Mon.Not.R.Astron.Soc.000,1–??(2014) Printed20January2014 (MNLaTEXstylefilev2.2) Strong RR Lyrae excess in the Hercules-Aquila Cloud Iulia T. Simion1(cid:63), Vasily Belokurov1, Mike Irwin1, Sergey E. Koposov1,2 1Institute of Astronomy, Madingley Rd, Cambridge, CB3 0HA, 2Moscow MV Lomonosov State University, Sternberg Astronomical Institute, Moscow 119992, Russia Accepted2014January16.Received2014January15;inoriginalform2013December16 4 1 0 2 ABSTRACT We map the large-scale sub-structure in the Galactic stellar halo using accurate 3D n positionsof∼14,000RRLyraereportedbytheCatalinaSkySurvey.Intheheliocentric a distancerangeof10-25kpc,intheregionoftheskyapproximatelyboundedby30◦ < J l<55◦ and−45◦ <b<−25◦,thereappearstobeastrongexcessofRRabstars.This 7 overdensity, peaking at 18 kpc, is most likely associated with the so-called Hercules- 1 AquilaCloud,previouslydetectedusingMainSequencetracersatsimilardistancesin the Sloan Digital Sky Survey data. Our analysis of the period-amplitude distribution ] A of RR Lyrae in this region indicates that the HAC is dominated by the Oosterhoff I G type population. By comparing the measured RR Lyrae number density to models of a smooth stellar halo, we estimate the significance of the observed excess and provide . h an updated estimate of the total luminosity of the Cloud’s progenitor. p - Key words: Galaxy: structure – Galaxy: stellar content – Galaxy: halo – stars: o variables: RR Lyrae – galaxies: individual: Milky Way – galaxies: photometry. r t s a [ 1 INTRODUCTION and the bulge, on the other side of the Galaxy (Belokurov 1 et al. 2007). Given their obvious proximity to the plane, v With the recent discoveries of both ongoing disruption claimshavebeenmadeastothegenesisofthesestructures, 4 events (such as the Sagittarius stream, see e.g. Majewski 5 linkingthemtoperturbationsinthestellardiskdensity.For et al. 2003) and of the remnants of past mergers (e.g. the 3 example,Larsenetal.(2011)arguethatwhiletheexcessof VirgoCloud,seeDuffauetal.2006;Juri´cetal.2008),there 4 faint Main Sequence (MS) stars around the HAC position isnowlittledoubtthatasubstantialportionoftheGalactic . isreal,thedistancetothesub-structurewasover-estimated 1 stellar halo has been accreted. Today, the focus has shifted andthetruelocationoftheCloudismuchcloser,only1to 0 to finding out what exactly fell onto the Milky Way and 4 6 kpc from the Sun rather than 10 to 20 kpc. when. The reconstruction of the stellar halo accretion his- 1 TheHACisachallengingstructuretomap.Itspresence tory necessarily involves brushing away the Galactic in-situ : is easily spotted due to the obvious asymmetry in the faint v components so that the faint fragments of broken satellites MS density in the directions towards the Galactic center. i canbepiecedtogether.Therefore,themajorityofthedetec- X However, characterizing its full extent is far from straight- tionsofstellarhalosub-structurehavebeenlimitedtohigh r (|b|>30◦) Galactic latitudes, where disk/bulge contamina- forward. The Sloan Digital Sky Survey (SDSS) data, which a provides the deepest wide area view of the sky, is not sym- tion is minimal. metric between positive and negative Galactic longitudes The current tally (see e.g. Belokurov 2013) indicates and latitudes. Moreover, the contiguous coverage stops at that as much as 50%−70% of the stellar halo is in well- latitudes|b|≈30◦,butthereexistsomelimitedSDSSimag- mixed, seemingly smooth components, with another 30%, ing data along several ∼ 2.5◦ wide stripes slicing through or perhaps even 50%, contributed by four structures: the the Galaxy at constant l at |b|<30◦. The analysis of these Sagittarius stream, the Galactic Anti-centre Stellar Struc- slices(Belokurovetal.2007)clearlyindicatesthattheHAC ture (GASS), the Virgo Cloud (containing the Virgo Over- signalcontinuestolowlatitudes,atleastaslowas|b|∼15◦. density and the Virgo Stellar Stream), and the Hercules- Inaddition,theCloudisnoticeablyasymmetricwithrespect Aquila Cloud (HAC). However, curiously two of the four to the disk plane, i.e. at l ∼ 50◦, there exists a prominent most massive structures in the stellar halo seem to reside over-abundanceofMSstarsintheGalacticSouthernhemi- very close to the disk plane. The GASS covers a large re- sphere.TheMSstarscontributingtotheover-densityarein gion of the low-latitude sky approximately in the direction a wide magnitude range 18 < i < 22, thus giving only an of the anti-centre and the HAC is hidden behind the disk approximate indication of the distance to the Cloud. RR Lyrae are an ideal tracer population to pin down (cid:63) email:[email protected] the distance and the luminosity of the HAC. To this end, (cid:13)c 2014RAS 2 I.T. Simion et al. 50 control field NH2 NH ) (◦ 0 b CSS completeness SH2 SH 0.8 0.7 0.6 0.5 F0.4 -50 0.3 0.2 0.1 0 5 10 15 20 25 30 DRR (kpc) 150 100 50 0 -50 -100 -150 l (◦) Figure 1. Spatial distribution of Catalina Schmidt Survey RRab Lyrae (DR13a) in Galactic coordinates. We highlight with blue our selectionofthecontrolfield(seeSection3),withredourselectionsofthe’on-cloud’fields,intheNorthernHemisphere(NH)andinthe SouthernHemisphere(SH)(seeSection4).Ingreenweshowanotherfieldselection(NH2andSH2)thatwewilluseinSection4.The insertshowsthecompletenessofthesurveyasafunctionofdistance,assumingAV =0. Watkins et al. (2009) and Sesar et al. (2010) exploit the RRLyraearealsorelativelyabundant,withtheirnum- multi-epochdatainthenarrowequatorialSDSSStripe82to ber density in the solar neighborhood between 4 and 6 detectRRLyraestarswithhighefficiency.Stripe82briefly kpc−3. These tracers allow for studies of the halo substruc- crossestheHACregionintheSouthandbothteamsreport ture with good spatial resolution (e.g. Preston et al. 1991; a significant overdensity of RR Lyrae at 45◦ <l<60◦ and Vivas & Zinn 2006). Several large-scale RR Lyrae surveys −45◦ <b<−25◦.Watkinsetal.