Table Of Content

Mon.Not.R.Astron.Soc.000,1–??(2010) Printed24January2014 (MNLATEXstylefilev2.2) Precision radial velocities of 15 M5 -M9 dwarfs 1 2 1 3 2 4 J.R. Barnes , J.S. Jenkins , H.R.A. Jones , S.V. Jeffers , P. Rojo , P. Arriagada , A. Jordań4, D. Minniti4,5, M. Tuomi1,6, D. Pinfield1, and G. Anglada-Escude´7 4 1CentreforAstrophysicsResearch,.UniversityofHertfordshire,.CollegeLane,Hatfield.Herts.AL109AB.UK. 1 2DepartamentodeAstronom´ıa,UniversidaddeChile,CaminodelObservatorio1515,LasCondes,Santiago.Chile. 0 3Institutfu¨rAstrophysik,Georg-August-Universita¨t,Friedrich-Hund-Platz1,Friedrich-Hund-Platz1,D-37077Gttingen.Germany 2 4InstitutodeAstrof´ısica,PontificiaUniversidadCato´licadeChile,Av.VicunãMackenna4860,7820436Macul,Santiago,Chile 5VaticanObservatory,V00120VaticanCityState,Italy n 6UniversityofTurku,TuorlaObservatory,DepartmentofPhysicsandAstronomy,Vaïsa¨la¨ntie20,FI-21500,Piikkio¨,Finland a 7AstronomyUnit,SchoolofMathematicalSciences,QueenMary,UniversityofLondon.UK. J 3 2 MNRAS,accepted ] P E ABSTRACT . h We present radial velocity measurements of a sample of M5V-M9V stars from p our Red-Optical Planet Survey, ROPS, operating at 0.652-1.025µm. Radial velocities - o for 15 stars, with r.m.s. precision down to 2.5ms−1 over a week long time scale r are achieved using Thorium-Argon reference spectra. We are sensitive to planets with t s mpsini>1.5M⊕ (3M⊕at2-σ) in the classical habitable zone and our observations cur- [a rentlyruleoutplanetswithmpsini>0.5MJat0.03AUforallourtargets.Atotalof9ofthe 15targetsexhibitr.m.s.<16ms−1,whichenablesustoruleoutthepresenceofplanetswith 2 m sini>10 M in0.03AUorbits. p ⊕ v Since the mean rotation velocity is of order 8kms−1 for an M6V star and 15kms−1 0 by M9V, we avoid observing only slow rotators that would introduce a bias towards low 5 axialinclination(i≪90◦)systems,whichareunfavourableforplanetdetection.Ourtargets 3 with the highest vsini values exhibit radial velocities significantly above the photon-noise 5 limitedprecision,evenafteraccountingforvsini.Wehavethereforemonitoredstellaractivity . 1 via chromospheric emission from the Hα and Ca II infrared triplet lines. A clear trend of 40 log10(LHα/Lbol)withradialvelocityr.m.s.isseen,implyingthatsignificantstarspotactivity isresponsiblefortheobservedradialvelocityprecisionfloor.TheimplicationthatmostlateM 1 dwarfsaresignificantlyspotted,andhenceexhibittimevaryinglinedistortions,indicatesthat : v observationstodetectorbitingplanetsneedstrategiestoreliablymitigateagainsttheeffects i ofactivityinducedradialvelocityvariations. X r Key words: (stars:) planetary systems stars: activity stars: atmospheres stars: spots tech- a niques:radialvelocities 1 INTRODUCTION dwarfs, the CRIRES survey (Beanetal. 2010) and the ROPS survey(Barnesetal.2012)(hereafterB12)havereportedprecisionra- Althoughthesolarneighbourhoodisdominatedbylowmassstars, dial velocities at the ∼10 ms−1 level for late M dwarfs (M6V- thelateMdwarfpopulationhasremainedlargelybeyondthereach M9V) withexistinginstrumentation. Reiners(2009) hasalsore- of optical precision radial velocity surveys. In order to address ported∼10ms−1stabilityontheflaringM6dwarfCNLeo.Work- thismajorparameterspace,dedicatedinstrumentshavebeenpro- ingintheinfraredKband,Beanetal.(2010)reported11.7ms−1 posedthatwouldinsteadoperateatlongerwavelengths,atthepeak forProximaCen,and5.4ms−1afterobservationswerebinnedto- of the energy distribution of low-mass stars (Jonesetal. 2008). gether.Ontheotherhand,B12,workinginthered-optical(0.62- Upcoming instruments are now being constructed, and include 0.90µm)foundthatwhilepropagatederrorswereatthe∼10ms−1 the Habitable Zone Planet Finder (Mahadevanetal. 2012) and level, the r.m.s. scatter was 16-35 ms−1 in the most stable tar- CARMENES,the Calar Alto high-Resolution search for M dwarfs gets. Until CRIRES is upgraded to a cross-dispersed, multi-order withExo-earthswithNear-infraredandopticalEchelleSpectrome- instrument,UVEShassubstantiallymorewavelengthcoveragewith ters(Quirrenbachetal.2012).However,whileanumberofwelles- reasonablesignal-to-noisefromwhichradialvelocitiesmaybede- tablishedinstrumentswithprovenstabilityatearlierspectraltypes rived. UVEShasalsoalreadydemonstrated2-2.5ms−1 precision have also reported precision radial velocities (RV) for early M (cid:13)c 2010RAS 2 J.R. Barneset al. over7yrsworkingwithanI2cell(Zechmeisteretal.2009)inthe a six day period (which we present in this paper), thus offer the 5000-6000A˚ range.However,by∼6500A˚,I2linesbecomeweak potentialtosample55percentofanM6Vhabitablezoneperiod, (<10percentofthenormalisedcontinuum)andarebarelyvisible and greater than a complete orbit for an M9V star. By defining beyond7000A˚.HenceI2 gascellscannot beusedintheredpart acontinuous habitable zone, therangeof possibleorbitalperiods oftheoptical,beyondthesewavelengths. for habitable planets are extended. For instance, Kopparapuetal. The first planet orbiting an M dwarf, was reported by (2013a)defineinnermoistgreenhouse andoutergreenhouse lim- Delfosseetal.(1998)andMarcyetal.(1998)nearlyadecade af- its, that extend the range of periods for an M6V habitable planet ter the first low mass companion to the main sequence star HD from∼6daystoamaximumof∼17days. 114762(Lathametal.1989),whichmaybeeitherabrowndwarf or massive planet, depending on the unknown orbital inclination. GJ876b, orbiting its parent M4V star in a 61day orbit is a gi- 1.2 ROPSSample antplanet,whichisperhapsnotsurprisinggiventhatclose-orbiting Ourchoiceoftargetswasbasedanumberoffactorsincludingvisi- companions are the easiest to detect with few epochs of obser- bilityandbrightness.InordertoobtainsufficientS/Ninthespectra vations using radial velocity techniques. However, while close- inexposureslimitedtonomorethan1800secswelimitedtheselec- orbiting planets have been predicted to be relatively common for early M dwarf samples (see §1.1 below), only ∼ 50 per cent tiontoM5-M9dwarfswithapparentIbandmagnitudes.14.5.A numberofstarsincommonwithourinitialobservationsmadewith of the M dwarf planets with mass estimates (16 from a total of 31)1 possess masses &0.3 MJ. The remaining 15 planets have theMIKEspectrographatMagellanClay(B12)havebeenretained. Additionaltargetswereselected,ensuringthatarangeofspectral minimummassesimplyingSuper-EarthtoNeptune-masscompan- ions.GJ876bisonlyoneoffourplanetssofardetected orbiting typeswereincludedwithlow-moderatevsinivalues.BecauseM stars,andparticularlylateMstarsonthewholearenoteffectively GJ876, andinfacttwooftheplanetspossessmassesofonly5.8 spundown,thosestarslaterthanM6Vtendtobemoderaterotators M and12.5 M . Inaddition,amongsttheKeplercandidatesfirst ⊕ ⊕ onthewhole.Jenkinsetal.