hebrardetal.1999). flow direction. This new fit to the data using the corrected
A simple temperature gradient between the three clouds does NOZOMI uncertainties should replace the fit that is presented
notexist.InthepresenceoftheISMF,theshocktransitionswill inPaperI.
bemodifiedbytheunknownanglebetweenthemagneticfield Inordertotesttherobustnessofthefittothesedata,wehave
andthenormaltothecloudsurface. performedasecondfitwithfouradditionalmodificationsmade
totheinputdataset:
1. The uncertainties assigned to the STP 72-1, Mariner 10,
3.REFITTINGTHEHISTORICALDATAONTHE and SOLRAD11B flow directions in Paper I represented
INTERSTELLARWINDLONGITUDE thecombinedlongitudeandlatitudeuncertaintiesreported
in the original publications (Section 4.3). The new fit
Anewstatisticalinterpretationofthehistoricaldata(Table1 uses only the longitude uncertainty. It is assumed that
in the online supplementary material, OSM, of Paper I) is the uncertainties for longitude and latitude are normally
required by a typographical error in the original NOZOMI distributed,sothattheuncertaintyusedinTable1ofPaper1
publication (Nakagawa et al. 2008). We first perform new fits canbereducedbyafactorof1.414.
to the same set of data used in the Paper I using the corrected 2. TheIBEXandUlyssesdataaretreatedequivalentlyinthe
uncertainty for the Heo flow longitude determined from the sense that the two independent data sets corresponding
NOZOMIdata(Nakagawaetal.2014,Section4.3.3),andnext to the 2009 and 2010 seasons are included separately,
foranalternateselectionofthefitteddata. ratherthancombinedasinPaperI.Thelongitudesforthe
The results of the refitting are presented in Table 2. The individualseasonsfromBzowskietal.(2012)areused.
firstresult,headingA,showstheoriginalfitfromPaperI.The
updatedfitusingthecorrectedNOZOMIuncertaintiesislisted 12 AspointedoutinS5oftheOSMinPaperI,theprobabilityoftakinga
underheadingB,fits3and4,ofTable2,andplottedinFigure1. resultχ2moreextremethantheobservedvalueofχ2 isgivenbythep-value
obs
The slope of the revised fit is 0.14 ± 0.06, which implies a thatequalstheminimumbetweenthetwoprobabilities,P(0(cid:2)χ2(cid:2)χ2 )
obs
directional change of δλ = 5◦.6±2◦.4 over the past 40 years andP(χo2bs(cid:2)χ2(cid:2)∞).Anullhypothesisassociatedwithap-valuesmaller
(instead of the 6◦.8±2◦.4 in Paper I). The reduced χ2 of the thanthesignificancelevelof0.05istypicallyrejected.Additionalinformation
aboutthestatisticaltechniquesusedinthefitscanbefoundLivadiotis(2014,
fit is 0.7, and the p-value of 0.82 is larger than the 0.05 value 2007),Deming(1964),Lybanon(1984).
4
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
Figure1.Left:thefitusingthecorrectNOZOMIuncertaintiesisshownbythedashedline,andthedottedlineshowstheoriginal(PaperI)fitwiththeuncorrected
NOZOMIuncertainties.Theshadedregionshowstheuncertaintiesforthenewfit.TheuncertaintiesontheIBEXdatapoint(reddotinupperright)areconstrainedby
theIBEXparameterrangeandtheLICtemperaturetowardSiriusbasedonFe+,Mg+,andCa+data(PaperI,OSM-S4).Right:samefitasontheleft,exceptthatthe
Prognoz6dataareomitted,theuncertaintiesfortheSTP72-1,Mariner10,andSOLRAD11Bdataarerevisedtoincludeonlythelongitudeuncertainty,andseasons
2009and2010oftheIBEXdataarelistedasindependentdatapoints.TheIBEXlongitudeintherighthandfiguredoesnotdependonanyconstraintsthatarebased
ontemperaturesobtainedfromLICabsorptionlines.ThesefitsareexplainedfurtherinSection3.
3. The uncertainties for the IBEX longitudes are based on New reductions and modeling of the Ulysses data have
theIBEXdataalone,withoutrecoursetoastronomicaldata appearedsincethepublicationofPaperI(Bzowskietal.2014;
on the LIC. The IBEX temperature uncertainty range of Wood et al. 2014; Katushkina et al. 2014). These results find
4400–8200K(Section4.1.2)canbeusedtosetlongitude a Heo flow longitude for the Ulysses data collected during the
uncertaintiesof±4◦.2(Figure1ofMcComasetal.2012).In 2007 polar pass that is close to the value of Witte (2004), but
PaperIaLICtemperaturebasedonastronomicaldatawas with a somewhat higher temperature (7500+1500 K; Bzowski
−2000
used,howevertheastronomicaldatareveallargevariations etal.2014).ThedifferencesbetweentheHeoflowvectorfound
intheLICtemperaturetowarddifferentstars(AppendixA). fromtheIBEX2009–2010seasonsofdata(Mo¨biusetal.2012;
4. ThePrognoz6dataareomittedfromthisfitbecauseboth Bzowski et al. 2012; McComas et al. 2012), and the Ulysses
thegeometricfitandthefitobtainedfrommodelingthedata vector, are not yet explained. For the latest seasons of IBEX
arenolongerconsideredvalidbycoauthorsoftheoriginal observations(2012–2014;McComasetal.2014;Leonardetal.