(2009)givethedistanceto existalreadyandausefulcomparisonbetweentheminterms theHACof22kpc,thisistheaverageoverthecross-section of their sky coverage and completeness is presented in Ta- of the Cloud and the Stripe. ble1ofMateuetal.(2012).However,2013sawtwonewRR In this work, we take advantage of the large sample Lyrae catalogues published: LINEAR II (Sesar et al. 2013) (inexcessof10,000)ofintermediatedistance(5-30kpc)RR which probes distances between 5 kpc and 30 kpc over ∼ LyraestarsoftheabtypedetectedbytheCatalinaReal-time 8,000 deg2; and the Catalina Schmidt Survey (Drake et al. Transient Survey (CRTS). The size of the sample and the 2013a,b, hereafter DR13a,b ) reaching distances between 2 distance range probed by these data is ample to search for and 60 kpc over ∼ 20,000 deg2 of the sky. the signatures of large-scale halo sub-structure, and in par- TheCatalinaSkySurveybeganin2004andusedthreedif- ticulartheHAC.Thepaperisstructuredasfollows.Section ferenttelescopestodiscovernear-Earthobjects(NEOs)and 2 briefly describes the properties of the RR Lyrae dataset potentiallyhazardousasteroids(PHAs).Eachofthesurvey analysed while Section 3 describes the smooth stellar halo telescopes is run as a separate sub-survey. In this work, we models used to reveal the Cloud and estimate its signifi- areusingthedatafromtheCatalinaSchmidtSurvey(CSS), cance.Finally,Section4focusesonthepropertiesoftheRR oneofthethreesub-surveysoftheCatalinaSkySurvey.The Lyrae stars in the Cloud itself. CSScontainsanimpressivesampleof∼14,500typeabRR Lyrae with magnitudes between 11.5 and 20 mag, spread overapproximatelyhalfofthesky(0◦ <RA<360◦,22◦ < DEC<65◦),thelargestGalacticvolumeeversurveyedwith 2 DATA RR Lyrae. The catalogue comes with reliable distance esti- RR Lyrae are radially pulsating horizontal branch stars, matesfromawell-studiedabsolutemagnitudevs.metallicity with pulsation periods between 5 and 15 hours and ampli- distribution (Catelan & Cort´es 2008). tudes in the range 0.2 -1.6 mag in the V-band. These stars The parameters of the CSS RRab stars are listed in have a well defined Period-Luminosity-Metallicity relation Table 1 of DR13a and Table 2 of DR13b. In what follows, which allows for accurate distance determination (normally we are going to use their published equatorial coordinates with less than 10% uncertainty). RR Lyrae are low-mass, RA,DEC,(seeFigure1),aswellasthemagnitudesV and 0 long-lived stars (typically more than 10 Gyrs old) and are the heliocentric distances D . The V magnitudes are al- RR 0 foundinlargequantitiesinglobularclustersandinthestel- ready extinction corrected using the Schlegel et al. (1998) lar halo of our Galaxy making them ideal tracers of the in- reddeningmaps.Theuncertaintiesonalloftheabovequan- ternal substructure of the Milky Way. They are sufficiently titiesarenotlistedbutestimatesareprovided.Theaverage bright,withanalmostinvariableabsolutemagnitude(M ≈ metallicity of the sub-sample with available SDSS spectra V 0.59) meaning that they can be detected out to large dis- is approximately −1.55 (see Figure 20, DR13a) therefore, tances in relatively shallow surveys, i.e. ∼ 100 kpc for a given that the absolute magnitude of the RRL has a lin- survey with a modest limiting magnitude between 20 and eardependenceonmetallicity(e.g.Chaboyer1999),wecan 21 mag. assume M ≈ 0.6 for the CSS data. The dispersion in the V (cid:13)c 2014RAS,MNRAS000,1–?? RR Lyrae in the Hercules-Aquila Cloud 3 Table 1.Singlepower-lawandbrokenpower-lawsmodelswiththeirparameters.Thenormalisationforthe Watkinsetal.(2009)modelisdividedby11,inagreementwithSesaretal.(2010)andDR13b. Model αin αout rb qH ρR(cid:12)R Nooftracers Area Survey Reference kpc−3 (approx) deg2 BPL 2.40 4.50 23 1.00 3.0 316RRab 290 Stripe82 Watkinsetal.(2009) SPL 2.77 0.64 5.6 366RRab 290 Stripe82 Sesaretal.(2010) BPL 2.30 4.60 27 0.61 7.3 6,800BHBs 14,000 SDSSDR8 Deasonetal.(2011) BPL 2.62 3.80 28 0.71 5.9 34,000MS 170 CFHTLS Sesaretal.(2011) SPL 2.42 0.63 5.6 4,000RRab 8,000 LINEARII Sesaretal.(2013) 0.0◦< l < 80.0◦, 45.0◦< b < 70.0◦ control field 0.15 693 - CSS data 0.10 755 - BPL Watkins+ 2009 el 475163 -- SBPPLL SDeesaasro+n +20 210011 mod 0.10 2eg−00..0068 656851 -- BSPPLL SSeessaarr++ 22001131 model)/ 0.05 dR a− NR0.04 dat 0.00 x ( 0.02 2 −−0.05 0 1 0.00 −0.10 0 5 10 15 20 25 30 0 5 10 15 20 25 30 D (kpc) D (kpc) RR RR Figure2.Leftpanel:ComparisonbetweentheheliocentricdistancenumberdensitydistributionofCSSRRabLyraeinthecontrolfield (selectedinblueinFigure1)andfivesimplepowerlawmodelpredictions(seelabelsandTable1).Thenumbersinthelegendarethe integrated number of RRab in the 5-28 kpc distance range. As our sample of RRab is incomplete, we corrected the model predictions for the effects of completeness, using equation 5. Right panel: Relative difference between the data and the model. The most suitable models for predicting the data in the control field are the Deason et al. (2011) and Sesar et al. (2011) broken power-law models (the Deasonetal.