(2009)foundthatM6Vstarsonaver- reported by Boruckietal. (2011) and confirmed by a number of authors (Fabryckyetal. 2012; Steffenetal. 2013; Muirheadetal. age possess vsini∼8kms−1, whereas this rises to ∼15kms−1 byM9V.Thisobviouslyhasimportantconsequencesforradialve- 2012),14planetshavebeenidentifiedwithradii6 3M ,whilst ⊕ locityprecision, especially if magnetic activityphenomena affect no transiting hot Jupiters have been detected. To date these find- therotationprofiles. Becausemoderaterotationisfoundonaver- ingsconfirmearlierpredictionsthatNeptunemassandEarth-mass planets are expected in greater numbers in orbit around M stars age,selectingonlytheslowestrotatingstarswithvsini65kms−1 (Ida&Lin2005). islikelytobiasatargetsampleto lowaxialinclination(i≪90◦) systems (i.e. with rotation axis aligned along the line of sight to theobserver),forwhichdetectionofplanetsislessfavourable.In 1.1 RockyplanetoccurrenceratesandtheMdwarf ordertocharacterisetheeffectsofactivityforthisandfuturesur- habitablezone veys, we included moderate rotators in our sample. The objects were selected for which vsini was on the whole well measured Bonfilsetal.(2013)havecalculatedphase-averageddetectionlim- (Mohanty&Basri 2003; Reiners&Basri 2010). In addition, fol- its for individual stars, which enable the survey efficiency of lowing the procedures detailed in Jenkinsetal. (2009), we have the HARPS (High Accuracy Radial velocity Planet Searcher) alsoobtainedthefirstvsinimeasurementsfortwoofthetargetsin early M dwarf sample to be determined. These detection lim- oursample,GJ3076andGJ3146,asindicatedinTable1. its enable corrections to be made for incompleteness, allow- Inthispaper weinvestigate themethods by whichprecision ing occurrence rates to be estimated. The frequency of HZ radialvelocitiescanbeachievedwithexistinginstrumentation,ex- planets, η⊕, (where1 M⊕<msini<10 M⊕) orbiting early tending the search of optical spectrometers into the 0.65-1.025 M dwarfs is found to be 0.36+0.25. From the Kepler sample, −0.10 µm wavelength region, where no established simultaneous refer- Dressing&Charbonneau(2013)estimatedη⊕ =0.90+−00..0043 forM encefiducialhasbeentested.Insection§3weoutlineourmaster dwarfplanetsupto4R .Withrevisedestimatesofhabitablezones ⊕ wavelength calibration procedure. The use of tellurics for wave- (Kopparapuetal. 2013a), Kopparapu (2013b) used the 95 Kepler lengthcalibrationisinvestigatedin§4usingananalysissimilarto planetcandidatesorbiting64low-masshoststarstosimilarlyplace that carried out by Figueiraetal. (2010) for HARPS observations conservativeestimatesofη⊕=0.51+−00..1200forMdwarfplanetswith of G type stars. We derive radial velocities for Proxima Centauri radiiintherange0.5-2R .Bonfilsetal.(2013)findthatthema- ⊕ using only telluric lines to enable us to determine the simultane- jorityof earlyMdwarfplanetsareclusteredinfew-daytotensof ouswavelengthsolution.Insection§5wepresenttheradialveloc- daysorbits,continuingthetrendwithsemi-majoraxisdistribution itymeasurementproceduresforourROPSsample,discussingour observedby(Currie2009).Byextrapolation,wemightexpect late wavelengthcalibrationprocedure(§5.2),applicableparticularlyto Mdwarfplanetsinorbitsuptoafew10sofdays. UVES observations, before presenting radial velocities for our 15 The centre of the continuous habitable zone for a M6V star M5V-M9V targets from 4 epochs of observations spread over a is estimated to be ∼0.045 AU (Kopparapuetal. 2013a). Hence, week-longtimescale(§5.3).Finally,wediscussourfindings(§5.4) a 7.5 M⊕ planet would induce a K∗=10ms−1 signal with an andprospectsforfutureobservations(§6). 11.0day period. Although (Kopparapuetal. 2013a) do not make habitable zone estimates for low masses, based on a simple flux and mass scaling, we estimate that an M9V star habitable zone would be centred at ∼0.023AU, with a 7.5 M⊕ planet inducing 2 OBSERVATIONS a15.8ms−1 signalwitha4.4dayperiod. Observationsspanning Inthispaper,weutiliseobservationsmadeduringourownobserv- ingcampaignin2012July.WealsousedatatakenfromtheEuro- 1 http://exoplanets.org peanSouthernObservatory(ESO)archive. (cid:13)c 2010RAS,MNRAS000,1–?? M dwarf radialvelocitieswithROPS 3 2.1 ROPSobservationswithUVES We observed 15 M dwarf targets with the Ultraviolet and Visual 2009 March 10 − airmass 1.27 1 2009 March 10 − airmass 2.00 E(reVcsLhoTelu,lltUeioTSn2p)o.efcORtrbos∼gerrav5pa4th,i0o0Un0sV.wEWSeereaotmbtsahedereve8wd.2itmohnaVf0oe.ur8ry′′hsLallaiftr,gnweighhTitceshlesgspcirvoeeapdea alised flux 00..68 m overaperiodofsixnightsintotal,on2012July23,24,26&29 Nor 0.4 0.2 (UTC).Shortorbitalperiodsmightbeexpectedbyextrapolatingthe tens of days orbits, found amongst early M dwarfs (Bonfilsetal. 09510 9520 9530 9540 9550 9560 9570 2013),tothelateMdwarfpopulation.Additionallytheobserving Wavelength [Å] strategyenabledthestabilityofUVESandourmeasurementpreci- 2009 March 12 − airmass 1.27 siontobecharacterisedonweeklongtimescales.Although UVES 1 2009 March 12 − airmass 2.01 offerstheabilitytosimultaneously recordobservations at shorter d flux 0.8 andlongerwavelengths,weoptedtomakeobservationsinthered alise 0.6 armonlysincemidtolate-Mstarsoutputlittlefluxshortof6000A˚. orm 0.4 InB12,wefoundtheratiooffluxinthe7000-9000A˚ regioncom- N 0.2 paredwiththe5000-7000A˚ regiontobe11.5and19forM5.5V 0 9510 9520 9530 9540 9550 9560 9570 andM9Vspectrarespectively.Thisestimateincludedthethrough- Wavelength [Å] putofthe6.5mMagellanClayandMIKEspectrograph. Working in the red-optical (i.e. 0.6-1.0 µm) poses a partic- Figure1.Spectralregion9510-9570A˚ illustratingthechangeinhumidity ular challenge in that there no currently operating ećhelle spec- between2009March10(lowhumidity:2-14percent)and2009March12 trometers coupled with 8m class telescopes that offer simultane- (highhumidity:48-54percent)atParanal.Notethatsomelinesbecome ouscalibration.Regularwavelengthobservationsforcalibrationare stronglysaturatedwhenthehumiditylevelsarehigh.Eventhoselinesthat crucial if precisions of order ms−1 are to be achieved from high donotsaturatearehighlyvariableinstrength,withsomelines(e.g.9550- resolution radial velocity information. Although UVES possesses 9551A˚)almostdisappearing. an iodine cell, the absorption lines of I2 do not extend far above 6500A˚,andarealreadyveryweak,withlinedepthsofonlyafew at vsini=2kms−1,means that spectral resolution (equivalent to percentofthenormalisedcontinuum.Wehavethereforeoptedto ∼6kms−1)dominatesthelinewidth. utilise near-simultaneous observations of Thorium-Argon (ThAr) arc lamp lines, coupled with the relative stability of UVES in or- dertoachievesub-ms−1 precisiononourtargetpopulationoflate 2.3 Dataextraction Mstars.SinceThArlampsexhibitmanylinesforcalibration,and aregenerallyalways availablebydefault withećhellespectrome- ThedatasetsforbothourROPSsample(§2.1)andProximaCen- tersworkingatopticalwavelengths,wemaderegularobservations tauri (§2.2) were flat field corrected by using combined expo- withthecomparisonlampavailablewithUVES.Acalibrationwas sures taken with an internal tungsten reference lamp. Since few includedintheobservingblockassociatedwitheachobservedtar- countsarerecordedinthereddestordersoftheMITLLCCD(ow- get and was taken immediately after each science frame. Further ing to the spectrograph efficiency and low quantum efficiency of detailsonthecalibrationproceduresaregivenin§3andfollowing the CCD longward of 1.0 µm), where of order 10,000 counts sections.Theobservingconditionsoverthefourhalfnightsnights could be achieved with 14 sec exposures compared with a peak wereverygood,withseeingestimatesintherange0.7-1.2fortar- of 40,000 counts, an additional 30 flatfield frames were taken in getsobservedatairmasses<1.5.OurtargetsarelistedinTable1. additiontothestandardcalibrationsfortheROPS(§2.1)dataset. The worst cosmic ray events were removed at the pre-extraction stageusingtheStarlinkFIGARO(Shortridge1993)routineBCLEAN (The Starlink software is currently distributed by the Joint As- 2.2 ProximaCentauriObservations tronomy Centre2). The spectra were extracted using ECHOMOP’s implementation of theoptimal extraction algorithmdeveloped by ProximaCentaurihasbeenshownbyEndl&Ku¨rster(2008)tobe Horne (1986). ECHOMOP rejects all but the strongest sky lines stable to 3.11 ms−1 over a 7 year period and thus we consider (Barnesetal. 2007b) and propagates error information based on this to be a good target to pursue as a calibrator for our tech- photon statisticsandreadout noisethroughout theextractionpro- niques. Data taken with UVES, spanning five nights, with obser- cess. vationsmadeonthreenightsandsinglenightgaps,wereobtained fromtheESOdataarchive.Thesedatawereinitiallytakenaspart ofamulti-wavelengthsurveyofProximaCentauri(GJ551)andare 3 WAVELENGTHCALIBRATION presented in Fuhrmeisteretal. (2011). Approximately 560 spectra of Proxima Centauri were continuously recorded on each of Wavelengthcalibrationatthems−1levelisrequiredifprecisionra- the three nights on 2009 March 10, 12 & 14, spanning 8 hours dialvelocitiesaretobeachieved.Tothisend,agreatdealofeffort pernightwithaltitudescorrespondingtoanairmassrangeof2.41- hasbeenexpendedinordertoobtainaccuratewavelengthsforspec- 1.27.Aslitwidthof1′′ givesaspectralresolution,R&43,000in tralcalibrationreferences(e.g.Gerstenkorn&Luc1978).Despite theredarmofUVES,whiletheextremeairmassrangeoftheobser- recentworkthathasidentifiednewsourcesforcalibration,suitable vationsledtoseeingthatvariedfrom1.4′′athighairmass,downto referencelinesareoftenlimitedinthewavelengthregionsthatthey ∼0.6′′ atlowairmass.TheCCDreadoutwasbinnedinthewave- lengthdirectionbyafactorof 2,resultinginanaveragepixelin- crementof2.4kms−1.TherotationvelocityofProximaCentauri, 2 http://starlink.jach.hawaii.edu/starlink (cid:13)c 2010RAS,MNRAS000,1–?? 4 J.R. Barneset al. span.Mahadevan&Ge(2009)haveidentifiedanumberofmolec- dominatedbyasystematicdifferencebetweenthewavelengthsand ulargascellsthatcouldbeusedtospantheHband,whileLASER thetwodimensionalfit.Thevariabilityinthewavelengthsolution comb technology has also been used to demonstrate ∼10ms−1 foragivensetofradialvelocitymeasurements(i.e.foreachstar)is precision on sky in the H band (Ycasetal. 2012). Although new thusimportantandultimatelydeterminestheprecisionthatcanbe calibration sources, rich in lines, have also been identified in the achieved.Wediscussthisfurther§5.2,butnoteherethatthisvari- redpartoftheoptical(Redmanetal.2011),ThArstillremainsthe ability is an order of magnitude smaller (i.e. <1ms−1) than the mostregularlyusedandonlyavailablecalibrationsourceforoptical zeropointr.m.s.valuesquotedabove. andinfraredhighresolutionspectrometers,althoughwithrelatively Theappropriatewavelengthsolutionforeachobservationcan fewlinesinthenearinfrared(>1µm). obtainedthroughasimultaneousmeasurement,byusingthetelluric lines, or a near-simultaneous measurement by using the nearby ThArreferenceframe.Ineachinstance,thecorrectionsaredeter- 3.1 Masterwavelengthcalibration minedaspixelshiftsandappliedtothemasterwavelengthsolution. This procedure enables wavelength corrections to be applied, al- ThAr wavelengths published by Lovis&Pepe (2007) were used lowingforlowordershiftsandstretches(duetomechanicaleffects toidentifystablelinesforwavelengthcalibration.Thislinelistis andtemperature/pressurechanges).Inotherwords,allowingmore estimated to enable a calibration (i.e. global) r.m.s to better than degrees of freedom for each solution can lead to poor fits in the 20 cms−1 for HARPS.Pixel positions were initiallyidentifiedfor firstandlastorder,neartheorderedges,andinregionswherethere asinglearcusingasimpleGaussianfit.Foreachsubsequent arc, may be fewer lines. Low order corrections correctly describe the across-match was made, followed by amultiple-Gaussian (upto changesintheinstrumentwhileminimisingvariabilityinthefits. three profiles) fit around each identified line using a Levenberg- Wedescribethetwomethodsadoptedinthispaperforupdatingthe Marquardt fittingalgorithm(Pressetal. 1986) toobtain thepixel wavelength,usingtelluriclines(§4.2)andThArlines(§5.2). positionofeach linecentre.TheLovis&Pepelinelistwasopti- misedforHARPSatR=110,000,whileourobservationsweremade at R ∼ 50,000 necessitating rejection of some lines that showed blending. Using a multiple Gaussian fit enables the effect of any 4 TELLURICSASAWAVELENGTHREFERENCE nearbylinestobeaccountedforinthefitthatalsoincludedafirst Thebenefitofutilisingtelluriclinestoobtainalocalwavelengthso- order (straightline)background. Anylinescloser thantheinstru- lutionisthatthewavelengthsarederivedfromtheveryobservation mentalFWHMwerenotused.Finallyforeachorder,anyremain- ofthestaritselfandarethereforesimultaneous.Thetelluricspec- ingoutlierswereremovedafterfittingacubic-polynomial.Inaddi- trumessentiallyfollowsthesamelightpathasthestarthroughthe tion,anylinesthatwerenotconsistentlyyieldingagoodfitforall earth’satmosphere,thetelescopeandthespectrograph,andisthus arcframesthroughout bothnights(towithin3-sigmaofthecubic subject to the same systematics. The calibration procedure could fits)wereremoved. beseenasanalogoustothatfirstadoptedbyMarcy&Butler(1992) TheThArobservationfollowingeachstaronthesecondnight andButleretal.