papers(Section4.3.2). 2014), the results from spacecraft pointings in the ecliptic
suggest a different portion of the coupled four-dimensional
Three fits were performed on this adjusted data set,a linear “tube”ofpossiblesolutionspresentedbyMcComasetal.(2012)
fit,afitassuminganconstantflowlongitude,andaparabolicfit with similar Heo velocity vectors to Ulysses but requiring a
(headingC,fits5,6and7).Thegoalofthefitsistotestwhethera significantlywarmercloud;othernewIBEXpointingdirections,
temporalvariationintheflowdirectionovertimeisstatistically atdifferentanglestotheeclipticplane,arenotconsistentwith
morelikelythanaconstantflowdirectionovertime.Thelinear theUlyssesdataforreasonsnotfullyunderstood.Comparisons
fits listed in Table 2 are highly likely and always more likely ofUlyssesandIBEXresultsareastudyinprogressandwedo
thanafitwithaconstantlongitudeovertime.Thelowχ2value not include results presented after 2013 in the refitting of the
oftheparabolicfitsuggeststhatthelimitedsetofavailabledata datainthispaper.
does not justify a second-order fit. Note that the finding that a
linear fit to the data is statistically more likely than a constant 4.DATAUSEDFORDETECTINGTEMPORAL
orparabolicfitisnotequivalenttoastatementthatachangein VARIATIONSINTHENEUTRALHELIUMWIND
longitudemusthaveoccurredinalinearfashion.
The linear fit is statistically the most likely fit and gives a LB14 have argued that most of the data incorporated into
temporal variation of 5◦.6±2◦.4 over the past 40 years (the fit themeta-analysisofPaperIwereeitherincorrectorwereused
isshowninFigure1,right).Thisresultisclosetotheresultin incorrectly.Thesedata,allofwhichweretakenfromtherefereed
PaperI.Howeverwhileaconstantlongitudecouldberejected scientificliterature,fallintothreegroups:remotemeasurements
inPaperI,theincreaseintheNOZOMIuncertaintiesmakesa ofthebackscatteredsolar584ÅemissionfrominterstellarHeo
constant longitude over time acceptable but less likely than a in the inner heliosphere (Weller & Meier 1974, 1979, 1981;
variationofthelongitudeovertimeLivadiotis(2014). Meier 1977; Broadfoot & Kumar 1978; Ajello 1978; Ajello
Theparabolicfitisincludedtoshowthatafitwithahigher et al. 1979; Dalaudier et al. 1984; Flynn et al. 1998; Vallerga
orderpolynomialyieldsaresultthatislesssatisfactorythanafit etal.2004;Lallementetal.2004;Nakagawaetal.2008,2014);
withalowerorderpolynomial.Astheorderofthepolynomial insitumeasurementsofthepickup-upionsthataresampledby
inthe fitincreases, theχ2 should inprinciple decrease untilit spacecrafttraversingthegravitationalfocusingconeandupwind
is <1 or it is χ2 ∼ 1. The parabolic fit under heading C gives crescent (Gloeckler et al. 2004; Drews et al. 2012; Gershman
χ2 ∼ 0.6, which tells us that the best statistical description etal.2013),andinsitumeasurementsofinterstellarHeoatoms
of the temporal variation in the flow direction is given by the (Witte 2004; Mo¨bius et al. 2012; Bzowski et al. 2012, 2014;
linearfit. McComasetal.2012).
5
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
Figure2.HyperbolictrajectorygeometryintheplaneoftheISNbulkflow.Thelocationoftheperihelion(orflowtangentialtotheEarth’sorbit)isdegenerateinall
combinationsVISM,∞(λISM,∞)thatsatisfyEquation(1).Thesolidandthedashedbluelineindicatetwosampletrajectoriesfrominfinitytotheirperihelionatθ =0◦
thatsatisfytherelation,butarrivefromdifferentinflowlongitudesλISM,∞asindicatedbythesolidanddashedblacklines.λObsisthelongitudeoftheobservations,
θ∞istrueanomalyoftheinflowtrajectory(blueline)frominfinitytoperihelionatrE=1AU.Thedashedbluelineshowsthetrajectoryofanatomwithadifferent
θ∞–VISM,∞combinationthatsatisfiesEquation(1).