(2011)modelwasscaledsuchthatitreproducesthenumberofRRLobservedinthisfield). metallicityofthesub-sample,∼0.3dex,introducesanerror numbercountsofoldmetal-poorstarsintheGalactichalo. on the absolute magnitude, while the apparent magnitudes According to these authors, within 20 kpc of the Galactic are affected by photometric error. These two factors trans- centrestellarhalosubstructure,astracedbyBHBs,reaches lateintoatotaluncertaintyonthedistancesoftheorderof thelevelsof10%−20%whentheknownlargeoverdensities 7% (e.g. at D=15 kpc the error will be δD≈± 1 kpc). such as the Sagittarius stream are removed. Since, apart from the HAC, there are no other halo structures known in theregionwearefocusingon,itissafetopositthatsimilar levels of smoothness should be seen in the RRab tracers, 3 SIMPLE SMOOTH HALO MODELS at least in the comparison field. This is not an unreason- able assumption to make given the earlier successful RR To probe the clumpiness of the Galactic RR Lyrae distri- Lyrae modelling attempts, though with admittedly smaller bution on the scales of several kpc, a reasonable first step datasets (e.g. Watkins et al. 2009; Sesar et al. 2010, 2011). is to compare the observed tracer volume density to the Simple stellar number count models of the halo are predictions of a simple smooth halo model. To guide this based on the following formalism. The heliocentric distance comparison we identify several locations of interest in the D and galactic coordinates (l,b) of a star and its Galacto- centralGalaxy.Figure1showsthespatialdistributionofall centric Cartesian coordinates are related as ∼14,400RRabLyraeintheCSSsample.Theprincipal“on- cloud” fields in the South and North Galactic hemispheres X =Dcos(l)cos(b)−R (cid:12) areplottedinredandmarkedwithNHandSHrespectively, Y =Dsin(l)cos(b) the secondary fields are marked in green. Additionally, a comparison field (control field) is chosen at suitably high Z =Dsin(b) Galactic latitude in the region free of known sub-structure; r2 =X2+Y2+Z2q−2 H this appears in blue. These fields are selected with the fol- lowinglogicinmind.Inthecomparisonfield,asmoothden- where q describes the flattening of the halo. H sitymodelcanbefoundthatdescribesthedatawell.Inthe HereR istheSun’sdistancefromtheGalacticcenterand (cid:12) primary“on-cloud”fields,thestrengthoftheresidualswith r is the star’s Galactocentric ellipsoidal distance. The Sun respecttothemodelcanbeusedtoascertainthepresenceof isat(X,Y,Z)=(-8,0,0)kpc,theX-axispointstowardsthe the Cloud. The neighboring secondary fields, labelled NH2 Galactic center, Y component in the direction of Galactic and SH2, are analysed to constrain the extent of the HAC. rotation and the Z-axis points toward the north Galactic As illustrated in e.g. Deason et al. (2011), power law pole. volumedensitymodelsarefairlysuccessfulatpredictingthe Thestellardensitydistributionintheinnerhalo(within (cid:13)c 2014RAS,MNRAS000,1–?? 4 I.T. Simion et al. ∼20kpc)appearstobewelldescribedwithasimplepower- Lyrae tracers, for example those by Watkins et al. (2009) law form andSesaretal.(2010)detectedintheSDSSStripe82data (cid:18)R (cid:19)α and the LINEAR II catalogue (Sesar et al. 2013). However, ρRmRodel(D,l,b)=ρR(cid:12)R r(cid:12) . (1) in two out of five modeling efforts, stars other than RR Lyrae were employed. For the BPL model fit to MS stars Atlargerdistances,asteeperpower-lawindexisrequiredto from the Canada France Hawaii Telescope Legacy Survey, fitthedatawell(seee.g.Deasonetal.2011),thereforelead- Table1givestheconversionofthedensitynormalisationto ingtoanoverallbrokenpower-lawmodelwithcharacteristic RR Lyrae counts derived by Sesar et al. (2013). “break radius” rb FortheBPLmodelofDeasonetal.(2011)derivedusing BHBstars,theRRLyraedensitynormalisationisobtained  using the following procedure. The model of Deason et al. R  (rrb)αin if r(cid:54)rb, (2011)andtheCSSRRabdataarebothintegratedoverthe ρRmRodel(r)=ρR(cid:12)R( r(cid:12))αin × (2) area covered by the control field shown in blue in Figure 1. b  (rrb)αout if r>rb. The control field is a 1,100 deg2 region with 0◦ < l < 80◦ and 45◦ < b < 70◦. To our knowledge, there are no stellar For a given density model, we can estimate the number of streams or satellites reported in this area within distances RR Lyrae in the given solid angle within an increment of D < 30 kpc. Equating the incompleteness-corrected in- RR heliocentric distance ∆DRR around DRR tegral of the BPL model of Deason et al. (2011) over the ∆N (D ,l,b)=ρRRD2 ρRR (D ,l,b) control field area out to D=30 kpc to the total number of exp RR (cid:12) RR model RR (3) RRab in the CSS data we obtain ρRR = 7.3 kpc−3, shown cosb∆D ∆l∆b, (cid:12) RR in italics in Table 1. whereρRR isnumberdensityofRRabinthesolarneighbor- Figure 2 compares the CSS RRab data in the control (cid:12) hood. field and the predictions of the five simple power-law mod- As the CSS RRab catalogue is incomplete, with