(1996)iftheatmosphereoftheEarthcouldbewell was chosen arbitrarily as the reference solution for that star. The characterisedandcalibratedfor.Inaddition, atthetimeof obser- wavelengths werethen incrementally updated for all other observations,therewerenooptical8mclassspectrometersthatenable vationsofeachstarusingthemethodsthatwedescribein§4.2and simultaneousThArobservationstobemade. §5.2,whichareaimedatminimisingsystematicsinthewavelength The stability of telluric lines as reference fiducials has been solutionsfromoneobservationtothenext.Atotalwavelengthspan of6519A˚ to10252A˚ iscoveredbytheEEVandMITLLchipsatthe investigated by a number of authors. Griffin&Griffin (1973), for example, made some initial attempts to identify lines in the nonstandard840nmsettingofUVES.Anorderthatfallsbetween 6841-7424A˚ region,estimatinguncertaintiesatthe1-2mA˚ level, thetwoCCDscannotbeusedandmustbeaccountedforcorrectly equivalent to∼40-90 ms−1. Themost completelist of abinitio in the two dimensional solution. The candidate ThAr lines were line strengths and positions for water have now been calculated subjectedtoatwodimensionalfitofwavelengthvsextractedorder by Barberetal. (2006) and are now routinely used in model at- (cross-dispersion) for each CCD independently. For each star, an mosphere databases that supply molecular information for many arbitraryreferencesolutionwithatwodimensionalpolynomialfit molecules Rothmanetal. (2009). Bands of telluricmolecular ab- using4coefficientsinthewavelengthdirectionand6coefficients sorption lines pose a challenge for any ground based observa- inthecross-dispersion(order)directionwasmade: tions and are seen from the mid-optical, becoming stronger and widerintotheinfra-red.Inthered-optical,atwavelengthsgreater λ(x,y)= 3 aixi 5 bjyj (1) athta∼n 66856050A˚A˚,ansdig∼nifi7c5a9n5tA˚O2apapbesaor,rptthieonlabttaenrdssh,owwiinthgbstarnodnhgeaadbs- Xi=0 Xj=0 sorptionwithsaturationinsomelines.H2Obandheadsat6450A˚, whereaandbarethepolynomial coefficientsthatwefitfor. 7170A˚, 8100A˚ exhibit increasing widths from ∼100A˚ to sev- x and y are thepixel number and order number respectively and eral100A˚,howeverthebandcovering8890A˚ to9950A˚ isbyfar i and j are the powers in x and y for each coefficient. By it- the most extensive at wavelengths short of 1µm. Gray&Brown eratively rejecting outlying pixels from the fit, we found that of (2006)wereabletoachieveempiricalprecisionsof∼25ms−1us- theinput573lines,clippingthefurthestoutliersyieldedthemost ingstrongH2Oabsorptionlinesinthe6222-6254A˚ regionformed consistent fitfromone solutiontothenext.Typically15-20 lines in the optical path of the Coude´ ećhelle spectrograph used. This wererejectedbeforeafinalfitwasproducedforeachobservation. procedurehadtheadvantageofminimisingatmosphericprojection Thezeropointr.m.s.(i.e.ther.m.s.bycombiningalllines)forthe effectssuchaschangeinairmass. masterwavelengthcalibrationsisfoundtobe∼5-5.5ms−1 and Figueiraetal.(2010)insteadtookadvantageofnightlongob- ∼6-6.5ms−1 fortheEEVandMITLLchipsrespectivelyandrep- servationsmadeonbrightstablestarswiththe HARPS,locatedat resents the goodness of fit of the polynomial. These values are theESO3.6matLaSilla.Theyfoundclearnightlytrendsofthe (cid:13)c 2010RAS,MNRAS000,1–?? M dwarf radialvelocitieswithROPS 5 Star SpT Imag Exp vsini S/N MeanS/N Nobs r.m.s. r.m.s. r.m.s. r.m.s. [s] [kms−1] [ms−1] [ms−1] [ms−1] [ms−1] Extracted Decon Nocorr Lcorr Tcorr L-Tcorr GJ3076 M5V 10.9 400 17.1* 77±10 6120 4 100.7 92.3 67.6 44.5 GJ1002 M5.5V 10.2 300 63 106±12 9110 4 29.4 5.1 12.9 23.6 GJ1061 M5.5V 9.5 300 65 142±12 12150 5 4.23 2.4 2.4 2.8 LP759-25 M5.5V 13.7 1500 13 38±5 2810 4 106.8 79.9 70.6 65.9 GJ3146 M5.5V 11.3 600 12.4* 60±10 4920 4 87.2 47.1 80.2 7.75 GJ3128 M6V 11.1 350 65 65±3 5590 4 24.4 11.5 24.1 15.6 ProximaCentauri M6V 6.9 100† 2 191±22 12900 561 5.2 - - - GJ4281 M6.5V 12.7 1200 7 49±6 4240 4 36.7 11.7 12.0 15.3 SOJ025300.5+165258 M7V 10.7 350 65 95±12 8280 4 15.2 12.4 12.5 14.6 LP888-18 M7.5V 13.7 1500 63 36±3 2790 4 45.3 35.3 31.6 38.0 LHS132 M8V 13.8 500 65 37±2 3160 4 12.3 12.3 7.7 9.09 2MASSJ23062928-0502285 M8V 14.0 1500 6 38±1 2890 4 29.0 14.2 16.9 10.0 LHS1367 M8V 13.9 1500 65 32±3 2470 4 22.7 15.3 16.1 22.5 LP412-31 M8V 14 1200 12 26±7 1850 3 253.2 222.6 248.9 119.6 2MASSJ23312174-2749500 M8.5V 14.0 1500 6 37±2 2890 4 37.2 36.7 29.5 22.3 2MASSJ03341218-4953322 M9V 14.1 1500 8 33±2 2810 4 11.2 6.37 6.92 8.37 Table1.ListoftargetsobservedwithUVESwithestimatedspectraltypes,Ibandmagnitudes, exposuretimesandvsinivalues(columns1to5).Themeasured vsinivaluesaretakenfromMohanty&Basri(2003),Jenkinsetal.(2009)andReiners&Basri(2010).WederivedvsinisforGJ3076andGJ3146(denoted bya*)usingtheproceduresweadoptedinJenkinsetal.(2009).WealsolistdetailsforProximaCentauri andthemeanr.m.s.of5.2ms−1,afteratmospheric correctionforallthreenights,isgiven.Theexposuretimes†forProximaCentauriwerevariable,rangingbetween11secsand500secs,however74percent oftheobservationsweremadewith100secexposures.ExtractedS/NratioandS/Nratioafterdeconvolutionaretabulatedincolumns6&7. Column8lists thetotalnumberofobservations,Nobsoneachtargetandcolumn9givesther.m.s.scatterusinganNobs−1correctiontoaccountforthesmallnumberof observationsforeachobject(seesection§5.3).Incolumns10,11&12,welistr.m.s.valuesafterapplyingbisectorcorrectionsderivedfromthestellarline (L),telluricline(T),andbothlines(L-T).Discussionoftheresultsisgivenin§5.3. radial velocity variations of the O2 absorption band as measured 2.34ms−1 in their measurements, indicating an additional unac- in the spectra of τ Ceti (HD 10700), µ Ara (HD 160691) and ǫ countedforsourceofnoise. Eri(HD20794).Empiricalfitsweremadetothevelocitiesusinga Theseeingvariationsof1.4′′athighairmass,downto∼0.6′′ simplemodelthatincludedalinearairmassterm(fixed,withmag- atlowairmass,whenviewedwitha1′′ slitofferalessthanideal nitude,of∼20ms−1),aprojectionofthewindvelocityalongthe matchsincethestardoesnotcompletelyfilltheslit.Thisresultsin line of sight of the telescope (encompassing both magnitude and changesinilluminationoftheećhelle,leadingtoradialvelocities direction),andafixedcalibrationoffsetterm.Suchaprocedureen- thatcanpotentiallyvaryatthems−1toseveraltensofms−1level. abled