Table 1 summarizes the basic LB14 arguments and our maximum observed by IBEX along the Earth orbit λ was
Peak
response to those arguments. Most of the arguments are determined to be 130±0◦.7 in (Mo¨bius et al. 2012), which
speculative. The LB14 hypothesis about an underestimated constrainsthefunctionalrelationshipinEquation(1)towithin
IBEX-Loinstrumentresponsefunctioniswrong(Section4.1.3, (cid:2)±0◦.7 in λISM,∞ and (cid:2)±0.5 kms−1 in VISM,∞, although a
AppendixB).Thebetter-supportedargumentthattheSTEREO large range of values along the function satisfy the observed
PUI data only provide upper limits because of the azimuthal location of the interstellar gas flow maximum at 1AU nearly
transport of PUIs inside of 1AU has since been shown to be aswell.
incorrect by the 0.1AU parallel mean-free-path of newborn Simultaneously, the ISN flow direction in latitude βISM,∞
PUIsfoundbyGershmanetal.(2014)fromMESSENGERdata (λISM,∞) and the ISN temperature TISM,∞ (λISM,∞) are deter-
(Section4.2). mined in the Mo¨bius et al. (2012) analysis from the ISN flow
peak location in latitude and its angular width in latitude, σ ,
4.1.EvaluationofHeo FlowLongitude ψ
the observation of which is shown in Figure 3, again varying
fromIBEXMeasurements
with the ISN flow longitude (also see Mo¨bius et al. 2014). As
4.1.1.FunctionalDependenciesofHeoFlow indicated in the left panel of Figure 3, the observed angular
DirectiononPeakNeutralFluxes distribution of the incoming interstellar neutrals constitutes a
localMachconeofwidthv /V ,basedonthebulkflow
Thetrajectoriesofneutralinterstellarheliumatomsthrough Th ISM,1AU
V andthethermalvelocityv ofthedistributionat1AU.
the heliosphere depend on the four parameters describe the ISM,1AU Th
interstellar Heo gas at “infinity” beyond the influence of the Therefore,theobservedangularwidthfixestheratioTISM,∞1/2/
V , and the derived temperature depends on the derived
hmspeeeliaeosdsuprVehIdeSMrbe,y,∞It,BhaeEnXdg-aLtseomfl,tpohewersaeltoupnraegriaTtmuISdeMet,,e∞rλs.IaSIrnMe,t∞ihne,teplradatreiatpumedneedteeβrnIStsMpwa,i∞cthe, IthSIuSNMs,bo1unAlUkλIflSoMw,∞s.pBeeedcaVuIsSeMt,∞he,tshtaetiosntilcyavlaurniacbelretaiinntVieISsMo,f1AthUe,aonbd-
servedpeaklocationandwidthoftheISNflowareonly2%and
coupleduncertainties.AsdiscussedbyMo¨biusetal.(2012),Lee
1%,respectively,thecombineduncertaintiesofthetworelations
et al. (2012) and Mo¨bius et al. (2014), since IBEX is nearly a
are mainly determined by the uncertainties in relation (1) and
Sun-pointedspinner,thedeterminationoftheISNsinflowvector
potentialsystematiceffects.Wewilladdresssucheffectsbelow,
istiedtotheIBEX-Loobservationofthefluxmaximumthatis
inparticular,ontheresultingtemperature.
inadirectionthatisperpendiculartotheEarth–Sunline.Then,
AsshowninFigure4,thelatitudeψ oftheflowat1AUin
as shown in Figure 2, the observed flux maximum is directly Peak
related to the bulk flow vector at 1AU, in the plane of the thesolarrestframeisconnectedtotheISNflowlatitudeβISM,∞
atinfinitythroughthetrueanomalyofthebulkflowas
ISNbulkflowtrajectory.HereVISM,∞ andλISM,∞ areuniquely
tcahonendntohebcestreeerdlvaebtdiyontvhebeleohtcwyitpeyeernabtnohdleicttrhtureeajflaenocowtomryvaeleycqtθou∞rat(aitothnein(aEfinnqgiulteaybt)ieootnwf(et1eh)ne) tan(ψPeak)= ta|ns(iβn(ISθM∞,∞)|) = |sin(λISMta,∞n(β+IS1M8,0∞◦)−λPeak)|
(2)
trajectory, the observer location when obtaining the peak λ ,
andinflowdirectionλI(cid:2)SM,∞,so(cid:3)thatθ∞=λIS(cid:4)M(cid:5),∞+180◦−λPPis laantdionthsucsoangsatiitnutteoathdeegIeSnNeraincflyoiwntlhoengdiettuedrme iλnIaSMtio,∞n.oTfhVeIsSeM,r∞e-,
GM −1 1/2 βISM,∞,andTISM,∞ asafunctionofλISM,∞.However,relation
VISM,∞ = S −1 . (1) (2)alsoprovidesatooltobreakthedegeneracyusingIBEXlat-
rE cos(θ∞) itudedistributionscombinedfordifferentpositionsofIBEXas
Without further information, this constitutes a functional theobservervariesinlongitudeλ .Then,λ replacesλ
Obs Obs Peak
relationship of VISM,∞ (λISM,∞). The location of the flow inEquation(2)asavariableandtheresultingrelationψ(λISM,∞)
6
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
Figure3.CompositeofangulardistributionmeasurementasalocalMachCone,whosewidth(shownontheleft)σψ(cid:7) =vth(1AU)/VISM(1AU)isdefinedbytheratioof
thelocalthermalandbulkvelocitiesoftheflow.Theangularwidthinlatitudeσ(cid:7) isobtainedfromthecutthroughthemaximumoftheneutralheliumratedistribution
ψ
ineclipticlongitude(alongthewhitearc)showninthecolorcodedmap(centerpanel).Thelatitudinalratedistributionofthatcut(rightpanel)isshownalongwitha
Gaussianfitandstatisticaluncertainties.