the els listed in Table 1. The data is shown in the blue his- completenessdroppingatfaintmagnitudesfromunder70% tograms using 2 kpc binning (errors on the distance are tounder40%,weneedtoscaledownthemodelpredictions ∼ 7%, i.e. less than the bin size). The uncertainty in the (andmoresoatlargedistances)totakethesurveydetection number of expected RR Lyrae is calculated as δNobs ≈ (cid:112) efficiencyintoaccount.ThefractionofRRLyraedetectedby (δNexp)2+(δF)2,whereδF istheuncertaintyinthecom- theCSSasafunctionofmagnitudeispresentedinFigure13 pleteness estimated in each distance bin and δNexp is the (cid:112) ofDR13a.TheinsertofFigure1showsthecompletenessasa Poissonuncertainty Nexp.AccordingtotheFigure,single functionofheliocentricdistance.Thiscurveisderivedfrom power-lawmodelscannotreproducethedataadequately.In Figure 13 of DR13a, using the well-known relation between ordernottoover-estimatethenumberoftracersatlargedis- apparent magnitude and distance modulus and assuming tances,SPLmodelsmustunder-predictthecountsatsmaller M =0.6: distances:notehowbothreddashedanddash-dottedcurves V fall low in the distance range 5<D<15 kpc. Of the three DRR =100.2(V−MV+5−AV) (4) broken power-law models, the ones by Deason et al. (2011) and Sesar et al. (2011) perform the best with the model of where V = V − A is the average magnitude from the 0 V Watkinsetal.(2009)consistentlyover-predictingRRLyrae Fourier fit to the RRab light curves listed in Table 1 of counts at D > 10 kpc, albeit at the level < 10% 2. In DR13aandTable2ofDR13b.Theefficiency-correctednum- whatfollows,wechoosetocomparethedatawiththemodel ber of stars predicted by the model is thus given by of Deason et al. (2011) simply because it gives marginally ∆N (D ,l,b)=F(D )∆N (D ,l,b) (5) higherRRabcountsatD>10kpc,andthereforecompared obs RR RR exp RR to the prediction of Sesar et al. (2011) the estimate of sub- where F is the completeness function shown in Figure 1. structure excess is slightly less optimistic. The efficiency curve needs to be shifted in distance as A V varies. Rather than fitting our own density model ρ to model the CSS RR Lyrae data we have re-used some of the halo 4 RR LYRAE IN THE HERCULES-AQUILA model parameters obtained in earlier studies.1 Table 1 lists CLOUD thebest-fitvaluesforthenumberdensitydistributionofdif- Figure 3 gives the all-sky density distribution of RR Lyrae ferent halo tracers: main-sequence stars (Sesar et al. 2010); with 10 < D < 25 kpc in the CSS and in the model RRab Lyrae between 5 and 23 kpc (Sesar et al. 2013); and of Deason et al. (2011) applying the re-normalisation de- blue horizontal branch stars (Deason et al. 2011). Also dis- duced in the previous section. The left (data) and the mid- played are the values obtained in two independent studies dle (model) columns of the Figure look very similar, and of the RR Lyrae density distribution in Stripe 82 (Watkins et al. 2009; Sesar et al. 2010). All normalisation values ρRR as illustrated in the right panel (residuals), the agreement (cid:12) is only broken in the few regions dominated by the known listedinTable1areestimatedfortheMilkyWayhaloRRab halo sub-structures. More precisely, at −100◦ < l < −50◦ stars. Some of these were obtained using the actual RR and b > 30◦, the Virgo Cloud, i.e. the nearer portion of 1 For a discussion of the difficulties of fitting a smooth density distributiontoanRRLyraedataset,seesection7ofSesaretal. 2 Note that the (Watkins et al. 2009) model has to be scaled (2013). downbyafactorof11inagreementwithDR13b. (cid:13)c 2014RAS,MNRAS000,1–?? RR Lyrae in the Hercules-Aquila Cloud 5 10 <DRR< 25 kpc DATA MODEL (DATA - MODEL) 50 50 50 ) ) ) (b◦ 0 (b◦ 0 (b◦ 0 VCilroguod OStrrpehaamn l e aaSrdgmirng HAC excess -50 -50 -50 log10(N) log10(N) Sgr N trailing 0.0 1.0 2.0 2.5 0.0 1.0 2.0 2.5 arm −20 −10 0 10 20 150 100 50 0 -50 -100 -150 150 100 50 0 -50 -100 -150 150 100 50 0 -50 -100 -150 l (◦) l (◦) l (◦) 10 <DRR< 15 kpc 50 50 50 ) ) ) (b◦ 0 (b◦ 0 (b◦ 0 log10(N) log10(N) N -50 0 1 2 -50 0 1 2 -50 −20 −10 0 10 20 50 0 -50 50 0 -50 50 0 -50 l (◦) l (◦) l (◦) 15 <DRR< 20 kpc 50 50 50 ) ) ) (b◦ 0 (b◦ 0 (b◦ 0 e xpceeasks log10(N) log10(N) N -50 0 1 2 -50 0 1 2 -50 −20 −10 0 10 20 50 0 -50 50 0 -50 50 0 -50 l (◦) l (◦) l (◦) 20 <DRR< 25 kpc 50 50 50 ) ()b◦ 0 ()b◦ 0 (b◦ 0 log10(N) log10(N) N -50 0 1 2 -50 0 1 2 -50 −20 −10 0 10 20 50 0 -50 50 0 -50 50 0 -50 l (◦) l (◦) l (◦) Figure 3. Data, model predictions and residuals (left, middle and right panels respectively) in Galactic coordinates, in 6◦ x 6◦ bins. Thenumberdensitymapwasnotsmoothedandittakesintoaccounttheefficiencycorrectionatvaryinglandb.Overlaidontheplots arecontoursofconstantreddeningE(B-V)(blackisE(B-V)=0.10andgrayisE(B-V)=0.04).Toppanels:all-skydensitydistribution ofRRLyraewith10<DRR<25kpcintheCSS(left)andthemodelDeasonetal.