typical measurement precisions over week long timescales SincetheCCDreadoutwasbinnedbyafactoroftwointhewave- of 4.5-10ms−1 to be made for observations at less than 1.5 air- lengthdirection, themean pixel increment of 2.4kms−1 istwice mass(>41.8◦)and 2.4-4ms−1 when restrictingobservations to that of the full 1.2 kms−1 mean readout increment used for the lessthan1.1airmass(>65.4◦).Overaperiodof6years,thepre- ROPStargets.Only oneThAr frameper night was recorded dur- cision wasfoundtobeoforder10ms−1. ingthe automated calibration procedures executed by UVES each night.Asaresult,itisimpossibletotrackanydriftofthespectro- graphthroughthenight,orinthisinstance,toinvestigateourability tousetheThArframesasanear-simultaneous referencefiducial. Duringextraction,wealsodiscoveredthaton 2009March10and 4.1 PrecisionradialvelocitiesofProximaCentauri 14,aregularhalfhour,cyclicshiftoforder1kms−1appearsinthe The opportunity to study the stability of telluric lines alone for radial velocities. The origin of this cyclical behaviour is unclear, updating the wavelength solution and providing a stable cross- butitappearstocoincidewithtimesatwhichtheseeingwasvery correlationreferenceagainstwhichtomakeprecisionradialveloc- good. We believe that it isrelated tothe mismatch of seeing and itymeasurementsisaffordedbythearchivalobservationsofProx- slitwidthwheretheautoguidermayhavebeenfooledintomaking ima Centauri, already outlined in §2.2 and initially published in onlyoccasionalcorrectionsthathaveresultedinsignificantećhelle Fuhrmeisteretal.(2011).Weintendedtocharacterisethestability illuminationchange. andbehaviour of UVESforourradialvelocitymeasurement tech- nique by making use of the ∼560 archival observations taken overthreenights,withanintentionofextendingthemethodtoour 4.2 Radialvelocitymeasurementprocedure ROPSsample.Inaddition,thiskindofstudyisnot possiblewith our2012Julyobservationssinceweonlyobservedeachtargetonce Startingwiththetwodimensionalwavelengthsolutiondescribedin pernight,whichprecludesmonitoringstabilityonminutetohour- §3.1,amethodofupdatingthelocalwavelengthsolutionforeach long timescales. Ku¨rsteretal. (1999) showed that Proxima Cen- observationmustbeobtained. Nospecialwavelengthcalibrations tauriisstabletothe54ms−1level,whilemorerecentresultsfrom were made during the observing sequence of Proxima Centauri, Endl&Ku¨rster(2008)haveshownittobestableto3.11ms−1over and only one ThAr spectrum was recorded during the standard a 7year period, but quote an average propagated uncertainty of calibrations foreachnight.Wethereforeinvestigatedtheuseofthe (cid:13)c 2010RAS,MNRAS000,1–?? 6 J.R. Barneset al. 2009 Mar 10 2009 Mar 12 2009 Mar 14 60 60 60 40 40 40 20 20 20 -1ms] Velocity [ -2 00 -2 00 -2 00 -40 -40 -40 Fit a, b, c and d Fixed a and g -60 -60 -60 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 2.1 2.15 2.2 2.25 2.3 2.35 2.4 2.45 4.1 4.15 4.2 4.25 4.3 4.35 4.4 4.45 15 15 15 10 10 10 -1ms] 5 5 5 Velocity [ - 05 - 50 - 05 -10 -10 -10 -15 ss == 64..0106 mmss--11 -15 ss == 55..5560 mmss--11 -15 ss == 75..4834 mmss--11 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 2.1 2.15 2.2 2.25 2.3 2.35 2.4 2.45 4.1 4.15 4.2 4.25 4.3 4.35 4.4 4.45 JD = 2454900.5 + t [d] JD = 2454900.5 + t [d] JD = 2454900.5 + t [d] Figure2.RadialvelocitiesofProximaCentauriforobservationsmadeon2009March10,12&14.Thetoppanelsshowtheheliocentricallycorrectedradial velocitiesoneachnightalongwiththefitswhichaccountforchangesinairmass,windvelocity,directionandoffset. Thesolid/redlinesindicatethefitsmade toeachnightindividuallywhilethegreen/dashedlinesareforfitsmadewithfixedγ=17.75.Seeingeffectsinthe1′′slit,whichvariedbetweenamaximum of1.4′′athighairmasstoaminimumof0.6′′atlowairmasswerenotaccountedforduringthemodelling.Theresidualsareplottedinthebottompanels (line typeandcolourcorrespondstothetoppanels).Thenightly-subtractedresidualsvarybetween4.16ms−1and5.84ms−1,whilethecorrespondingγ=17.75 fitresidualsgiver.m.s.valuesof5.56-7.43ms−1. abundantH2OandO2 inthered-opticaltoupdatethewavelength miningtheshiftofeveryrecordedinstanceofeachline.Amaskis solutions. made, toinclude alltheO2 lineswithinto4 FWHM.Only these The weather conditions on the first night were particularly linesareusedtodeterminethetransform. dry, with relative humidity variations in the 2-14percent range Wefoundthemostreliableprocedureforupdatingthewave- (as recorded for the telescope dome in the observation headers). lengthsolutionsviatelluriclinesistocalculatethetransformthat Onthesecondandthirdnights,therelativehumidityvariedinthe mapsthereferencespectrumtoeachindividualobservationinturn. ranges48-54%and22-28percent.Theincreasedwatercolumn The normalised master spectrum tj is thus scaled to the current isclearlyevidentintheH2OlinesasillustratedinFig.1.Thisaddi- normalisedobservedspectrum,sj byminimisingthefunction rtiaodniaalllvyesleorcviteyswtooirlklucsatrnatperowvheycthhaelluesnegionfgw.Wateitrhlicnaerseffuolrsperleeccitsiioonn, χ2= si−(ξ0+ξ1ti+ξ2t2i) 2 (2) itisinfactpossibletoselectH2Olinesthatarenotblended with Xi (cid:18) σi+τi (cid:19) other lines and that also do not vary so greatly in strength as to where become insignificant relative to the continuum level noise. Since Fig.1illustratestheextremesofthetelluriclinevariationsduring fi=ξ0+ξ1ti+ξ2t2i (3) theProximaCentauriobservations,wefoundthattheoptimalpro- isthetransformednormalisedmasterspectrumandξ1,ξ2&ξ3are cedure was to manually select the appropriate lines that fit these thequadratictransformcoefficientsforeachO2linepixel,i,desig- criteria.Overthe0.65-1.025µminterval,aninitiallistoftellurics natedbythemask.σiandτiaretheuncertaintiesontheobserved comprising∼1700lines,withnormalisedlinedepthsintherange spectrumandthemaster spectrumrespectively. Thisprocedureis 0.1-1.0, results in a subset of only ∼300 non-blended H2O and implementedsuchthatallmaskdesignatedlinesarefittedsimulta- O2 lineswithnormalisedlinesinthe0.6-0.95range.Changesin neously. Inother words, thesametransformcanbeappliedtoall instrumental resolution are likely to affect the selection, with the orderstoupdatethewavelengthsolution.Sinceξ1,ξ2 &ξ3 arein expectationthatmorelinescouldbeusedwithahigherinstrumen- pixelunits,thewavelengthincrementperpixeliscalculatedfrom talresolution. themasterwavelengthframeforallpixelsoveralltheordersused DespiteselectingonlythestrongestunblendedH2Olines,we fordeterminingradialvelocities.Themasterwavelengthincrement foundthatthemoststableprocedureentailedutilisingonlytheO2 mapismultipliedbythepixelincrementsandaddedtothemaster linesthat arerecorded intwobands on the EEVchip. Theuseof wavelengthstoupdatethewavelengthsolution. O2lineswasadvocatedandadoptedbyFigueiraetal.