Figure4.Schematicviewoftheinterstellargasflowinitstrajectoryplane(blue),whichisinclinedtotheecliptic(yellow)atthelatitudeinflowangleβISM,∞.ψPeak
isthelatitudeoftheflowinthesolarrestframe,θ∞istheperihelionanglewithrespecttotheinterstellarflowatinfinity,andβISM,∞isthelatitudeoftheinterstellar
flowatinfinity.Thex,y,z(andprimed)coordinatesdefinethecartesiancoordinatesystems.
dependsdistinctlyonλISM,∞.ThecomparisonofISNflowdis- distributions for the Witte (2004) parameter set produced by
tributionsobservedoverentire2009–2010observationseasons Bzowskietal.(2012)fromthebest-fittingχ2analysis.However,
with the predicted relation (2) as a function of λISM,∞ and the difference in the distributions between these two sets is
respective χ2-minimization in Mo¨bius et al. (2012) re- totallydominatedbythelargerangularwidthsofthecountrate
sulted in the reported optimum inflow longitude λISM,∞ = sequencesmeasuredbyIBEX,whichleadstothelargerderived
79◦.0(+3◦.0,−3◦.5)withasimilarresultinBzowskietal.(2012). temperature,andnotbyunaccounted-fordatalosses.Asshown
Bzowskietal.(2012)usedaχ2 analysistofindanuncertainty byLeeetal.(2012),thewidthsofthedistributioninlatitudeand
range for the interstellar flow longitude of ∼±4◦ (blue con- longitudearetightlycoupledandrepresenttheISNtemperature.
tour in Figure 22 of Bzowski et al. 2012). These uncertainties To illustrate the basis for the inflow direction in the IBEX
were the basis for the bounding values of the resulting pa- analysis,Figure5showsthepeaklocationandwidthoftheISN
rameter range from the IBEX observations used in McComas flow distributions in latitude as simulated for the observation
etal.(2012). periods during the 2009 ISN season orbits. Shown are the
In contrast, Lallement & Bertaux (2014, in top box of results for two ISN flow parameters sets that lie on the center
Figure2)maintainthatthevariationoftheobservedfluxesasa line of the parameter tube (e.g., Figure 1 in McComas et al.
functionofobserverlongitudeλ ,withasubstantiallywider 2012) as defined in Equations (1) and (2), using the inflow
Obs
maximumthanobservedwithUlyssesGAS(Witte2004),may longitudeλISM,∞ ofWitte2004(red)andofBzowskietal.and
haveledtothedifferentinflowlongitude,whichmaybeaffected Mo¨biusetal.(2012,blue).Forcomparisonthewidthisshown
byinstrumentalinfluences.Tosupporttheirclaim,theycompare asobtainedwiththetemperaturefoundbyWitte(2004),which
the observed IBEX count rate distributions with simulated stands out as much narrower than that from IBEX. However,
7
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
of their hypothesis that the helium vector of Witte (2004)
is correct, LB14 took the orbit-by-orbit IBEX data, and the
simulations where Bzowski et al. (2012, Figure 18) compared
the IBEX data with the Ulysses flow parameters. Those rates
anddatasimulationswerereplottedfortwocases(Figure2of
LB14).OnelineontheplotshowedtheresultsoftheBzowski
et al. (2012) analysis, while the other line plotted an adjusted
count-rate based on the LB14 hypothesized 7 ms dead-time
correction for the IBEX-Lo instrument. They argued that the
Witte (2004) flow direction agreed with the IBEX count rates
for their hypothesized dead-time correction.14 This argument
to reject the IBEX analysis because of a hypothesized 7 ms
dead-timecorrectionisnotvalid.
TheeffectthatinfluencedtheIBEXdataflowisaknownfi-
nitedataprocessingtime,notasensordead-time,andthetypes
ofeventsmostlyresponsibleforthiseffectwereunanticipated
electrons,whichdonotaltertheshapeoftheobservedinterstel-
lar flow distribution significantly. Also, these are excluded by
anewinstrumentmodethatstartedmid-2012.Thisoperational
modification has eliminated the count rate reduction effect al-
mostentirely.Comparisonbetweentheoldandnewoperational
modeshasdemonstratedthatthefinitedataprocessingtimedoes
Figure 5. Peak location in latitude (top) and width (bottom) of ISN flow notsignificantlyaffectthecountratedistributionsusedtoreach
distributionsfortheIBEXorbitsduringthe2009ISNflowseason,simulatedfor theconclusionsreportedinMo¨biusetal.(2012),Bzowskietal.