(2011)(middle).Theresiduals(right)highlightthe halo substructure, in particular the Virgo Cloud (possibly mostly the VSS), the Orphan Stream, and the Sagittarius stream. A strong excessofRRLyraeispresentintheSHregion.Lowerpanels:zoom-indensitymapsofthecentral180◦ oftheGalaxyinthreedistance bins:10<DRR <15kpc(2ndrow),15<DRR <20kpc(3rdrow)and20<DRR <25kpc(4throw).Thepeakoftheexcessinthe SH field is located in the 15<DRR <20 kpc distance bin, as also shown in Figure 4. Note that the left and middle panels show the logarithmofthestellarnumberdensityineach6x6deg2 bin. it, the VSS (see e.g. Zinn et al. 2014) can be clearly seen kpc. Judging by the evolution of the stellar density in the connectingtotheleadingtailoftheSgrstream.In-between residual map inside the SH on-cloud field, the peak of the thesetwolargestructures,atapproximatelyconstantGalac- HAC is clearly located in the 15<D <20 kpc distance RR tic latitude (at the resolution of the map) b∼50◦ runs the range,asalsoshowninFigure4.IntheNorth,theRRLyrae narrow Orphan Stream. The Sgr trailing tail is under the overdensitystrengthintheon-cloudfieldisgreatlyreduced disk,at50◦ <l<180◦ andb<−60◦.ComparingtheRRab ascomparedtotheSouth.ItisquitepossiblethattheHAC densityintheGalacticNorthandSouthat30◦ <l<50◦,an is asymmetric with respect to the plane of the disk. How- excess at −50◦ <b<−20◦ is obvious, roughly bounded by ever, there could be a more prosaic explanation for such theSHon-cloudfield.Whilethetoprowgivesthedistribu- an observation. Grey contours in the Figure 3 reveal the tionofRRabatalldistancesoverthewholesky,thesecond, levels of the dust extinction in the Galaxy as deduced by third and fourth row of the Figure zoom-in on the central Schlegel et al. (1998). According to this map, the redden- 180◦ in the Galaxy and map the RRab in three distance ingispatchyonratherlargeangularscales.Inparticular,a bins: 10<D <15 kpc, 15<D <20 kpc and 20<D <25 spur of dust reaches up to higher Galactic latitudes in the (cid:13)c 2014RAS,MNRAS000,1–?? 6 I.T. Simion et al. 0.25 28◦< l < 55◦, 25◦< b < 45◦ (NH) 661 - CSS data NH BPL - NH 0.20 657873 -- CBPSLS NdHata SH model 0.4 BPL - SH / 0.3 2 Ndeg−RR00..1105 datamodel)−00..12 x ( 0.05 2 − 0.0 0 1 0.000 5 10 15 20 25 30 −0.10 5 10 15 20 25 30 D (kpc) D (kpc) RR RR 0.25 31◦< l < 55◦, -45◦< b < -25◦ (SH) 661 - CSS data NH BPL - NH 0.20 excess 647876 -- CBPSLS SdHata SH model 0.4 excess BPL - SH 2 Ndeg−RR00..1105 datamodel)/−000...312 peak x ( 0.05 2 − 0.0 0 1 0.000 5 10 15 20 25 30 −0.10 5 10 15 20 25 30 D (kpc) D (kpc) RR RR Figure 4.Leftpanels:heliocentricdistancenumberdensitydistributionforCSSRRabinbinsof2kpcandDeasonetal.(2011)model predictionswitherrorbars(ingreen),completeness-corrected,intheNHandSHfields(redselectioninFigure1).Rightpanels:Residuals between the data and the model. The star density distributions are asymmetric in the NH and SH fields, with a strong excess of stars in the SH between 12 and 25 kpc and a peak at ∼ 18 kpc, also revealed in Figure 3. The excess in the NH field is weaker and not particularlysignificanttakingintoaccountthelargeerrorbars(seelegendforplotlabels). Northernhemisphere.Therefore,thisextraextinctioncould with 10 < D < 30 kpc in the three regions containing the be responsible for diminished star counts in the NH field. control field (0◦ < l < 85◦, 45◦ < b < 70◦, top row), the Clearly, the extinction by dust plays some role in shaping NH and NH2 fields (0◦ < l < 85◦, 25◦ < b < 45◦, mid- the apparent stellar halo density at lower Galactic l and b dle row), as well as the SH and SH2 fields (31◦ < l < 85◦, as illustrated in the top right panel of the Figure, where an −45◦ < b < −25◦, bottom row). The data is shown in the under-density is observed directly above the bulge, falling Left, the model predictions in the Middle and the residuals neatly inside the reddening contours. in the Right panels. The top panels confirm that the Dea- sonetal.(2011)modelisabletoreproducethebroad-brush According to previous studies (e.g. Belokurov et al. features of the data in the control-field, as also seen in Fig- 2007;Watkinsetal.2009),theHACissituatedintherange 20◦ <l<80◦, matching closely our detection in the Galac- ure 2. In the middle panels, an under-density is visible at lowGalacticlatitudesatl<30◦ (mostlikelycausedbythe tic South. To investigate in detail the line-of-sight distri- unaccounted dust extinction) as described above. However bution of RRab stars in the Cloud, we focus on two fields thesealsorevealasmallexcessfor28◦ <l<55◦andR>12 in the inner Galaxy, one in the Northern Hemisphere (NH: 28◦ < l < 55◦, 25◦ < b < 45◦) and one in the Southern kpc. Finally, the most dramatic over-density of RRab stars Hemisphere (SH: 31◦ < l < 55◦, −45◦ < b < −25◦). These is located below the Galactic plane, at the projected helio- centric distances between 12 and 16 kpc, for 30◦ <l <50◦ fieldsaremarkedinredinFigure1.Figure4showsthehelio- as evident from the maps shown in the bottom panels. centricdistancenumberdensitydistributionofRRabinthe NH (top) and SH (bottom) fields together with the model predictionsintheleftcolumn,andtherelativeresidualsbe- tweenthedataandthemodelintherightcolumn.Theerrors 4.1 Oosterhoff dichotomy in the North and the onthepredictednumberofRRLyraeδN asafunctionof South fields obs distancehavebeenestimatedasintheprevioussection,and TheRRLyraeperiod-amplitudeplaneisnotpopulateduni- take into account the uncertainty in the completeness δF formly.Thereappearstobesignificantclumpinginthedis- and in the expected number of counts δN . As evidenced exp tribution, also known as the Oosterhoff dichotomy (Ooster- inthetopleftpanel,overalltheRRabdistributionisinrea- hoff 1939). This