(2010)since AsinBarnesetal.(2012)wecarryoutaleastsquaresdecon- theyaremorestablethanH2Olineswhichoccurinaverynarrow volutionusinglineliststhatrepresentboththetelluriclineandthe layer and are highly variable, being correlated with weather and stellarlinepositions.WeusetheLineByLineRadiativeTransfer humidity patterns. Wethusmade useof only theorders recorded Model(LBLRTM)code(Cloughetal.1992,2005)toobtaintelluric onthischipfortheProximaCentauridataset,whichspan6519- linelists,whilewederivedthestellarlinelistsempirically.Inthe 8313AA.SincethetwoO2bandsspan5ordersintotal,withsome lattercase,weusedhighS/NobservationsofGJ1061 madewith linesrecordedtwice,wemakeuseofthefullinformationbydeter- a 0.4′′ slit. The GJ1061 line list was used for deconvolution of (cid:13)c 2010RAS,MNRAS000,1–?? M dwarf radialvelocitieswithROPS 7 theM5V-M7Vtargets.FortheM7.5V-M9Vtargetsweusedthe stabilisedHARPSobservationsofτ Ceti.Nevertheless,byholding spectraofLHS132(alignedandco-addedtoaugmenttheS/Nra- γ fixedat the mean velocity(forthethreenights) and fixingthe tio). Theprocedureforderivationofthestellartemplatesusedin 17.75ms−1valueforαfoundbyFigueiraetal.(2010),thecorre- thispaperisgiveninAppendixA.TwohighS/Nratiolinesarethus spondingcorrectedradialvelocityr.m.s.valuesforeachnightare calculatedforeachspectrum,withthefinalvelocitycalculationbe- σ = 6.00,5.60&7.43ms−1 onMarch10,12&14respectively. ing made by subtracting the telluriclineposition fromthe stellar ThecorrectedvelocitiesusingthisprocedurearelistedinTableA1 lineposition(measuredviacross-correlation). (column5,entitled“Acorr”).Morereasonablewindspeedsof115, 74and53ms−1arefound,butagainwestressthattheseareprob- ablybiasedbytheunconstrainedeffectsdiscussedabove.Mostno- 4.3 RadialvelocitystabilityofProximaCentauri tably,thecurvatureisnotfitwellinthesefits(Fig.2,upperpanel greencurves) indicating theprobableinvolvement of seeingvari- TheradialvelocitiesforthethreenightsareshowninFig.2. All ations. Forcomparison, whenconsideringthedatatakenwithan radialvelocitiespresentedinthissectionarelistedinA1.Thereis airmass range up to 1.5, Figueiraetal. (2010) found r.m.s. scat- acleartrendduringeachnightandanoffset,particularlywhenthe terofbetween4.54ms−1and5.81ms−1forτ Ceti(G8.5V)using firstnightiscompared withthesecondandthirdnights.Boththe thesamemethod asdescribed here. Theradial velocity of τ Ceti slope, curvature and offset changes from night tonight. Wehave isknown tobeverystablewithastandarddeviation of1.7ms−1 used the empirical procedure outlined in Figueiraetal. (2010) to Pepeetal.(2011). modelthetrendsseenintheRVsoneachnight.Theradialvelocity correction 4.4 Concludingremarks 1 Ω=α −1 +βcos(θ)cos(φ−δ)+γ (4) The study in this section was motivated by a desire to charac- (cid:18)sin(θ) (cid:19) terise a simultaneous reference fiducial in order to obtain a local wasshowntobesufficienttoadequatelyremoveatmospheric wavelengthsolutionforour deconvolution procedure. Withafew effects.Theparametersα,β,γandδcanbedeterminedwhenob- caveats, we areable to reproduce similar precision with an M6V servations are made throughout the night at the telescope eleva- star(ProximaCentauri)tothatachievedwithaG8Vstar(τ Ceti) tions(θ)andazimuthangles(φ)ofafixedtarget.αrepresentsthe with HARPS. Undoubtedly, a stabilised spectrograph, a narrower linear radialvelocitydrift per airmass(1/sin(θ)) duetochanges slit(oratleastaslitwidthwellmatchedwiththemedianseeing) in the line shape as different layers of the atmosphere are sam- shouldremovesomeoftheadditionaltrendsinthedatathatequa- pled. β is effectively the wind speed at the time of the observa- tion4cannotdescribe.Despitethesepromisingfindings,themajor tionandδisthewinddirection.γ isanadditionaloffsettermthat drawback of this procedure is that regular observations of a sin- describestheoffsetof theobservations fromzero, whenallother gletargetthroughouteachnightwouldbenecessaryforsuccessful termsarezero;inourcase,thistheheliocentricvelocitycorrection. implementation. Wewould never realisticallyexpect toobservea Wehave enabled all parameters to be fit in order to optimise the giventargetatsucharangeofairmasses,andindeedFigueiraetal. fitforeachindividualnight. Aftersubtractingthenightlyfits,the (2010) found that restricting observations to a narrower airmass residuals yield r.m.s. values of σ = 4.16, 5.50 &5.84 ms−1 on rangewasnecessarytoachievetheprecisionsreported. each of 2009 March 10, 12 & 14 respectively (Fig. 2). See also Giventhatthetrendsthroughouteachnightarealsoapproxi- TableA1 for a list of all corrected velocities (column 4, entitled matelylinearorquadratic,correctingforatmosphericeffectswitha “Icorr”).Thesevaluesappearreasonableconsideringtheexpected fourparameterfitsuchasEquation4clearlyrequiresveryhighS/N PoissonlimitedS/Nof∼2ms−1 (Barnesetal.2013).From pre- ratio.Obtainingfewms−1precisionviathismethodhasbeenpos- viouslyunpublishedarchivalHARPSdata3 andUVESobservations siblefor Proxima Centauri observations that enable S/N ratiosof (Zechmeisteretal. 2009), we find the radial velocity of Proxima afewhundred. However,typicalobservationsoflateMstarswill Centauri to show r.m.s. scatter at the 2.3ms−1 (27 observations) onlyachieve S/Nratiosofseveral tens, whichwillmoreseverely and4.3ms−1 (339observations)levelsrespectively (Tuomietal. restricttheprecisionachievable.Internalcalibrationreferencesare 2013,MNRAS,submitted). thereforealwaysapreferred,andmorerealisticoptionforobtain- The typical wind speed values we determine (130, 150 and ingthelocalwavelengthsolutionfordeconvolution.Wethussub- 190ms−1 foreachnight) arelargeandpotentiallynot physically sequently adopt this procedure for our ROPS sample of late M realistic.Inaddition,wefindrespectivevaluesforα,thevariation dwarfs,describedinthefollowingsections. perairmass,of31,11&23ms−1 whilethevalueofγ variesbe- tween -169.1ms−1and100.9ms−1(i.e.270ms−1variation).As noted by Figueiraetal. (2010), γ and α should be fixed. How- 5 UVESOBSERVATIONSOFALATEMDWARF ever the observations are not ideal, with varying humidity (see SAMPLE Gray&Brown(2006)foradiscussionoftemperature,pressureand humidityeffects,thatcanreachkms−1levels).Theadditionalprob- ForthelateMstarsobservedwithUVES,ourstrategycomprisedof lemswiththecyclicalbehaviourduringgoodseeingandtheappar- observingthesamesequenceof15targetsduringeachoffourhalf enttrendofuncorrectedradialvelocitydriftwithseeing,especially nights.Theobservationsweremadeoverasixdayperiodon2012 whentheseeingFWHMfallsinthe0.6-0.8′′rangeinthe1′′slit, July23,24,26&29.Thisenablesatimespanthatissufficientto arelikelytoyieldsystematics.Forthisreason,webelievethatthe discernshortperiodsignalsoforderafewdays.Sinceweareun- data are not able to reliably constrain wind speed values and di- abletoimplementtheproceduredescribedintheprevioussection, rections for theProxima Centauri observations, unlike thehighly whichmadeuseofthetelluriclinestoupdatethewavelengthso- lutionfordeconvolutionofeachspectrum(see§4.4),weusedthe near-simultaneousThArframerecordedaftereachobservationas 3 http//archive.eso.org/eso/eso archive main.html alocalwavelengthsolution. (cid:13)c 2010RAS,MNRAS000,1–?? 