theactualobservationperiodswithtwoISNflowparametersetsalongthecenter
(2012), and McComas et al. (2012, Appendix B). In particu-
oftheIBEXtubeinparameterspace.TheinflowlongitudeofWitte(2004)is
lar, the data processing time did not influence the deduced He
showninredandtheoneofBzowskietal.(2012)andMo¨biusetal.(2012)
in blue. The latitudinal width of the distribution, σ, is also shown in green temperature, which LB14 tried to reconcile by exploring this
for the temperature found by Witte (2004). These simulations are based on effect to an extent that was already excluded by Mo¨bius et al.
themeasuredpropertiesoftheIBEX-Loinstrument.Atemperatureof8180K (2012).TheinstrumentresponsefunctionofIBEX-Lodoesnot
wouldberequiredfortheWitte(2004)flowvectortomatchthelatitudinalwidth
contributeinanysignificantwaytotheuncertaintiesofthein-
measuredbyIBEX.
terstellarHeoflowdirectionderivedfromtheIBEXdata.Thisis
discussedinmoredetailinAppendixB.
thevisibledifferencesbetweenthetwoparametersetsalongthe In summary, the inflow longitude parameter hinges mostly
IBEXtubeare,first,asteepervariationofthepeaklocationasa on the evolution of the peak location in latitude as a function
functionoforbitforλISM,∞asfavoredbyBzowskietal.(2012) of observer longitude in the IBEX observations. While the
than for the value of Witte (2004) and, second, a somewhat temperature in Mo¨bius et al. (2012) is obtained by fitting
strongerincreaseinwidthtowardtheearlyorbits.Toarriveata the height and width of the count rate distributions, these
valueforλISM,∞,solelythefirsteffecthasbeenusedbyMo¨bius observables were ignored in LB14. The temperature derived
etal.(2012).InrelationtoλISM,∞ theχ2-minimizationinthe from the IBEX observations, for the case where one assumes
comparisonbetweenobservedandsimulatedflowdistributions the inflow velocity of the Ulysses results (Witte, 2004), is
usedbyBzowskietal.(2012,seeFigure22)issensitivetoboth 30% higher than the temperature reported from Ulysses GAS
effectsasillustratedinFigure5. observationsofHeo.Systematicinstrumentaleffectslargerthan
This leaves only the argument of LB14 that the width 2%intheIBEXobservations,suchasanerroneousinstrument
of the angular distribution, which now solely influences the response function, can be ruled out as the reason for the
derivedtemperature,couldbebroughtintoagreementwiththe difference between the Ulysses and IBEX Heo flow directions
Ulysses value by considering the influence of a strong and (alsoseeAppendixB).
unaccounted-forinstrumentaleffectintheIBEXdatareduction.
AsderivedfromextensivetestswiththeactualIBEXdatasystem 4.1.2.IBEXResultsfor2009–2010SeasonsofData
hardware simulator and Monte Carlo simulations of realistic
data throughput of a wide range of particle rates, the actual IBEX-Lo directly samples the trajectories of interstellar
effectofdatasuppressionduetolimitedtransmissionbandwidth He atoms at 1AU after they have survived photoionization,
on the width of the observed distributions is <2% and thus is electron-impactionization,andcharge-exchangewiththeSolar
negligible(seeAppendixB,includingFigure6). wind to reach the 1AU orbit of IBEX (McComas et al. 2009;
LB14(p.1)makethepointthatthedegeneracybetweenthe Fuselier et al. 2009a, 2009b; Mo¨bius et al. 2009, 2014, 2012;
longitudeandvelocityoftheHeodistributionfunctionallowsthe Bzowskietal.2012).IBEX-Locollectsinterstellarheliumatoms
longitudeofthepeakcountratesofIBEXdatatoalsocorrespond overseveralorbitsduringlatewinterandearlyspringwherethe
totheexpectationsfortheUlyssesflowvector,providingthere atomsarenearperihelionoftheirtrajectories(Leeetal.2012).
is an additional suppression of count rates (detector “dead- Intheearliestorbits,IBEX-Losamplesthe“warmbreeze”that
time”) that was neglected in the IBEX data flow.13 In support consists of secondary Heo atoms that are deflected beyond the
heliopause(Kubiaketal.2014).Afterthatthedirectinterstellar
Heowinddominatesthecountrate.
13 Morespecifically,theystatethat“Wehaveexaminedtheconsequencesof
dead-timecountingeffects,andconcludethattheirinclusionatarealisticlevel
issufficienttoreconcilethe[IBEX2009–2010]datawiththeold[Witte 14 LB14(p.5)state:“...itisprematuretousethenewlongitudeandvelocity
Ulysses]parameters,callingforfurtherinvestigations.”(LB14,p.1) [ofIBEX]beforethequestionofthedead-timecorrectionisfullysettled.”