bimodality, most pronounced for members sonable agreement with the model, although it exceeds the oftheGalacticstarclusters,issuspectedtooriginatemostly predicted counts slightly, especially at 15 < D < 20 kpc. fromthemetallicitydifferencesamongstthepopulation(see As the lower panel of Figure 4 illustrates, in the Southern e.g.areviewbyCatelan2009).TheOosterhoffI(OoI)type on-cloud field, SH, there exists a strong over-abundance of globular clusters (< P >∼ 0.55 days) are more metal- ab RRab stars. The data exceeds the model at all distances richcomparedtotheOosterhoffII(OoII)globularclusters beyond 10 kpc, however the observed distribution shows a (< P >∼ 0.64 days). One hypothesis is that Oo II type ab clear peak in the 17-19 kpc distance bin. clusters formed early in the proto-Galaxy while Oo I type TofurtherinvestigatetheextentoftheCloud,Figure5 clustersmighthaveformedsome2to3Gyrslater(e.g.Lee presentstheGalacticplaneprojectionsofthedensityofstars &Carney1999).However,GCagesmeasuredwiththehelp (cid:13)c 2014RAS,MNRAS000,1–?? RR Lyrae in the Hercules-Aquila Cloud 7 20 0◦< l < 85◦, 45◦< b < 70◦ DATA 20 MODEL 20 (DATA-MODEL) N N 20 0 3 6 9 12 20 =60◦20 −8 0 8 Y(kpc) 1105 8 12 16 1105 8 12 16 1105 8 1 2 l16 5 5 5 4 4 4 0 0 0 0 0 0 −8 −4 0 4 8 12 −8 −4 0 4 8 12 −8 −4 0 4 8 12 X (kpc) X (kpc) X (kpc) 20 0◦< l < 85◦, 25◦< b < 45◦ DATA 20 MODEL 20 (DATA-MODEL) N N 20 0 3 6 9 12 20 =60◦20 −8 0 8 Y(kpc) 1105 8 12 16 1105 8 12 16 1105 8 1 2 l16 5 5 5 4 4 4 0 0 0 0 0 0 −8 −4 0 4 8 12 −8 −4 0 4 8 12 −8 −4 0 4 8 12 X (kpc) X (kpc) X (kpc) 20 31◦< l < 85◦, -45◦< b < -25◦ DATA 20 MODEL 20 (DATA-MODEL) N N 20 0 3 6 9 12 20 =60◦20 −8 0 8 Y(kpc) 1105 8 12 16 1105 8 12 16 1105 8 1 2 l16 5 5 5 4 4 4 0 0 0 0 0 0 −8 −4 0 4 8 12 −8 −4 0 4 8 12 −8 −4 0 4 8 12 X (kpc) X (kpc) X (kpc) Figure 5. Top panels: X-Y number density maps over a 1132 deg2 field for the data (left) and the model (middle) in the control-field outlined in blue in Figure 1. These maps indicate the good agreement between the data and the model in this region of the sky where no known halo substructure is present in the distance range considered. Middle panels: X-Y maps over a 1385 deg2 field that includes ourselectionoftheNHandNH2fields.Weseeasmalloverdensityat30◦<l<50◦ butalsoastrongunderdensityforl<20◦,which isduetotheincompletenessofthesurveyinaregionofhighextinction,asshownbytheblackcontoursofconstantE(B-V)reddening inFigure3.Lowerpanels:X-Ymapovera880deg2 regionwhichincludestheSHansSH2fields.TheresidualsrevealtheRRLexcess in the SH at projected distances on the Galactic plane between 12 and 16 kpc. The maps have been smoothed with with a σ = 0.6 Gaussianfilterandarelimitedtothe10<D<30kpcdistancerange.TheoriginmarksthelocationoftheGalacticcenter(theSunis atX=−8;Y =0kpc). ofHSTACSphotometry(e.g.Dotteretal.2010)revealthat globular clusters (blue and magenta lines respectively) us- the age differences might be substantially smaller. Many of ingtheperiod-amplituderelationdefinedbyZorotovicetal. the neighbouring dwarf spheroidal galaxies and their glob- (2010)anddeterminetheperiodshifts∆P(A)(i.e.theoffset ular clusters do not display the Oosterhoff dichotomy but in period at constant amplitude from the Oo I locus line) fall instead in the so-called Oosterhoff gap on the period- for each RRab. In the bottom panels of Figure 6 we show amplitude(Bailey)diagramwith0.58<P <0.62days.It the ∆P period-shift distribution for the RRab stars in the ab is therefore clear that the distribution of lightcurve proper- NH ( 28◦ < l < 55◦, 25◦ < b < 45◦, left) and in the SH ( tiesoftheGalacticRRLyraecontainscomplementaryinfor- 31◦ <l<55◦,−45◦ <b<−25◦,right)fields.Thedistribu- mation about the accretion history of the stellar halo itself tion of ∆P values is centered on ∆P= 0 by definition (the harboring a larger fraction of Oo I type stars (e.g. Catelan position of the Oo I locus). However a tail at long periods & Cort´es 2008). isnoticeable;thisisduetotheOoIIcomponent.Byfitting ThetwotoppanelsofFigure6showtheso-calledBailey theperiodshiftdistributionwithatwo-componentGaussian diagramsforCSSRRabwithheliocentricdistancesbetween model we estimate that in the NH field, ∼ 69% of type-ab 12 and 22 kpc in the NH and SH fields (c.f. Figure 3 in the RRL belong to the Oo I component and 31% to the Oo II previoussection).WetracethelociforOoIandOoIItype in the distance range considered (left panels, Figure 6). In (cid:13)c 2014RAS,MNRAS000,1–?? 8 I.T. Simion et al. 28◦< l < 55◦, 25◦< b < 45◦ NH 31◦< l < 55◦, -45◦< b < -25◦ SH 1.6 NOoI/(NOoI+NOoII) = 69% 1.6 NOoI/(NOoI+NOoII) = 81% 1.4 12<DRR<22 kpc 1.4 12<DRR<22 kpc 1.2 1.2 g)1.0 g)1.0 a a m m A (0.8 A (0.8 0.6 0.6 0.4 0.4 0.2 Oo I Oo II 0.2 Oo I Oo II 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 P (days) P (days) NH SH 50 fOoI = 0.69, µOoI = 0.0 , σOoI = 0.03 50 fOoI = 0.81, µOoI = 0.0 , σOoI = 0.03 fOoII = 0.31, µOoII = 0.09, σOoII = 0.04 fOoII = 0.19, µOoII = 0.09, σOoII = 0.04 40 40 ∆Pb=0.055 ∆Pb=0.055 OoI OoII OoI OoII 30 30 NRR NRR 20 20 10 10 0 0 −0.1 0.0 0.1 0.2 0.3 −0.1 0.0 0.1 0.2 0.3 ∆ P (days) ∆ P (days) Figure 6. Top panels: Bailey diagrams for the RRL in the NH and SH fields, at 12<D <22 kpc. The continuous lines indicate the loci for the Oo I and Oo II components as