8 J.R. Barneset al. 1] 400 -ms 300 hift [ 120000 oTemp. [C]Vel. s-1013.013.500 4374474512.012.5 4574674710.010.5 427437446.57.0 41742743 Pressure [hPa] 0.2 0.3 0.4 7 1.2 1.3 1.4 7 3.2 3.3 3.4 7 6.26.36.4 7 HJD = 2456131.5 + t [d] Pressure [hPa] 743 744 745 745 746 747 742 743 744 741 742 743 1] 400 -ms 300 hift [ 120000 el. s-1000 V 13.0 13.5 12.0 12.5 10.0 10.5 6.5 7.0 Temp. [oC] Figure3.StabilityofUVESduringobservationsinJuly2012.Thetoppanelsshowthedriftinms−1oneachnightandthemiddlepanelsplotthetemperatureof theredcamera.While0.4◦-0.6◦driftisseenoneachnight,theabsolutetemperaturevaluesaredifferent.Thebottompanelsshowthedriftvsthetemperature. Temperaturesareplottedasfilledredcircles(scalesontheleftandbottomaxes),whilepressureisplottedasfilledgreensquares(scaleinhPaontherightand topaxes). 5.1 RadialvelocitystabilityofUVES 0.3 0.2 0.1 Pixel shiftPixel shift 0 -0.1 InB12,wedeterminedanincrementaldriftrelativetoareference -0.2 wavelength solutioninorder toobtainthelocal wavelength solu- -0.3 tionineachorder.TheMIKEspectrographhoweverexhibitedshifts EEV ofuptoafewhundredms−1 overshorttimescales,whichweat- SurfaMceIT fLitLs tributedtomechanical stabilityand possiblegravitational settling ofthedewarasthecoolantboilsoffduringthenight.UVESappears toexhibitamuchmorepredictablebehaviourinthatamoremono- 0 1000 2000 3000 0 5 10 15 20 25 30 35 tonic drift in wavelength isseen through asingle night, although Pixel Order thereisanoffsetbetweeneachnightasshowninFig.3(toppan- Figure4.ExampleoftheThArlinepixelshiftsforEEV(bluecircles)and els).Againthenightly offsetmayberelatedtobothdewar refills MITLL(redsquares)CCDsforthe33extractedorders.Theblack“+”sym- and to re-configuration of UVES which regularly observes at dif- bolsrepresentthefitted3(wavelength)by2(cross-dispersion/order)poly- ferentwavelengths.Shiftsoforder50ms−1 canbeexpectedwith nomial surface. Shifts are relative to the master wavelength frame taken UVESwhendifferentThArspectraaretakenafterchangingthein- witheachobservationonthesecondnightofobservations. strumentconfiguration4.Inaddition,shiftsoforder1/20pixelper 1hPa(millibar)changeinpressureandthesameshiftforachange of 0.3◦ in temperature are typical. The recorded 0.4◦-0.6◦ vari- 5.2 LocalThArwavelengthsolution ation throughout each night (Fig. 1, filled red circles) during our observations,wouldthusleadustoexpect100-150ms−1 wave- Subsequenttoobtainingamastersolutionforeachstar,asoutlined lengthshift.Thepressuredriftoneachnightisoforder1hPa(Fig. in §3.1, we have adopted a method for obtaining the local wave- 1, filled green squares) and hence presumably contributed to the lengthsolutionforeachframethatisdifferentfromthatdescribed observeddrift.Whileattributingtheobservedshiftstotemperature in §4.2, which made use of telluric lines. For our ROPS targets, changes alone is in agreement with expectation on nights 2, 3 & weobtainthelocalwavelengthframetakenaftereachobservation 4 of our observations, the first night, which was the least humid, byinsteadupdatingthewavelengthpositionsofalltheThArlines showed∼400ms−1driftthroughthenight.Atthesametime,the usedtodeterminethemastersolution.Thepixelpositionsofallthe temperatureswerehighestonthefirstnight,possiblyindicatingthat driftrateiscorrelatedwithtemperature.Thisincreaseddriftrateis discussedlaterinlightofourderivedradialvelocities. 4 http://www.eso.org/sci/facilities/paranal/instruments/uves/doc (cid:13)c 2010RAS,MNRAS000,1–?? M dwarf radialvelocitieswithROPS 9 linesarecalculatedasoutlinedin§3beforesubtractingthelinepo- vsiniwasalsosimulatedinBarnesetal.(2012,2013),andwefur- sitionsofthemasterwavelengthframe.Thisprocedurehasthead- ther discuss and illustrate the “excess” r.m.s. (i.e. above that ex- vantagethatlowerordercorrectionscanthenbeappliedtoupdate pectedfromvsiniandS/Nratioalone)in§5.4,§5.5.3andFig.8. themasterwavelengthsolution.Atwodimensionalfitismadefor For the early M dwarf sample targeted by HARPS, pixelpositionvsorderforallthemeasuredpixels.Inotherwordsa Bonfilsetal.(2013)foundananti-correlationwhenplottingbisec- twodimensionalpixelshiftsurfaceisdeterminedandwefindthat torspans(BIS)againstthemeasuredradialvelocities.Forinstance a polynomial of degree 3 (quadratic) in the wavelength direction aclearcorrelation(withaPearson’scorrelationofr=-0.81)was and2(linear)inthecross-dispersiondirection(Fig.4)issufficient identifiedforGl388(ADLeo).Subtractionofthetrenddecreased todescribe thedriftingwavelengthsolutionrelativetothemaster ther.m.s.from24ms−1 to14ms−1.WehavecalculatedtheBIS solutionwhichwascalculatedviaa4×6polynomial(§3.1).The (Gray1983;Toner&Gray1988;Mart´ınezFiorenzanoetal.2005) fittedpixelshiftsurfacecanbewrittenas forallourstarsandsubtractedthebestfitlineartrendwiththede- rived RVs. The uncorrected RVs are listed in column 9 of Table 1,whiletheBIScorrectedRVsarelistedincolumn10andshow 2 1 ∆p(x,y)= aixi bjyj (5) thatanumberofourstars alsodemonstratetrendsthatarelinked Xi=0 Xj=0 withthelinebisectorspan(BIS).ThesestellarlineBIScorrected velocitiesand subsequent r.m.s.values areplotted inFig.5, and where∆p(x,y)isthepixeldriftsurfacedefinedateachpixel,x, werefertothesecorrectedvaluesinthefollowingdiscussion.Sig- andextractedordernumber,y.Thecoefficientsaandbscalethex nificantimprovementsinther.m.sareseenforanumberoftargets, andy termsofpower iand j respectively. Thepixelsurfacesare wherether.m.s.ishalved.ThecorrectedRVshowevershowlittle converted toan updated wavelength surface by calculatingwave- improvementinthestarsthatexhibitthelargestvsiniandderived length increments from the master wavelength frame and adding r.m.s.values.Improvementsarealsoseenifacorrelationwiththe tothemasterwavelengthframe.Thisprocedurehastheadvantage telluric