8
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
The interstellar helium wind direction was obtained from IBEX data analysis of the 2009–2010 observing seasons. An
these measurements using two independent analysis methods. explanation of the incorrectness of this speculation is given in
Mo¨biusetal.(2012)usedananalyticalapproachtofittheevolu- AppendixB,wherethedetailsoftheIBEXinstrumentalfunction
tionoftheheliumcountratesacrosstheorbitswhereheliumis arediscussed.
observed,wheretheouterboundaryofthecomputationswasat In addition, the ISN flow direction is driven by the varia-
infinity.Bzowskietal.(2012)comparedsimulationsofthehe- tionofthepeakinlatitudewiththeeclipticlongitudeofIBEX,
liumdistributionobtainedfromtheWarsawTestParticlemodel which is not affected by data flow restrictions. IBEX is a Sun-
withtheIBEXdata,andperformedχ2−fittingbetweenthemod- pointed spinner that measures vertical swaths of the sky with
els and data toobtain the besthelium wind parameters, where each spin (McComas et al. 2009). The Heo particle trajecto-
theouterboundaryofthecomputationswasat150AU,thedis- riesaremeasuredoverbothlongitudeandlatitude,andthatare
tance at which the interaction is already underway. McComas insensitivetothespeculationsofLB14regardingtheinstrumen-
etal.(2012)resolvedthesmalldiscrepancycausedbythediffer- talresponseparameter(Section4.1).Thebest-fittingISNflow
entouterboundariesusedinthetwocomputationsandshowed direction found from IBEX observations (Section 4.1.2) is not
that these independent analyses of the IBEX-Lo data yield dominatedbythelongitudinalheightandwidthoftheobserved
similarresults. fluxdistribution.
Independentanalysisofthe2009and2010seasonsofIBEX LB14havealsoclaimedthatthesecondindependentanalysis
data yielded the same flow direction within a large acceptable of IBEX data by Bzowski et al. (2012) omits “dead-time”
range. For the IBEX flow direction during the 2009 observing corrections.ThisargumentisinvalidbecausetheBzowskietal.
season,Bzowskietal.(2012)foundaHeflowvectorofλISM,∞= results are based on data that used an instrument response
79◦.2, βISM,∞ = −5◦.06, VISM,∞ = 22.831 kms−1, and TISM,∞ functionthathadbeentestedandvalidated.
= 6094 K. For the 2010 season, the Heo flow vector was
λISM,∞=79◦.2,βISM,∞=−5◦.12,VISM,∞=22.710kms−1,and 4.2.STEREOMeasurementsofPickupIons
TISM,∞ = 6254 K. These solutions also provided acceptable Measurements of He, Ne and O PUIs by the STEREO
χ2 values for mutual fits of the four interdependent parame- PLASTIC instrument between 2007 and 2011 are reported by
ters (Figure 22 in Bzowski et al. 2012). The acceptable range Drewsetal.(2012,hereafterD12).NeutralinterstellarHe,Ne
forthelongitudeis75◦.2–83◦.6,correspondingtotemperaturesof and O atoms penetrate to the inner heliosphere where they
4400–8200K,witheachflowlongitudecoupledtoatightlycon- are ionized by photoionization, electron-impact ionization, or
strainedrangeoftheotherparametersbecausetheuncertainties charge-exchange with the solar wind (Bzowski et al. 2013).15
arecorrelated.Incontrast,whentheUlyssesheliumflowlongi- The charged particles that form inside of 1AU couple to the
tude(Witte2004)wasusedasaconstrainttotheIBEXlongitude solar wind magnetic field and are convected radially outward
data, the best reduced χ2 fit was an order-of-magnitude larger withthesolarwindwheretheycanbeobservedat1AU.Helium
(Figure17inBzowskietal.2012).Independently,andwiththe andNesurvivethelargerionizationlossesassociatedwithclose
analytic method described by Lee et al. (2012), Mo¨bius et al. passage to the Sun and become gravitationally focused in the
(2012) determined a flow vector for the combined 2009–2010 downwind direction. Oxygen is mainly ionized in the upwind
seasons of λISM,∞ = 79.0+−33..05 deg, βISM,∞ = −4.9±0.2 deg, regions. Helium, Ne and O PUIs form a region of enhanced
VISM,∞ = 23.5−+32..00 kms−1, and TISM,∞ = 6200+−21020000 K. The PUI density with a crescent-like distribution in the upwind
best-fitting direction of the interstellar helium wind using hemisphere (D12). Drews et al. determined the interstellar
two different analysis techniques are in good agreement with wind longitude independently from the He and Ne focusing
eachother. cones, and the upwind He, Ne and O crescents, and obtained
Thelargerangeofpossiblelongitudesthatisobtainedwhen fiveindependentmeasurementsoftheeclipticlongitudeofthe
the correlated uncertainties of the four-dimensional parameter interstellar wind direction (Table 2) that were used in Paper I
areincludedcanbeconstrainedfurtherifindependentinforma- to evaluate the statistical likelihood that the interstellar wind
tionontheLICtemperatureisavailable.InPaperIweusedthe directionhadchangedoverthepastfortyyears.