defined by Zorotovic et al. (2010). Blue and magenta indicate the Oo I and Oo II RRL respectively, tentatively classified using the ∆P = 0.055 boundary. The dashed black lines represent the lower (A = 2.3−3.4P) and upper(A=3.3−3.4P)limitsexpectedformostoftheRRab.Bottompanels:HistogramoftheperiodshiftscalculatedforeachRRLat constant amplitude from the Oo I locus (blue line in the top panels) in the period-amplitude plane defined by Zorotovic et al. (2010). IntheNHfield69%ofthestarsareOoItypewhileintheSH,81%ofthem.TheexpectedproportionofOoIRRabis∼75%.Seethe textformoredetails. theSHfieldtheproportionissignificantlydifferent,withthe Table 2. Poisson Statistics for the RRab in the HAC region, majority of RRab (81%) belonging to Oo I population and shown in the top panels of Figure 7. We list two values for each only 19% to the Oo II population (right panels, Figure 6). region:RRLwith0<R<10kpc/10<R<20kpc.The’NH’ TosplitthesampleintoOoIandOoIIpopulationsasimple fieldhas0<Z<15kpcwhilethe’SH’,−15<Z<0kpc. boundary at constant period-shift of ∆P = 0.055 (red line inthebottompanels)isused:OoItypelietotheleftofthe RRLType Region Nexp Ndata Significance boundaryandOoIItypetotheright.IntheBaileydiagrams OoI,II NH 223/247 234/300 0.74/3.37 (top panels in Figure 6) for the NH and SH fields the Oo OoI,II SH 223/247 241/368 1.21/7.70 I and Oo II types RRL are marked with blue and magenta OoI NH 168/185 179/221 0.85/2.65 respectively while black indicates the stars that failed this Oo I SH 168/185 168/307 0.00/8.97 simpleclassification,becausesituatedabovethedefinedloci. OoII NH 56/62 55/79 0.13/2.16 Previous studies (e.g. Miceli et al. 2008; Sesar et al. 2013; OoII SH 56/62 73/61 2.27/0.37 Drake et al. 2013a) have shown that ∼ 75% of halo field RRab belong to the Oo I population. The fact that the SH field contains ∼ 81% of these indicates that the progenitor of 28◦ <l <55◦, to allow for symmetry with the SH field) of the HAC in the Southern Hemisphere is either an Oo I and SH fields. The left column shows the observed number populationglobularclusteroradwarfgalaxyfallingintothe density distributions of the RR Lyrae, the middle column Osterhoff gap. In fact, using the boundary ∆P = 0.055 we gives the smooth model predictions and finally the right classifymostofthestarsintheOosterhoffgapasOoItype. panels report the resulting residuals. From top to bottom TypeabstarsthatlietotheleftoftheOoIcurvemayeither the rows show: the entire RR Lyrae sample (top); the Oo I be metal-rich or have smaller mean amplitudes because of type(middle);andtheOoIItype(bottom).Themodelden- the Blazhko effect (Blaˇzko 1907). sitynormalisationsfortheindividualOotypesarechosenby We show the spatial distribution of the Oo I and Oo simply assuming that Oo I types make up 75% of the total II components along the sight-lines towards the HAC re- RRabpopulationandOoIItheremaining25%.Theresidual gion in Figure 7. The density map is shown in the plane of maps (right panels) reveal asymmetries in the North-South √ heliocentric distance R = X2+Y2 and height above the distributions. We quantify the statistical significance of the plane Z for both NH (for this plot, 31◦ < l < 55◦ instead overdense regions relative to the model for the (Oo I +Oo (cid:13)c 2014RAS,MNRAS000,1–?? RR Lyrae in the Hercules-Aquila Cloud 9 DATA MODEL DATA-MODEL 15 15 15 N N N 0 6 12 0 6 12 −5 0 5 10 10 10 5 5 5 Z(kpc) 0 Z(kpc) 0 Z(kpc) 0 HAC excess −5 −5 −5 −10 −10 −10 Oo I + Oo II Oo I + Oo II Oo I + Oo II −150 5 10 15 20 −150 5 10 15 20 −150 5 10 15 20 R (kpc) R (kpc) R (kpc) DATA MODEL DATA-MODEL 15 15 15 N N N 0 6 12 0 6 12 −5 0 5 10 10 10 5 5 5 Z(kpc) 0 Z(kpc) 0 Z(kpc) 0 HAC excess −5 −5 −5 −10 −10 −10 Oo I Oo I Oo I −150 5 10 15 20 −150 5 10 15 20 −150 5 10 15 20 R (kpc) R (kpc) R (kpc) DATA MODEL DATA-MODEL 15 15 15 N N N 0 6 12 0 6 12 −5 0 5 10 10 10 5 5 5 Z(kpc) 0 Z(kpc) 0 Z(kpc) 0 −5 −5 −5 −10 −10 −10 Oo II Oo II Oo II −150 5 10 15 20 −150 5 10 15 20 −150 5 10 15 20 R (kpc) R (kpc) R (kpc) Figure 7.RvsZnumberdensitymapsofalltheRRabLyrae(toprow),OoIRRL(middlerow)andOoIIRRL(bottomrow)forthe data(leftcolumn)andtheDeasonetal.(2011)model(middlecolumn)intheN-Ssymmetricfields31◦<l<55◦,|b|<45◦.Theresidual maps(rightcolumn)revealanoverdensityofOoIRRabintheSouthernHemisphere,forR>10kpc.Amuchweakersignalispresent alsointheNorthernHemisphereforR>12kpc,asalreadyseeninFigure5.Themodelcalculationstakeintoaccountthecompleteness functionofthesurvey.Thebinsizeis0.5kpcx0.5kpc.ThestatisticalsignificancesoftheoverdenseregionsarelistedinTable2.The plotshavebeensmoothedwithaGaussiankernelofσ=0.6. II), Oo I and Oo II populations at heliocentric distances R asignificantexcessofRRL(morespecificallyOoItype)for smaller and greater than 10 kpc. For each population we R>10kpc.Theexcesshasastatisticalsignificanceof∼9σ compute the statistical significance by finding the absolute for the Oo I population and ∼ 8σ if we consider the whole differencebetweentheobserved(N )andexpected(N sample.WhileintheSHtheoverdensityseemstobedueto data exp ) number of RR Lyrae stars, then dividing it by the Pois- the Oo I population, in the NH there seems to be a small (cid:112) son uncertainty ( N ). The results of this analysis are excessduetoboththeOoI(∼2.6σ)andOoIIpopulations exp summarizedinTable2.Welistseparatelythevaluesforthe (∼2.2σ).Thesevaluesexplaintheratios(f =69%forSH OoI regionsabove(NH)andbelow(SH)theGalacticplane(see andf =81%SH)weobtainedinFigure6.Highlightedin OoI column’region’)andforeachregionwecomputethesignifi- blackarethevaluesfortheOoIcomponentintheSH.Our canceoftheoverdensityforR<10kpcandR>10kpc(we results are broadly consistent with those reported in earlier listtwovaluesineachcolumn).Thevaluesconfirmthatthe studies.Forexample,inFigure24inSesaretal.