BIS is removed (column 11), indicating that atmospheric of maintaining stability as any order edge effects are minimised variationmayalsocontributetolimitingtheprecisionthatcanbe inalow-orderfit.Zeropointr.m.s.valuesinthewavelengthsolu- achieved using the methods outlined above. Also, variability in tionsof2.01±0.20ms−1 and2.63±0.24ms−1 fortheEEVand theslitillumination(e.g.duetoseeingchanges)affectstheinstru- MITLLchipsrespectivelyarefound.Asalreadynotedin§3,these mentalpoint-spread-function,thusaffectingbothstellarandtelluric valuescouldbereducedbyusingadditionalcalibrationlamps,but linestosomedegree.Thiswillgosomewaytoexplainingwhystel- we note that the 1-σ variability is an order of magnitude lower larortelluriclinescanimprovethemeasuredr.m.s.Howeveronly at20cms−1and24cms−1,andwellbelowthephotonnoisepre- thestellarlinescontainlineshapevariabilityintroducedbythestar cision that can be achieved with UVES using the techniques de- itself.Finally,wehavealsoinvestigatedincorporatingboththeline scribedinthispaper. andtelluricBISmeasurements.Sincethefinalradialvelocitiesare measured by subtracting thetelluriclineposition fromthe stellar position,wealsolistRV-BIScorrectionsforastellar-telluricBIS 5.3 Radialvelocitiesof15lateMdwarfs correction(column12). The mean-subtracted radial velocities for our ROPS targets are plotted in Fig.5 with details of r.m.s. estimates listed in Table1. 5.4 Discussion Appendix A gives full details of all radial velocities, which are listed in Tables A2 & A3. The radial velocities are measured as The r.m.s. velocities demonstrate that near-photon noise limited outlinedinB12bysubtractingthedeconvolved telluriclineposi- precision is achievable using our red optical survey. Following tionfrom the simultaneously observed stellar line. The lineposi- B12, where photon noise limited simulations were made with tionsaremeasuredbycrosscorrelatingeachstellarlinerelativeto the MIKE spectrograph at the 6.5m Magellan Clay telescope, we the mean deconvolved stellar line for each target, and similarly have estimated that 1.5-2 ms−1 should be achieved with UVES for the telluric lines. We use the HCROSS algorithm of Heavens (Barnesetal.2013). Theobservations, inparticularforGJ1061, (1993)whichisamodificationoftheTonry&Davis(1979)cross- GJ1002and2MASSJ03341218-4953322(Table1)thusshowcon- correlationalgorithm.HCROSSutilisesthetheoryofpeaksinGaus- siderableimprovementsoverrecentmeasurementsthathavemade siannoisetodetermineuncertaintiesinthecross-correlationpeak. use of telluric lines as a reference fiducial (e.g. Reiners 2009; Wehavemadeaminormodificationoftheroutine,whichbelongs Rodleretal.2012;Baileyetal.2012).TheBIScorrected2.4,5.1 to the Starlink package, FIGARO, in order to directly output both and6.4ms−1measurementsfortheseobjectscomparefavourably thepixelshift,andshiftuncertainty. with those that we obtained with HARPS for the brightest targets From Table 1, it can be seen that a range of exposure times inoursample.WhileGJ1061andGJ1002havenotbeenactively and S/N values were obtained, depending on the brightness of monitored with HARPS, 4 observations for each target (that re- the target, which ranged from mI = 9.5 to mI = 14.1. In ad- main unpublished) exist in the European Southern Observatory’s dition, not all observed targets possess slow rotation, which we archive.UsingTERRA,theTemplate-EnhancedRadialvelocityRe- define as, at, or below the instrumental resolution of 54,000, or analysis Application (Anglada-Escude´&Butler 2012), a pipeline 5.55kms−1.Jenkinsetal.(2009)foundthatatM6V,starspossess suitedesignedtoimprovetheRVsachievedbythestandardHARPS vsini=8kms−1 onaverage, whilethisincreasesto∼15kms−1 DataReductionSoftware(DRS),wehavefound2.04ms−1&2.32 by M9V. Table 1 and Fig. 5 demonstrate that those stars with ms−1precisionsforGJ1061andGJ1002 (seeTableA4forRVs). slower vsini values on the whole appear to enable better radial WenotethatonlythereddestordersofHARPSintheverybrightest velocityprecisiontobedetermined,asfirstnotedbyButleretal. midMtargetsenableprecisionofafewms−1tobeachieved. (1996).Thisisnotsurprisingsincetheresolutioniseffectivelyde- Despitethesub-10ms−1 r.m.s.values,anumberofourstars gradedandlineblendingincreaseswithincreasingvsini.Thecor- exhibitradialvelocitiesthataresignificantlyinexcessofthepho- relation between photon limited precision and r.m.s. for a given tonnoiselimitedprecisionthatweexpect fromourtargets, even (cid:13)c 2010RAS,MNRAS000,1–?? 10 J.R.Barneset al. 150 100 50 V [m/s] − 55 000 (v sGsinJ = i3 =90 271.673 . (4mM ks5m−V1s)−1) 0 L(Pv s8s i8=n8 3i− 5£1. 833 (mkMms7−s.1−51V)) R−100 −50 −150 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 30 10 20 V [m/s] 0 (Gv Jss i1 n=0 i62 £1.4 5( mM kms5−.s51−V1)) − 11 000 (vL sHs i=Sn 1i1 2£3. 235 (mkMms8−sV1−)1) R−10 −20 −30 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 10 20 V [m/s] 0 (Gv Jss i1 n=0 i05 £2.0 3( mM kms5−.s51−V1)) 0 2M(vJ s2s3 i=n0 61i −4=0. 265 0mkm2s −s(M1−18)V) R−10 −20 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 200 30 20 V [m/s] 10 00 (LvP ss7i 5n=9 i7 −=92 .1593 (m Mkms5−.s51−V1)) − 11 000 (LvH ss Si=n 11i 35£ 6. 357 mk(Mms8−s1−V1)) R −20 −100 −30 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 100 300 200 V [m/s] 0 (v GssiJn 3=i 1 =44 716.2 1(. M4m 5ks.m−51Vs−)1) − 1100 000 (Lv sPs i=4n 1 2i 22=−2 13.621 mk(Mms8−sV1−1)) R −200 −100 −300 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 20 50 s] 10 GJ 3128 (M6V) 2MJ2331−2749 (M8.5V) m/ 0 (v sin i £ 5 kms−1) 0 (v sin i = 6 kms−1) V [−10 s = 11.5 ms−1 s = 36.7 ms−1 R −20 −50 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 20 20 V [m/s]− 11 000 (Gv sJs i4=n2 1i8 1=1. 77( M mkm6s.−s51−V1)) 1 00 2M(vJ 0ss3i n3= 4 i6 −=.44 89 m 5km3s− s(1M−19)V) R −10 −20 −20 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 JD = 2456131.5 + n [days] 20 RV [m/s]− 11 000 SO(Jv s0s 2i=n5 1i3 2£+. 1456 mk5m2s −s(1−M17)V) −20 0 1 2 3 4 5 6 7 JD = 2456131.5 + n [days] Figure5. Heliocentrically correctedradialvelocities plottedforour15UVESROPStargets.Observations weremadeon2012July22,23,25&28.The samplecontainsatotalofsevenM5V-M6.5VandeightM7V-M9Vtargets.Aradialvelocityprecisionof2.4&5.0ms−1 ismeasuredforquiet,slowly rotatingtargetsatspectraltypeM5.5V(GJ1061andGJ1002),while6.4ms−1isfoundforourlatest,M9V,target(2MASSJ03341218-4953322).Thetargets showinghigherr.m.s.inTable2eitherexhibitsignificantrotation(vsini&10kms−1),significantvariabilityinthechromosphericindicatorsCaIIandHα, orboth. (cid:13)c 2010RAS,MNRAS000,1–??