gastemperaturethatisobtainedfromastronomicalobservations LB14 conclude that the STEREO PUI data are only valid
of the spectra of refractory interstellar elements toward Sirius as upper limits on the interstellar wind longitude, and that the
(seeAppendixA). oxygen data should be ignored entirely. In more detail they
One outcome of the Bzowski et al. (2014) reanalysis of the argue that: (1) PUI transport inside of 1AU causes deviations
Ulyssesdatasetisthatthe“suspicious”coincidenceremarked of the location of the focusing cone and crescent of up to 5◦
upon by LB14, that the latitude of the Heo flow did not vary so that values for the flow directions obtained from STEREO
between the IBEX and Ulysses Heo measurements, may not PUIcanonlybetakenasupperlimits.(2)Therequirementofa
be so settled after all with the new reduction and modeling of smoothlyvaryingionizationrateusedintheSTEREOanalysisis
the Ulysses data. Bzowski et al. (2014) found that the best-fit notsatisfied.TheyclaimthatD12onlyconsiderphotoionization
Ulyssesupwinddirectionoftheflowlatitudeisonedegreenorth andnotconsideredionizationbyelectronimpacts.(3)Theinter-
of the center of the IBEX range, at βISM,∞ = 6◦ ±1◦.0 versus comparison of the data between orbit groupings are invalid
βISM,∞=4◦.98±1◦.21forIBEX(McComasetal.2012). because the data are not statistically independent. (4) The
statistical simulations of the direction uncertainties of the
4.1.3.IBEXInstrumentalResponseFunction
15 PhotoionizationmodelsoftheLICpredictneutralfractionsatthe
In order to support their perspective that the interstellar
heliosphereforhydrogen,helium,neon,andoxygenof∼0.78,∼0.61,∼0.20,
wind direction through the heliosphere is invariant with time, and∼0.80respectively(Slavin&Frisch2008).Upto60%,30%,and4%of
LB14 speculated that the instrumental response function used theseneutralinterstellarHe,NeandOatoms,respectively,survivetothe
to analyze the IBEX-Lo data is incorrect, and argued that Earth’sorbitbeforebeingionizedbyamixofelectron-impactionization,
charge-exchangewiththesolarwind,andphotoionization(Bzowskietal.
the true “dead-time” correction is 7 ms, which is larger than
2013).PUIdatasampleparticlescreatedbetweentheSunandtheobservation
the value of the instrument response function underlying the point.
9
TheAstrophysicalJournal,801:61(15pp),2015March1 Frischetal.
STEREOanalysisareincorrect.(5)Oxygenshouldnotbeused cant.HoweverthelargeruncertaintiesontheHeoflowlongitude
to infer the interstellar flow since there is also a population of obtainedfromtheMESSENGERmeasurements,77◦.0±6◦.0,do
secondaryoxygenatoms. notallowpossibletransporteffectstoberuledoutconclusively.
The IBEX 2009 in situ measurements were collected over a
4.2.1.TransportofPUIsintheExpandingSolarWind similartimeperiodandagreewiththePUIfocusingconelon-
gitude(Mo¨biusetal.2012;Bzowskietal.2012).Comparisons
PUIsarecreatedintheturbulentenvironmentinsideof1AU.
between the ACE PUI results (Gloeckler et al. 2004) and the
If the mean-free-path for transport of the newly formed PUIs
is large, λ(cid:8) ∼ 1AU, PUI ions are transported parallel to the Ulysses in situ Heo results (Witte 2004), collected at similar
timesbutdifferentradialdistances,identifysimilarheliumflow
magneticfieldlinessothattheirspatialdistributionisnolonger
directions.
a valid marker of the ISN wind direction (Chalov 2014).16
Although the possible transport of PUIs inside of 1AU
HoweverifthePUIsareformedinanon-turbulentenvironment,
e.g., λ(cid:8) ∼ 0.1AU, the parallel transport is negligible. A new remains an important question because of large uncertainties
MESSENGERstudyhasshownthatλ(cid:8) ∼0.1AUintheregion ontheheliumfocusingconelongitudeintheSTEREOdata,and
because ofthelargeelectronimpactionizationratethatvaries
0.1–0.3AU from the Sun (Gershman et al. 2014). These new
withtime,availabledatasuggestthatitisnotsignificant.
results invalidate the main argument of LB14 that the PUI
data should not be used for determining the direction of the
interstellar neutral wind. However, we reach this conclusion
4.2.2.TemporalVariationsinPUICountRates
with the understanding that new modeling of the ionization of
ISNsclosetheSunisprovidingnewinsightsintothestabilityof The geometric properties of the gravitational focusing cone
the focusing cone and crescent locations that may affect these andupwindcrescentvarywithboththesolarmagneticactivity
conclusions(J.Sokol2014,privatecommunication). cycle and the solar rotation. The PUIs observed at 1AU are
LB14basedtheirconclusionsregardingthePUIsonastudy seeded by the ionization of neutrals inside of 1AU (Bzowski
ofChalov&Fahr(2006),whofoundthattransporteffectscould etal.2013),andsubsequentoutwardconvection aspartofthe
generate a relative azimuthal displacement of PUIs to positive solar wind with a distinct energy distribution (Vasyliunas &
longitudesofuptofivedegrees,butthatthetransporteffectwas Siscoe1976).