(2010),the model agrees with the data for R < 10 kpc, while there is Hercules–AquilaCloudisseenat310◦ <R.A.<330◦ inthe (cid:13)c 2014RAS,MNRAS000,1–?? 10 I.T. Simion et al. 0.25 55◦< l < 85◦, 25◦< b < 45◦ 453 - CSS data NH2 0.3 BPL - NH2 510 - CSS data SH2 el BPL - SH2 0.20 452 - BPL mod 0.2 / 2eg−0.15 model) 0.1 NdRR0.10 ata−0.0 d x( 0.05 2 −−0.1 0 1 0.00 −0.2 0 5 10 15 20 25 30 0 5 10 15 20 25 30 D (kpc) D (kpc) RR RR Figure 8.NumberdensitydistributionsofCSSRRabinthefieldsNH2andSH2adjacenttotheNHandSHfields(seeFigure1).The Deasonetal.(2011)modelagreeswiththeobserveddistribution.Theabsenceofasignificantoverdensityintheseregionssuggeststhat theHACextendsmainlytotheNHand(mostly)SHfields. distributionofmain-sequencestarsinStripe82,atdistances Table 3. Examples of old and intermediate systems and esti- intherange10-25kpc,withafactorof∼1.6overdensity.In mates of the luminosity of the HAC progenitor. The values in the same study, it is found that the HAC is dominated by the first three columns are from Clement et al. (2001) (updated Oo I stars (end of section 4.5, Sesar et al. (2010)). catalogue of Variable Stars in Globular Clusters), Harris (1996, 2010) (the 2010 revision of his Catalog of Parameters for Milky WayGlobularClusters),andTable1inSmithetal.(2009)(RRL 4.2 Luminosity of the progenitor propertiesofDwarfSpheroidalGalaxies). To estimate the total luminosity of the Cloud we need to constrain its spatial extent. Figure 8 shows the heliocentric Oldpop. NRsyRs MVsys Lsys LHAC MVHAC number density distribution of RRab stars in two regions GCs (L(cid:12)) (L(cid:12)) symmetric with respect to the Galactic plane, and neigh- N5272(M3) 187 -8.9 3·105 6·105 -9.6 bouring the NH and SH fields (note boxes marked in green N3201 72 -7.4 8.1·104 4·105 -9.2 and labelled NH2 and SH2 in Figure 1). Field NH2 encom- N6333(M9) 9 -7.9 1.3·105 5·106 -12 passes stars with 55◦ < l < 85◦, 25◦ < b < 45◦, while field Interm.pop. SH2 contains stars with 55◦ < l < 85◦, −45◦ < b < −25◦. Dwarfgalaxies Thesedistributionsshownoclearexcess.Therefore,wecan conclude that at the Galactic latitudes probed by the CSS, LeoI 47 -12.0 5.4·106 3·107 -14.2 Draco 214 -8.8 2.8·105 5·105 -9.4 theHACmembersarepredominatelylocatedintheNHand Carina 54 -9.1 3.7·105 3·106 -11.2 SH fields. Fornax 396 -13.4 1.9·107 2·107 -13.3 Assuming that within the CSS footprint the majority of stars belonging to the Cloud are concentrated in the NH and SH fields, it is possible to compute a rough estimate of at all distances while the completeness decreases down to theinitialluminosityoftheprogenitor.Wecounttheexcess 40%at30kpc,leadingustounderestimatetheactualnum- RRab stars in both the NH and SH fields with respect to ber of RRL. It is also possible that the HAC is not limited the BPL model, in the 10 to 28 kpc distance range. In the to the NH and SH fields: if the NH and SH overdensities North,thisamountsto76RRab,andmorethandouble,i.e. are associated, it is likely that the overdensity extends at 181RRabstarsintheSouth.Theluminosityoftheprogen- lower galactic latitudes. The total luminosity of the Cloud itor is then estimated for a range of disruption scenarios. MHAC = −13 estimated by Belokurov et al. (2007), is in For example, the parent of the Cloud might have been an V agreement with the results found in this section. old and evolved system, with stellar populations similar to thatofaglobularcluster,packedwithnumerousRRLyrae. Alternatively,itcouldhavebeenasystemwithaninterme- diateagepopulationandasmallerproportionofRRLgiven 5 CONCLUSIONS the total luminosity. In Table 3 we list a series of possible We have used a sample of ∼ 14,000 RR Lyrae from the progenitor systems indicating the number of RRab Nsys in RR Catalina Schmidt Survey to map out the Hercules Aquila each one, their luminosity L and the inferred luminosity sys Cloudlocatedon theother sideof theGalaxy.To illustrate of the HAC’s progenitor L , assuming: HAC thelocationandtheextentoftheCloud,aschematicdraw- L /NHAC =L /Nsys (6) ingoftheCloud’ssignalastracedbytheRRLyraeisgiven HAC RR sys RR inFigure9.Thefiguregivesagraphicalsummaryofthere- where NHAC = 367, counting for a ∼70% completeness RR sultspresentedinthisPaper,whicharealsodetailedbelow. of the survey. The total absolute magnitude of a system, knowing its luminosity is Msys =M −2.5log (L /L ). • In the Galactic Southern hemisphere, there is a promi- V (cid:12) 10 sys (cid:12) AccordingtoTable3,theluminosityoftheprogenitorcould nent overdensity of RRab stars in the direction coincident be anywhere between MHAC ≈ −9 and MHAC ≈ −14. with the previous detections of the Hercules-Aquila Over- V V However, we have assumed a constant 70% completeness density(seee.g.Belokurovetal.2007;Watkinsetal.2009). (cid:13)c 2014RAS,MNRAS000,1–??