insignificantfortherangeofw =v /v >1.4intheSWICS IntheapproachusedbyD12,bothslowlyandrapidlyvarying
ion sw
datafromACEbecauseofinstrumentalproperties.Howeverthe influences on the observed cone and crescent structure are
STEREO data used by D12 is for PUIs with w > 1.5 so the included in the analysis, but due to the different origins of
Chalov&Fahr(2006)modelingdoesnotapply. long-term and short-term modulations they are included in
In addition to the new MESSENGER results, three other distinctly different ways. Slowly changing modulations of the
setsofmeasurementsoftheISNflowlongitudeobtainedfrom PUIcountratesincludepossiblevariationsoftheionizationrate
focusingconemeasurementsafter2007agreewitheachother. inducedbysolarUVradiationorbyslowdecayofinstrumental
ThelongitudefortheHeo ISNflowfoundfromSTEREO data detectionefficiency.Eitherwouldresultinasystematicshiftof
collected 2007–2011 is 77◦.4 ± 1◦.9 (Drews et al. 2012). It the focusing cone and crescent location in ecliptic longitude.
is consistent with the ACE SWICS and MESSENGER FIPS Theseshiftscangoineitherdirection,andtheydependpurely
data collected over the first two years of this interval. Also, on the nature of the modulation. Temporal variations in the
Gershmanetal.(2013,G13)analyzedACESWICS1AUdata interstellarwinddirectionwereassumedtobenegligible.
from 2007–2009 and found a longitude 77◦.0 ± 1◦.5 that was Drews et al. (2012) estimated the influences of the slowly
consistent with the ISN flow direction, 76◦.0±6◦.0, collected varying count rate modulation of interstellar PUIs on the re-
by MESSENGER at 0.3AU during three passes through the sults and explicitly included those variations in the analysis.
focusing cone during 2007–2009. Gershman et al. (2013) ComparisonsbetweenSWICSPUIdataandthesolarUVemis-
concluded that the similar longitudes found from the 1AU sion over a ten-year period of time showed that the influence
SWICSdataandthe0.3AUFIPSdataindicatedthatazimuthal of slowly varying ionization rates induced by solar UV radia-
transporteffectshaveaminimaleffectontheHeoflowlongitude, tion is negligible. Influences of a slowly decreasing detection
andthatelectronimpactionizationissignificantinsideof1AU. efficiency induced by aging of the PLASTIC’s microchannel
However, as noted above, transport effects appear to plates were considered explicitly in the STEREO analysis by
have a minimal effect on the direction of the He focusing estimating the slowly varying detection efficiency decay with
cone. STEREO (Drews et al. 2012), and MESSENGER and time. Other slowly varying modulations of the observed PUI
ACE (Gershman et al. 2013) measurements of the helium fo- count rates, e.g., the effect of the eccentricity of the orbit
cusingconewerecollected conducted duringsimilartemporal of STEREO around the Sun, have been thoroughly discussed
intervalsof2007–2009,duringsolarminimum,andfoundfocus- inD12.
ingconedirectionsinagreement.TheMESSENGERdatawere PUIcountratemodulationovershorttimescales,ontheother
collectedat0.3–0.7AUwhileACEandSTEREOdatawereob- hand, are included implicitly through the statistical approach
tainednear1AU,suggestingthattransporteffectsareinsignifi- used by D12. The short-term modulations mainly stem from
solarwinddensitycompressions,changesofthesolarmagnetic
16 Chalov(2014)hasinvestigatedtheinfluenceoftransporteffectsonthe field, or solar wind variabilities such as the electron impact
heliumfocusingconepositionfortheunusualsolarminimumconditions ionizationrates(thatLB14claimareignored).Asdemonstrated
between2007and2009,anintervalthatisincludedintheSTEREO
inD12,thestatisticalapproachisespeciallywell-suitedtodeal
measurements.BynumericallysolvingthetransportequationofHe+PUIs,the
possibledisplacementoftheheliumfocusingconewasevaluatedfordifferent withthesekindsoffluctuations.
valuesofthemeanfreepathoftheatomparalleltothemagneticfield,λ(cid:8).For Theinfluenceofelectronimpactionizationat1AUisstrongly
themostextremeconditions,λ(cid:8)=1AUgaveanoffsetanglefortheobserved coupledtotheoccurrenceofdensesolarwindstreams(Bzowski
focusingconecenterof3◦,whichwouldthenneedtobesubtractedfromthe
et al. 2013) and therefore fluctuates on short time scales.
measuredvaluetogetthetrueinflowingwinddirection.Howeverfor
λ(cid:8)=0.1AUoftheMESSENGERdata,notransportisfound. LB14 argue that the statistical analysis does not include the
10